commit fe3288f0f599a57b43f37ed3d9d2f26aedf7e74d Author: ModelHub XC Date: Mon Aug 24 22:08:28 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.17 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..1e6e860 --- /dev/null +++ b/README.md @@ -0,0 +1,59 @@ +--- +base_model: meta-llama/Llama-3.1-8B-Instruct +datasets: Neelectric/Replay_0.01.MoT_science.wildguardmix.Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_SFT_sciencev00.17 +tags: +- generated_from_trainer +- open-r1 +- trl +- sft +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_SFT_sciencev00.17 + +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_0.01.MoT_science.wildguardmix.Llama3_4096toks](https://huggingface.co/datasets/Neelectric/Replay_0.01.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_sciencev00.17", 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/26yn0joi) + + +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..7f87134 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.1035281197777814e+19, + "train_loss": 0.9530878761967747, + "train_runtime": 21335.3439, + "train_samples": 145693, + "train_samples_per_second": 13.657, + "train_steps_per_second": 0.854 +} \ 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..4a4832b --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:155478a6c705ecc88346a34027906828b2f9a564284c7aebf5bcd99a8ed15123 +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..e8acf88 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9dcc24c89107e2b250c05f729e6d8103db37a5afe5560a0445bfe0c871a9421d +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..0f4787a --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02a9d3761707da323b55bce5cd574ba55f466a578b39277764addbf82b9ea780 +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..1a6ab8e --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd4460a1a11c6cee904061e57719cb2a53ff7443f9b477b44c741359f1a482a4 +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..7f87134 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.1035281197777814e+19, + "train_loss": 0.9530878761967747, + "train_runtime": 21335.3439, + "train_samples": 145693, + "train_samples_per_second": 13.657, + "train_steps_per_second": 0.854 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..148c3ad --- /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": 7.300925254821777, + "learning_rate": 0.0, + "loss": 1.4841, + "mean_token_accuracy": 0.6132287979125977, + "num_tokens": 25099.0, + "step": 1 + }, + { + "epoch": 0.00021963540522732265, + "grad_norm": 6.589354038238525, + "learning_rate": 5.48847420417124e-10, + "loss": 1.4857, + "mean_token_accuracy": 0.6123285293579102, + "num_tokens": 54691.0, + "step": 2 + }, + { + "epoch": 0.000329453107840984, + "grad_norm": 6.737575531005859, + "learning_rate": 1.097694840834248e-09, + "loss": 1.4471, + "mean_token_accuracy": 0.617202639579773, + "num_tokens": 81754.0, + "step": 3 + }, + { + "epoch": 0.0004392708104546453, + "grad_norm": 7.5840678215026855, + "learning_rate": 1.646542261251372e-09, + "loss": 1.5597, + "mean_token_accuracy": 0.6015745401382446, + "num_tokens": 106869.0, + "step": 4 + }, + { + "epoch": 0.0005490885130683066, + "grad_norm": 7.931573867797852, + "learning_rate": 2.195389681668496e-09, + "loss": 1.5562, + "mean_token_accuracy": 0.6061089038848877, + "num_tokens": 128291.0, + "step": 5 + }, + { + "epoch": 0.000658906215681968, + "grad_norm": 7.768801212310791, + "learning_rate": 2.7442371020856205e-09, + "loss": 1.4751, + "mean_token_accuracy": 0.6183398962020874, + "num_tokens": 150815.0, + "step": 6 + }, + { + "epoch": 0.0007687239182956292, + "grad_norm": 7.086880207061768, + "learning_rate": 3.293084522502744e-09, + "loss": 1.4697, + "mean_token_accuracy": 0.6157780289649963, + "num_tokens": 177612.0, + "step": 7 + }, + { + "epoch": 0.0008785416209092906, + "grad_norm": 7.966911315917969, + "learning_rate": 3.841931942919868e-09, + "loss": 1.5507, + "mean_token_accuracy": 0.5960360169410706, + "num_tokens": 200608.0, + "step": 8 + }, + { + "epoch": 0.000988359323522952, + "grad_norm": 6.685125350952148, + "learning_rate": 4.390779363336992e-09, + "loss": 1.3834, + "mean_token_accuracy": 0.6347727179527283, + "num_tokens": 227801.0, + "step": 9 + }, + { + "epoch": 0.0010981770261366132, + "grad_norm": 8.334959983825684, + "learning_rate": 4.939626783754116e-09, + "loss": 1.6375, + "mean_token_accuracy": 0.5836213827133179, + "num_tokens": 249896.0, + "step": 10 + }, + { + "epoch": 0.0012079947287502745, + "grad_norm": 6.189165115356445, + "learning_rate": 5.488474204171241e-09, + "loss": 1.5229, + "mean_token_accuracy": 0.6069564819335938, + "num_tokens": 285639.0, + "step": 11 + }, + { + "epoch": 0.001317812431363936, + "grad_norm": 7.454898357391357, + "learning_rate": 6.0373216245883644e-09, + "loss": 1.539, + "mean_token_accuracy": 0.6075851321220398, + "num_tokens": 308916.0, + "step": 12 + }, + { + "epoch": 0.0014276301339775973, + "grad_norm": 6.882843017578125, + "learning_rate": 6.586169045005488e-09, + "loss": 1.5005, + "mean_token_accuracy": 0.6127958297729492, + "num_tokens": 335588.0, + "step": 13 + }, + { + "epoch": 0.0015374478365912585, + "grad_norm": 7.327468395233154, + "learning_rate": 7.135016465422612e-09, + "loss": 1.5263, + "mean_token_accuracy": 0.6071827411651611, + "num_tokens": 361190.0, + "step": 14 + }, + { + "epoch": 0.0016472655392049198, + "grad_norm": 6.2751569747924805, + "learning_rate": 7.683863885839736e-09, + "loss": 1.5065, + "mean_token_accuracy": 0.6069755554199219, + "num_tokens": 396266.0, + "step": 15 + }, + { + "epoch": 0.0017570832418185812, + "grad_norm": 7.5034942626953125, + "learning_rate": 8.23271130625686e-09, + "loss": 1.4101, + "mean_token_accuracy": 0.6236786842346191, + "num_tokens": 420318.0, + "step": 16 + }, + { + "epoch": 0.0018669009444322424, + "grad_norm": 7.060347080230713, + "learning_rate": 8.781558726673984e-09, + "loss": 1.4645, + "mean_token_accuracy": 0.6110981702804565, + "num_tokens": 445206.0, + "step": 17 + }, + { + "epoch": 0.001976718647045904, + "grad_norm": 8.628910064697266, + "learning_rate": 9.330406147091108e-09, + "loss": 1.5176, + "mean_token_accuracy": 0.6095705032348633, + "num_tokens": 465341.0, + "step": 18 + }, + { + "epoch": 0.002086536349659565, + "grad_norm": 7.919843673706055, + "learning_rate": 9.879253567508231e-09, + "loss": 1.5033, + "mean_token_accuracy": 0.6071434617042542, + "num_tokens": 486885.0, + "step": 19 + }, + { + "epoch": 0.0021963540522732263, + "grad_norm": 7.990423202514648, + "learning_rate": 1.0428100987925357e-08, + "loss": 1.5379, + "mean_token_accuracy": 0.6065486669540405, + "num_tokens": 509692.0, + "step": 20 + }, + { + "epoch": 0.002306171754886888, + "grad_norm": 7.754836559295654, + "learning_rate": 1.0976948408342482e-08, + "loss": 1.5982, + "mean_token_accuracy": 0.5912879705429077, + "num_tokens": 532743.0, + "step": 21 + }, + { + "epoch": 0.002415989457500549, + "grad_norm": 7.039846897125244, + "learning_rate": 1.1525795828759604e-08, + "loss": 1.4029, + "mean_token_accuracy": 0.637151837348938, + "num_tokens": 557866.0, + "step": 22 + }, + { + "epoch": 0.0025258071601142102, + "grad_norm": 6.614126205444336, + "learning_rate": 1.2074643249176729e-08, + "loss": 1.5048, + "mean_token_accuracy": 0.6118674278259277, + "num_tokens": 586481.0, + "step": 23 + }, + { + "epoch": 0.002635624862727872, + "grad_norm": 7.085023403167725, + "learning_rate": 1.2623490669593852e-08, + "loss": 1.5105, + "mean_token_accuracy": 0.6053236722946167, + "num_tokens": 612224.0, + "step": 24 + }, + { + "epoch": 0.002745442565341533, + "grad_norm": 6.345484733581543, + "learning_rate": 1.3172338090010976e-08, + "loss": 1.3612, + "mean_token_accuracy": 0.6446989178657532, + "num_tokens": 640880.0, + "step": 25 + }, + { + "epoch": 0.0028552602679551946, + "grad_norm": 6.305335998535156, + "learning_rate": 1.3721185510428101e-08, + "loss": 1.5175, + "mean_token_accuracy": 0.6023125052452087, + "num_tokens": 673462.0, + "step": 26 + }, + { + "epoch": 0.0029650779705688557, + "grad_norm": 6.251530170440674, + "learning_rate": 1.4270032930845225e-08, + "loss": 1.5199, + "mean_token_accuracy": 0.601014256477356, + "num_tokens": 704958.0, + "step": 27 + }, + { + "epoch": 0.003074895673182517, + "grad_norm": 7.016903877258301, + "learning_rate": 1.4818880351262348e-08, + "loss": 1.501, + "mean_token_accuracy": 0.6115278601646423, + "num_tokens": 730724.0, + "step": 28 + }, + { + "epoch": 0.0031847133757961785, + "grad_norm": 7.830250263214111, + "learning_rate": 1.536772777167947e-08, + "loss": 1.551, + "mean_token_accuracy": 0.5986595153808594, + "num_tokens": 752409.0, + "step": 29 + }, + { + "epoch": 0.0032945310784098397, + "grad_norm": 7.377598285675049, + "learning_rate": 1.5916575192096597e-08, + "loss": 1.5083, + "mean_token_accuracy": 0.6185790300369263, + "num_tokens": 775937.0, + "step": 30 + }, + { + "epoch": 0.003404348781023501, + "grad_norm": 6.9750800132751465, + "learning_rate": 1.646542261251372e-08, + "loss": 1.5112, + "mean_token_accuracy": 0.6117210388183594, + "num_tokens": 803955.0, + "step": 31 + }, + { + "epoch": 0.0035141664836371624, + "grad_norm": 6.894033908843994, + "learning_rate": 1.7014270032930844e-08, + "loss": 1.4984, + "mean_token_accuracy": 0.6083914041519165, + "num_tokens": 832955.0, + "step": 32 + }, + { + "epoch": 0.0036239841862508236, + "grad_norm": 7.338194847106934, + "learning_rate": 1.756311745334797e-08, + "loss": 1.4518, + "mean_token_accuracy": 0.6219106912612915, + "num_tokens": 857912.0, + "step": 33 + }, + { + "epoch": 0.0037338018888644848, + "grad_norm": 6.991853713989258, + "learning_rate": 1.811196487376509e-08, + "loss": 1.4438, + "mean_token_accuracy": 0.6188669800758362, + "num_tokens": 883926.0, + "step": 34 + }, + { + "epoch": 0.0038436195914781464, + "grad_norm": 6.396656036376953, + "learning_rate": 1.8660812294182216e-08, + "loss": 1.4573, + "mean_token_accuracy": 0.6150898337364197, + "num_tokens": 914246.0, + "step": 35 + }, + { + "epoch": 0.003953437294091808, + "grad_norm": 6.301985263824463, + "learning_rate": 1.920965971459934e-08, + "loss": 1.4308, + "mean_token_accuracy": 0.6316434741020203, + "num_tokens": 944191.0, + "step": 36 + }, + { + "epoch": 0.004063254996705469, + "grad_norm": 6.834984302520752, + "learning_rate": 1.9758507135016463e-08, + "loss": 1.5077, + "mean_token_accuracy": 0.607927680015564, + "num_tokens": 972265.0, + "step": 37 + }, + { + "epoch": 0.00417307269931913, + "grad_norm": 8.327461242675781, + "learning_rate": 2.030735455543359e-08, + "loss": 1.5396, + "mean_token_accuracy": 0.5999776721000671, + "num_tokens": 993492.0, + "step": 38 + }, + { + "epoch": 0.004282890401932792, + "grad_norm": 7.98877477645874, + "learning_rate": 2.0856201975850713e-08, + "loss": 1.5018, + "mean_token_accuracy": 0.6170031428337097, + "num_tokens": 1015535.0, + "step": 39 + }, + { + "epoch": 0.004392708104546453, + "grad_norm": 7.2313737869262695, + "learning_rate": 2.1405049396267835e-08, + "loss": 1.4849, + "mean_token_accuracy": 0.6129239797592163, + "num_tokens": 1041629.0, + "step": 40 + }, + { + "epoch": 0.004502525807160114, + "grad_norm": 6.46075439453125, + "learning_rate": 2.1953896816684964e-08, + "loss": 1.5876, + "mean_token_accuracy": 0.5898911952972412, + "num_tokens": 1073829.0, + "step": 41 + }, + { + "epoch": 0.004612343509773776, + "grad_norm": 8.000155448913574, + "learning_rate": 2.2502744237102085e-08, + "loss": 1.4993, + "mean_token_accuracy": 0.6211850047111511, + "num_tokens": 1095730.0, + "step": 42 + }, + { + "epoch": 0.0047221612123874365, + "grad_norm": 7.149877548217773, + "learning_rate": 2.3051591657519207e-08, + "loss": 1.3967, + "mean_token_accuracy": 0.6410075426101685, + "num_tokens": 1120004.0, + "step": 43 + }, + { + "epoch": 0.004831978915001098, + "grad_norm": 8.04874038696289, + "learning_rate": 2.3600439077936336e-08, + "loss": 1.4998, + "mean_token_accuracy": 0.6150715947151184, + "num_tokens": 1141766.0, + "step": 44 + }, + { + "epoch": 0.00494179661761476, + "grad_norm": 8.158447265625, + "learning_rate": 2.4149286498353458e-08, + "loss": 1.5911, + "mean_token_accuracy": 0.5896943807601929, + "num_tokens": 1163196.0, + "step": 45 + }, + { + "epoch": 0.0050516143202284204, + "grad_norm": 7.008297920227051, + "learning_rate": 2.469813391877058e-08, + "loss": 1.4906, + "mean_token_accuracy": 0.6076422929763794, + "num_tokens": 1190415.0, + "step": 46 + }, + { + "epoch": 0.005161432022842082, + "grad_norm": 7.520852088928223, + "learning_rate": 2.5246981339187705e-08, + "loss": 1.3742, + "mean_token_accuracy": 0.6374220848083496, + "num_tokens": 1212589.0, + "step": 47 + }, + { + "epoch": 0.005271249725455744, + "grad_norm": 7.061659336090088, + "learning_rate": 2.579582875960483e-08, + "loss": 1.5168, + "mean_token_accuracy": 0.6034730672836304, + "num_tokens": 1240399.0, + "step": 48 + }, + { + "epoch": 0.005381067428069404, + "grad_norm": 6.296067237854004, + "learning_rate": 2.6344676180021952e-08, + "loss": 1.4032, + "mean_token_accuracy": 0.6294281482696533, + "num_tokens": 1270595.0, + "step": 49 + }, + { + "epoch": 0.005490885130683066, + "grad_norm": 7.0035505294799805, + "learning_rate": 2.6893523600439077e-08, + "loss": 1.4185, + "mean_token_accuracy": 0.6313836574554443, + "num_tokens": 1297738.0, + "step": 50 + }, + { + "epoch": 0.0056007028332967276, + "grad_norm": 8.43636417388916, + "learning_rate": 2.7442371020856202e-08, + "loss": 1.4779, + "mean_token_accuracy": 0.613957405090332, + "num_tokens": 1317699.0, + "step": 51 + }, + { + "epoch": 0.005710520535910389, + "grad_norm": 6.7988505363464355, + "learning_rate": 2.7991218441273324e-08, + "loss": 1.3821, + "mean_token_accuracy": 0.6364263892173767, + "num_tokens": 1343282.0, + "step": 52 + }, + { + "epoch": 0.00582033823852405, + "grad_norm": 7.523954391479492, + "learning_rate": 2.854006586169045e-08, + "loss": 1.4698, + "mean_token_accuracy": 0.6223812699317932, + "num_tokens": 1365867.0, + "step": 53 + }, + { + "epoch": 0.0059301559411377115, + "grad_norm": 7.38977575302124, + "learning_rate": 2.9088913282107574e-08, + "loss": 1.5116, + "mean_token_accuracy": 0.607897937297821, + "num_tokens": 1390260.0, + "step": 54 + }, + { + "epoch": 0.006039973643751373, + "grad_norm": 6.4841532707214355, + "learning_rate": 2.9637760702524696e-08, + "loss": 1.4631, + "mean_token_accuracy": 0.6154390573501587, + "num_tokens": 1421611.0, + "step": 55 + }, + { + "epoch": 0.006149791346365034, + "grad_norm": 5.986716270446777, + "learning_rate": 3.018660812294182e-08, + "loss": 1.5732, + "mean_token_accuracy": 0.5956559777259827, + "num_tokens": 1460080.0, + "step": 56 + }, + { + "epoch": 0.006259609048978695, + "grad_norm": 7.8722004890441895, + "learning_rate": 3.073545554335894e-08, + "loss": 1.5338, + "mean_token_accuracy": 0.5995888710021973, + "num_tokens": 1481970.0, + "step": 57 + }, + { + "epoch": 0.006369426751592357, + "grad_norm": 7.955368518829346, + "learning_rate": 3.128430296377607e-08, + "loss": 1.4886, + "mean_token_accuracy": 0.6131999492645264, + "num_tokens": 1503691.0, + "step": 58 + }, + { + "epoch": 0.006479244454206018, + "grad_norm": 7.52812385559082, + "learning_rate": 3.1833150384193193e-08, + "loss": 1.4887, + "mean_token_accuracy": 0.6172623634338379, + "num_tokens": 1528090.0, + "step": 59 + }, + { + "epoch": 0.006589062156819679, + "grad_norm": 6.561055660247803, + "learning_rate": 3.2381997804610315e-08, + "loss": 1.459, + "mean_token_accuracy": 0.6227182745933533, + "num_tokens": 1557214.0, + "step": 60 + }, + { + "epoch": 0.006698879859433341, + "grad_norm": 6.863100528717041, + "learning_rate": 3.293084522502744e-08, + "loss": 1.4879, + "mean_token_accuracy": 0.6071408987045288, + "num_tokens": 1584825.0, + "step": 61 + }, + { + "epoch": 0.006808697562047002, + "grad_norm": 6.89917516708374, + "learning_rate": 3.3479692645444566e-08, + "loss": 1.5111, + "mean_token_accuracy": 0.6179922819137573, + "num_tokens": 1612567.0, + "step": 62 + }, + { + "epoch": 0.006918515264660663, + "grad_norm": 8.708135604858398, + "learning_rate": 3.402854006586169e-08, + "loss": 1.4635, + "mean_token_accuracy": 0.6245152354240417, + "num_tokens": 1632095.0, + "step": 63 + }, + { + "epoch": 0.007028332967274325, + "grad_norm": 7.851266384124756, + "learning_rate": 3.4577387486278816e-08, + "loss": 1.5115, + "mean_token_accuracy": 0.6115695238113403, + "num_tokens": 1653332.0, + "step": 64 + }, + { + "epoch": 0.007138150669887986, + "grad_norm": 7.528713226318359, + "learning_rate": 3.512623490669594e-08, + "loss": 1.5883, + "mean_token_accuracy": 0.6024693846702576, + "num_tokens": 1677458.0, + "step": 65 + }, + { + "epoch": 0.007247968372501647, + "grad_norm": 7.197704792022705, + "learning_rate": 3.567508232711306e-08, + "loss": 1.4898, + "mean_token_accuracy": 0.6037963628768921, + "num_tokens": 1701768.0, + "step": 66 + }, + { + "epoch": 0.007357786075115309, + "grad_norm": 6.743851661682129, + "learning_rate": 3.622392974753018e-08, + "loss": 1.4926, + "mean_token_accuracy": 0.6111395359039307, + "num_tokens": 1727997.0, + "step": 67 + }, + { + "epoch": 0.0074676037777289695, + "grad_norm": 7.707272052764893, + "learning_rate": 3.677277716794731e-08, + "loss": 1.5555, + "mean_token_accuracy": 0.6019924879074097, + "num_tokens": 1751298.0, + "step": 68 + }, + { + "epoch": 0.007577421480342631, + "grad_norm": 6.819432258605957, + "learning_rate": 3.732162458836443e-08, + "loss": 1.5118, + "mean_token_accuracy": 0.6028596758842468, + "num_tokens": 1778613.0, + "step": 69 + }, + { + "epoch": 0.007687239182956293, + "grad_norm": 6.093446731567383, + "learning_rate": 3.787047200878156e-08, + "loss": 1.3883, + "mean_token_accuracy": 0.6358729600906372, + "num_tokens": 1810153.0, + "step": 70 + }, + { + "epoch": 0.007797056885569954, + "grad_norm": 7.426247596740723, + "learning_rate": 3.841931942919868e-08, + "loss": 1.5774, + "mean_token_accuracy": 0.6002684831619263, + "num_tokens": 1833467.0, + "step": 71 + }, + { + "epoch": 0.007906874588183616, + "grad_norm": 7.207817554473877, + "learning_rate": 3.8968166849615804e-08, + "loss": 1.5973, + "mean_token_accuracy": 0.5897461175918579, + "num_tokens": 1859498.0, + "step": 72 + }, + { + "epoch": 0.008016692290797276, + "grad_norm": 7.225587844848633, + "learning_rate": 3.9517014270032926e-08, + "loss": 1.5594, + "mean_token_accuracy": 0.5943459272384644, + "num_tokens": 1883805.0, + "step": 73 + }, + { + "epoch": 0.008126509993410937, + "grad_norm": 7.270538330078125, + "learning_rate": 4.006586169045005e-08, + "loss": 1.4359, + "mean_token_accuracy": 0.6174066066741943, + "num_tokens": 1907402.0, + "step": 74 + }, + { + "epoch": 0.008236327696024599, + "grad_norm": 7.466093063354492, + "learning_rate": 4.061470911086718e-08, + "loss": 1.5369, + "mean_token_accuracy": 0.5997008681297302, + "num_tokens": 1930296.0, + "step": 75 + }, + { + "epoch": 0.00834614539863826, + "grad_norm": 6.725176811218262, + "learning_rate": 4.1163556531284305e-08, + "loss": 1.4626, + "mean_token_accuracy": 0.6144124865531921, + "num_tokens": 1958064.0, + "step": 76 + }, + { + "epoch": 0.008455963101251922, + "grad_norm": 6.356276988983154, + "learning_rate": 4.1712403951701427e-08, + "loss": 1.4702, + "mean_token_accuracy": 0.6252585649490356, + "num_tokens": 1986399.0, + "step": 77 + }, + { + "epoch": 0.008565780803865584, + "grad_norm": 7.0442328453063965, + "learning_rate": 4.226125137211855e-08, + "loss": 1.4927, + "mean_token_accuracy": 0.6126309633255005, + "num_tokens": 2010413.0, + "step": 78 + }, + { + "epoch": 0.008675598506479244, + "grad_norm": 6.398924350738525, + "learning_rate": 4.281009879253567e-08, + "loss": 1.5005, + "mean_token_accuracy": 0.6030904650688171, + "num_tokens": 2040488.0, + "step": 79 + }, + { + "epoch": 0.008785416209092905, + "grad_norm": 7.594489097595215, + "learning_rate": 4.335894621295279e-08, + "loss": 1.4294, + "mean_token_accuracy": 0.6262327432632446, + "num_tokens": 2064223.0, + "step": 80 + }, + { + "epoch": 0.008895233911706567, + "grad_norm": 6.374741554260254, + "learning_rate": 4.390779363336993e-08, + "loss": 1.5256, + "mean_token_accuracy": 0.6049121618270874, + "num_tokens": 2094926.0, + "step": 81 + }, + { + "epoch": 0.009005051614320228, + "grad_norm": 7.078765869140625, + "learning_rate": 4.445664105378705e-08, + "loss": 1.5554, + "mean_token_accuracy": 0.5897557139396667, + "num_tokens": 2120904.0, + "step": 82 + }, + { + "epoch": 0.00911486931693389, + "grad_norm": 7.882656097412109, + "learning_rate": 4.500548847420417e-08, + "loss": 1.4535, + "mean_token_accuracy": 0.6235867738723755, + "num_tokens": 2141475.0, + "step": 83 + }, + { + "epoch": 0.009224687019547552, + "grad_norm": 6.418519496917725, + "learning_rate": 4.555433589462129e-08, + "loss": 1.4553, + "mean_token_accuracy": 0.6140104532241821, + "num_tokens": 2167715.0, + "step": 84 + }, + { + "epoch": 0.009334504722161213, + "grad_norm": 7.251123905181885, + "learning_rate": 4.6103183315038415e-08, + "loss": 1.5847, + "mean_token_accuracy": 0.5864617824554443, + "num_tokens": 2193244.0, + "step": 85 + }, + { + "epoch": 0.009444322424774873, + "grad_norm": 7.240039348602295, + "learning_rate": 4.6652030735455537e-08, + "loss": 1.4674, + "mean_token_accuracy": 0.6196328997612, + "num_tokens": 2216743.0, + "step": 86 + }, + { + "epoch": 0.009554140127388535, + "grad_norm": 6.579141139984131, + "learning_rate": 4.720087815587267e-08, + "loss": 1.478, + "mean_token_accuracy": 0.60977703332901, + "num_tokens": 2244475.0, + "step": 87 + }, + { + "epoch": 0.009663957830002196, + "grad_norm": 6.593840599060059, + "learning_rate": 4.7749725576289793e-08, + "loss": 1.4699, + "mean_token_accuracy": 0.6082249879837036, + "num_tokens": 2270864.0, + "step": 88 + }, + { + "epoch": 0.009773775532615858, + "grad_norm": 7.724915504455566, + "learning_rate": 4.8298572996706915e-08, + "loss": 1.4186, + "mean_token_accuracy": 0.624160647392273, + "num_tokens": 2291320.0, + "step": 89 + }, + { + "epoch": 0.00988359323522952, + "grad_norm": 5.792138576507568, + "learning_rate": 4.884742041712404e-08, + "loss": 1.3858, + "mean_token_accuracy": 0.634220540523529, + "num_tokens": 2319374.0, + "step": 90 + }, + { + "epoch": 0.009993410937843181, + "grad_norm": 6.368673324584961, + "learning_rate": 4.939626783754116e-08, + "loss": 1.4658, + "mean_token_accuracy": 0.620029866695404, + "num_tokens": 2346205.0, + "step": 91 + }, + { + "epoch": 0.010103228640456841, + "grad_norm": 5.803673267364502, + "learning_rate": 4.994511525795828e-08, + "loss": 1.4078, + "mean_token_accuracy": 0.6299054622650146, + "num_tokens": 2375396.0, + "step": 92 + }, + { + "epoch": 0.010213046343070502, + "grad_norm": 5.868958473205566, + "learning_rate": 5.049396267837541e-08, + "loss": 1.4065, + "mean_token_accuracy": 0.628637433052063, + "num_tokens": 2405734.0, + "step": 93 + }, + { + "epoch": 0.010322864045684164, + "grad_norm": 6.296507358551025, + "learning_rate": 5.104281009879254e-08, + "loss": 1.4609, + "mean_token_accuracy": 0.6101961135864258, + "num_tokens": 2432431.0, + "step": 94 + }, + { + "epoch": 0.010432681748297826, + "grad_norm": 6.463015556335449, + "learning_rate": 5.159165751920966e-08, + "loss": 1.4359, + "mean_token_accuracy": 0.6181434988975525, + "num_tokens": 2460274.0, + "step": 95 + }, + { + "epoch": 0.010542499450911487, + "grad_norm": 6.956847190856934, + "learning_rate": 5.214050493962678e-08, + "loss": 1.4976, + "mean_token_accuracy": 0.6064534187316895, + "num_tokens": 2484767.0, + "step": 96 + }, + { + "epoch": 0.010652317153525149, + "grad_norm": 6.030062675476074, + "learning_rate": 5.2689352360043903e-08, + "loss": 1.4635, + "mean_token_accuracy": 0.610021710395813, + "num_tokens": 2514201.0, + "step": 97 + }, + { + "epoch": 0.010762134856138809, + "grad_norm": 6.843840599060059, + "learning_rate": 5.3238199780461025e-08, + "loss": 1.4691, + "mean_token_accuracy": 0.6222818493843079, + "num_tokens": 2536658.0, + "step": 98 + }, + { + "epoch": 0.01087195255875247, + "grad_norm": 6.963970184326172, + "learning_rate": 5.3787047200878154e-08, + "loss": 1.5017, + "mean_token_accuracy": 0.6069198846817017, + "num_tokens": 2560479.0, + "step": 99 + }, + { + "epoch": 0.010981770261366132, + "grad_norm": 6.589203357696533, + "learning_rate": 5.433589462129528e-08, + "loss": 1.5001, + "mean_token_accuracy": 0.6034599542617798, + "num_tokens": 2586011.0, + "step": 100 + }, + { + "epoch": 0.011091587963979794, + "grad_norm": 6.632607936859131, + "learning_rate": 5.4884742041712404e-08, + "loss": 1.4804, + "mean_token_accuracy": 0.6149855256080627, + "num_tokens": 2611789.0, + "step": 101 + }, + { + "epoch": 0.011201405666593455, + "grad_norm": 6.649400234222412, + "learning_rate": 5.5433589462129526e-08, + "loss": 1.4753, + "mean_token_accuracy": 0.6143431663513184, + "num_tokens": 2636967.0, + "step": 102 + }, + { + "epoch": 0.011311223369207117, + "grad_norm": 7.239114284515381, + "learning_rate": 5.598243688254665e-08, + "loss": 1.5928, + "mean_token_accuracy": 0.5948206186294556, + "num_tokens": 2660630.0, + "step": 103 + }, + { + "epoch": 0.011421041071820778, + "grad_norm": 8.112085342407227, + "learning_rate": 5.653128430296377e-08, + "loss": 1.5071, + "mean_token_accuracy": 0.6075838804244995, + "num_tokens": 2679781.0, + "step": 104 + }, + { + "epoch": 0.011530858774434438, + "grad_norm": 6.989120006561279, + "learning_rate": 5.70801317233809e-08, + "loss": 1.5253, + "mean_token_accuracy": 0.6038544178009033, + "num_tokens": 2704189.0, + "step": 105 + }, + { + "epoch": 0.0116406764770481, + "grad_norm": 6.615602493286133, + "learning_rate": 5.762897914379802e-08, + "loss": 1.3863, + "mean_token_accuracy": 0.6282256245613098, + "num_tokens": 2728547.0, + "step": 106 + }, + { + "epoch": 0.011750494179661761, + "grad_norm": 7.647705554962158, + "learning_rate": 5.817782656421515e-08, + "loss": 1.3795, + "mean_token_accuracy": 0.6350748538970947, + "num_tokens": 2748282.0, + "step": 107 + }, + { + "epoch": 0.011860311882275423, + "grad_norm": 8.133593559265137, + "learning_rate": 5.872667398463227e-08, + "loss": 1.423, + "mean_token_accuracy": 0.6302414536476135, + "num_tokens": 2767045.0, + "step": 108 + }, + { + "epoch": 0.011970129584889085, + "grad_norm": 6.896756172180176, + "learning_rate": 5.927552140504939e-08, + "loss": 1.4301, + "mean_token_accuracy": 0.6221820116043091, + "num_tokens": 2792013.0, + "step": 109 + }, + { + "epoch": 0.012079947287502746, + "grad_norm": 7.547605037689209, + "learning_rate": 5.982436882546651e-08, + "loss": 1.4832, + "mean_token_accuracy": 0.6156705617904663, + "num_tokens": 2814118.0, + "step": 110 + }, + { + "epoch": 0.012189764990116406, + "grad_norm": 6.521027565002441, + "learning_rate": 6.037321624588364e-08, + "loss": 1.5073, + "mean_token_accuracy": 0.6078979969024658, + "num_tokens": 2843003.0, + "step": 111 + }, + { + "epoch": 0.012299582692730068, + "grad_norm": 6.449631690979004, + "learning_rate": 6.092206366630077e-08, + "loss": 1.438, + "mean_token_accuracy": 0.6141859292984009, + "num_tokens": 2870416.0, + "step": 112 + }, + { + "epoch": 0.01240940039534373, + "grad_norm": 7.007363319396973, + "learning_rate": 6.147091108671789e-08, + "loss": 1.4127, + "mean_token_accuracy": 0.6232775449752808, + "num_tokens": 2893723.0, + "step": 113 + }, + { + "epoch": 0.01251921809795739, + "grad_norm": 6.952868938446045, + "learning_rate": 6.201975850713501e-08, + "loss": 1.3676, + "mean_token_accuracy": 0.6378432512283325, + "num_tokens": 2917388.0, + "step": 114 + }, + { + "epoch": 0.012629035800571052, + "grad_norm": 6.349638938903809, + "learning_rate": 6.256860592755214e-08, + "loss": 1.4315, + "mean_token_accuracy": 0.6207841634750366, + "num_tokens": 2944663.0, + "step": 115 + }, + { + "epoch": 0.012738853503184714, + "grad_norm": 6.065464019775391, + "learning_rate": 6.311745334796927e-08, + "loss": 1.3771, + "mean_token_accuracy": 0.6286150217056274, + "num_tokens": 2973469.0, + "step": 116 + }, + { + "epoch": 0.012848671205798374, + "grad_norm": 7.447420120239258, + "learning_rate": 6.366630076838639e-08, + "loss": 1.4271, + "mean_token_accuracy": 0.6176100969314575, + "num_tokens": 2995171.0, + "step": 117 + }, + { + "epoch": 0.012958488908412035, + "grad_norm": 6.910494804382324, + "learning_rate": 6.421514818880352e-08, + "loss": 1.4587, + "mean_token_accuracy": 0.6091834306716919, + "num_tokens": 3020114.0, + "step": 118 + }, + { + "epoch": 0.013068306611025697, + "grad_norm": 7.096086502075195, + "learning_rate": 6.476399560922063e-08, + "loss": 1.4117, + "mean_token_accuracy": 0.6282980442047119, + "num_tokens": 3042750.0, + "step": 119 + }, + { + "epoch": 0.013178124313639359, + "grad_norm": 6.209129810333252, + "learning_rate": 6.531284302963776e-08, + "loss": 1.453, + "mean_token_accuracy": 0.6121433973312378, + "num_tokens": 3071328.0, + "step": 120 + }, + { + "epoch": 0.01328794201625302, + "grad_norm": 5.540743827819824, + "learning_rate": 6.586169045005487e-08, + "loss": 1.2739, + "mean_token_accuracy": 0.6520963311195374, + "num_tokens": 3098402.0, + "step": 121 + }, + { + "epoch": 0.013397759718866682, + "grad_norm": 5.604063034057617, + "learning_rate": 6.6410537870472e-08, + "loss": 1.487, + "mean_token_accuracy": 0.6052236557006836, + "num_tokens": 3128496.0, + "step": 122 + }, + { + "epoch": 0.013507577421480343, + "grad_norm": 5.896417617797852, + "learning_rate": 6.695938529088913e-08, + "loss": 1.3915, + "mean_token_accuracy": 0.6274953484535217, + "num_tokens": 3153443.0, + "step": 123 + }, + { + "epoch": 0.013617395124094003, + "grad_norm": 6.048513412475586, + "learning_rate": 6.750823271130625e-08, + "loss": 1.3762, + "mean_token_accuracy": 0.6289961338043213, + "num_tokens": 3177160.0, + "step": 124 + }, + { + "epoch": 0.013727212826707665, + "grad_norm": 5.489530563354492, + "learning_rate": 6.805708013172338e-08, + "loss": 1.4071, + "mean_token_accuracy": 0.6164348721504211, + "num_tokens": 3205607.0, + "step": 125 + }, + { + "epoch": 0.013837030529321327, + "grad_norm": 5.285059928894043, + "learning_rate": 6.86059275521405e-08, + "loss": 1.3582, + "mean_token_accuracy": 0.6296685338020325, + "num_tokens": 3232675.0, + "step": 126 + }, + { + "epoch": 0.013946848231934988, + "grad_norm": 5.726156711578369, + "learning_rate": 6.915477497255763e-08, + "loss": 1.3351, + "mean_token_accuracy": 0.6392382979393005, + "num_tokens": 3256531.0, + "step": 127 + }, + { + "epoch": 0.01405666593454865, + "grad_norm": 5.6614603996276855, + "learning_rate": 6.970362239297475e-08, + "loss": 1.4126, + "mean_token_accuracy": 0.6196485161781311, + "num_tokens": 3281505.0, + "step": 128 + }, + { + "epoch": 0.014166483637162311, + "grad_norm": 5.65325927734375, + "learning_rate": 7.025246981339188e-08, + "loss": 1.4826, + "mean_token_accuracy": 0.606367826461792, + "num_tokens": 3305320.0, + "step": 129 + }, + { + "epoch": 0.014276301339775971, + "grad_norm": 5.331496238708496, + "learning_rate": 7.0801317233809e-08, + "loss": 1.339, + "mean_token_accuracy": 0.6459329128265381, + "num_tokens": 3331724.0, + "step": 130 + }, + { + "epoch": 0.014386119042389633, + "grad_norm": 4.777414798736572, + "learning_rate": 7.135016465422612e-08, + "loss": 1.3958, + "mean_token_accuracy": 0.6282868385314941, + "num_tokens": 3363052.0, + "step": 131 + }, + { + "epoch": 0.014495936745003294, + "grad_norm": 5.283354759216309, + "learning_rate": 7.189901207464325e-08, + "loss": 1.4644, + "mean_token_accuracy": 0.6105809211730957, + "num_tokens": 3389640.0, + "step": 132 + }, + { + "epoch": 0.014605754447616956, + "grad_norm": 5.33246374130249, + "learning_rate": 7.244785949506036e-08, + "loss": 1.4394, + "mean_token_accuracy": 0.6020941138267517, + "num_tokens": 3413559.0, + "step": 133 + }, + { + "epoch": 0.014715572150230618, + "grad_norm": 4.549609184265137, + "learning_rate": 7.299670691547749e-08, + "loss": 1.4285, + "mean_token_accuracy": 0.616729199886322, + "num_tokens": 3447287.0, + "step": 134 + }, + { + "epoch": 0.01482538985284428, + "grad_norm": 5.088230133056641, + "learning_rate": 7.354555433589462e-08, + "loss": 1.3727, + "mean_token_accuracy": 0.6259582042694092, + "num_tokens": 3473926.0, + "step": 135 + }, + { + "epoch": 0.014935207555457939, + "grad_norm": 5.598947048187256, + "learning_rate": 7.409440175631174e-08, + "loss": 1.4251, + "mean_token_accuracy": 0.6106061935424805, + "num_tokens": 3497069.0, + "step": 136 + }, + { + "epoch": 0.0150450252580716, + "grad_norm": 5.505355358123779, + "learning_rate": 7.464324917672886e-08, + "loss": 1.4546, + "mean_token_accuracy": 0.6090430617332458, + "num_tokens": 3521199.0, + "step": 137 + }, + { + "epoch": 0.015154842960685262, + "grad_norm": 4.968281269073486, + "learning_rate": 7.519209659714599e-08, + "loss": 1.416, + "mean_token_accuracy": 0.6104863882064819, + "num_tokens": 3548692.0, + "step": 138 + }, + { + "epoch": 0.015264660663298924, + "grad_norm": 4.552020072937012, + "learning_rate": 7.574094401756312e-08, + "loss": 1.4193, + "mean_token_accuracy": 0.6216394901275635, + "num_tokens": 3579040.0, + "step": 139 + }, + { + "epoch": 0.015374478365912585, + "grad_norm": 5.056906700134277, + "learning_rate": 7.628979143798024e-08, + "loss": 1.3573, + "mean_token_accuracy": 0.6333290338516235, + "num_tokens": 3604814.0, + "step": 140 + }, + { + "epoch": 0.015484296068526247, + "grad_norm": 5.240631103515625, + "learning_rate": 7.683863885839736e-08, + "loss": 1.3166, + "mean_token_accuracy": 0.6365993022918701, + "num_tokens": 3627270.0, + "step": 141 + }, + { + "epoch": 0.015594113771139909, + "grad_norm": 5.50866174697876, + "learning_rate": 7.738748627881449e-08, + "loss": 1.3277, + "mean_token_accuracy": 0.631197452545166, + "num_tokens": 3650188.0, + "step": 142 + }, + { + "epoch": 0.01570393147375357, + "grad_norm": 5.156933307647705, + "learning_rate": 7.793633369923161e-08, + "loss": 1.3632, + "mean_token_accuracy": 0.6407966613769531, + "num_tokens": 3673529.0, + "step": 143 + }, + { + "epoch": 0.015813749176367232, + "grad_norm": 5.053804874420166, + "learning_rate": 7.848518111964874e-08, + "loss": 1.3013, + "mean_token_accuracy": 0.6357213258743286, + "num_tokens": 3697036.0, + "step": 144 + }, + { + "epoch": 0.01592356687898089, + "grad_norm": 5.130709171295166, + "learning_rate": 7.903402854006585e-08, + "loss": 1.403, + "mean_token_accuracy": 0.6196437478065491, + "num_tokens": 3720702.0, + "step": 145 + }, + { + "epoch": 0.01603338458159455, + "grad_norm": 5.433174133300781, + "learning_rate": 7.958287596048298e-08, + "loss": 1.4518, + "mean_token_accuracy": 0.6055370569229126, + "num_tokens": 3742127.0, + "step": 146 + }, + { + "epoch": 0.016143202284208215, + "grad_norm": 5.055339813232422, + "learning_rate": 8.01317233809001e-08, + "loss": 1.3504, + "mean_token_accuracy": 0.6362881660461426, + "num_tokens": 3766054.0, + "step": 147 + }, + { + "epoch": 0.016253019986821875, + "grad_norm": 4.253845691680908, + "learning_rate": 8.068057080131722e-08, + "loss": 1.3587, + "mean_token_accuracy": 0.6262515187263489, + "num_tokens": 3797469.0, + "step": 148 + }, + { + "epoch": 0.016362837689435538, + "grad_norm": 5.161014556884766, + "learning_rate": 8.122941822173437e-08, + "loss": 1.3411, + "mean_token_accuracy": 0.6380307674407959, + "num_tokens": 3819680.0, + "step": 149 + }, + { + "epoch": 0.016472655392049198, + "grad_norm": 4.550171375274658, + "learning_rate": 8.177826564215148e-08, + "loss": 1.3648, + "mean_token_accuracy": 0.6334359049797058, + "num_tokens": 3847068.0, + "step": 150 + }, + { + "epoch": 0.01658247309466286, + "grad_norm": 4.533081531524658, + "learning_rate": 8.232711306256861e-08, + "loss": 1.4299, + "mean_token_accuracy": 0.6067022681236267, + "num_tokens": 3875808.0, + "step": 151 + }, + { + "epoch": 0.01669229079727652, + "grad_norm": 4.802399635314941, + "learning_rate": 8.287596048298572e-08, + "loss": 1.2878, + "mean_token_accuracy": 0.6496145129203796, + "num_tokens": 3900146.0, + "step": 152 + }, + { + "epoch": 0.01680210849989018, + "grad_norm": 4.912590026855469, + "learning_rate": 8.342480790340285e-08, + "loss": 1.343, + "mean_token_accuracy": 0.638085126876831, + "num_tokens": 3924969.0, + "step": 153 + }, + { + "epoch": 0.016911926202503844, + "grad_norm": 4.50677490234375, + "learning_rate": 8.397365532381998e-08, + "loss": 1.354, + "mean_token_accuracy": 0.6322039365768433, + "num_tokens": 3952295.0, + "step": 154 + }, + { + "epoch": 0.017021743905117504, + "grad_norm": 4.065097808837891, + "learning_rate": 8.45225027442371e-08, + "loss": 1.3999, + "mean_token_accuracy": 0.6204909086227417, + "num_tokens": 3986429.0, + "step": 155 + }, + { + "epoch": 0.017131561607731167, + "grad_norm": 4.427964687347412, + "learning_rate": 8.507135016465423e-08, + "loss": 1.2997, + "mean_token_accuracy": 0.6419214606285095, + "num_tokens": 4011516.0, + "step": 156 + }, + { + "epoch": 0.017241379310344827, + "grad_norm": 4.8929853439331055, + "learning_rate": 8.562019758507134e-08, + "loss": 1.3903, + "mean_token_accuracy": 0.6255956888198853, + "num_tokens": 4039315.0, + "step": 157 + }, + { + "epoch": 0.017351197012958487, + "grad_norm": 4.389498233795166, + "learning_rate": 8.616904500548847e-08, + "loss": 1.2906, + "mean_token_accuracy": 0.6411471962928772, + "num_tokens": 4065021.0, + "step": 158 + }, + { + "epoch": 0.01746101471557215, + "grad_norm": 4.668491363525391, + "learning_rate": 8.671789242590558e-08, + "loss": 1.38, + "mean_token_accuracy": 0.6228213310241699, + "num_tokens": 4090001.0, + "step": 159 + }, + { + "epoch": 0.01757083241818581, + "grad_norm": 5.012348651885986, + "learning_rate": 8.726673984632271e-08, + "loss": 1.4042, + "mean_token_accuracy": 0.6196765899658203, + "num_tokens": 4113208.0, + "step": 160 + }, + { + "epoch": 0.017680650120799474, + "grad_norm": 4.287522315979004, + "learning_rate": 8.781558726673985e-08, + "loss": 1.2924, + "mean_token_accuracy": 0.6492425203323364, + "num_tokens": 4141843.0, + "step": 161 + }, + { + "epoch": 0.017790467823413134, + "grad_norm": 4.4571533203125, + "learning_rate": 8.836443468715697e-08, + "loss": 1.3851, + "mean_token_accuracy": 0.624024510383606, + "num_tokens": 4167956.0, + "step": 162 + }, + { + "epoch": 0.017900285526026797, + "grad_norm": 4.499305248260498, + "learning_rate": 8.89132821075741e-08, + "loss": 1.3742, + "mean_token_accuracy": 0.6225917339324951, + "num_tokens": 4192489.0, + "step": 163 + }, + { + "epoch": 0.018010103228640457, + "grad_norm": 4.485260486602783, + "learning_rate": 8.946212952799121e-08, + "loss": 1.3158, + "mean_token_accuracy": 0.6487626433372498, + "num_tokens": 4220226.0, + "step": 164 + }, + { + "epoch": 0.018119920931254117, + "grad_norm": 4.157156467437744, + "learning_rate": 9.001097694840834e-08, + "loss": 1.3211, + "mean_token_accuracy": 0.6409006118774414, + "num_tokens": 4249816.0, + "step": 165 + }, + { + "epoch": 0.01822973863386778, + "grad_norm": 4.089326858520508, + "learning_rate": 9.055982436882546e-08, + "loss": 1.4119, + "mean_token_accuracy": 0.616904616355896, + "num_tokens": 4279590.0, + "step": 166 + }, + { + "epoch": 0.01833955633648144, + "grad_norm": 4.493042469024658, + "learning_rate": 9.110867178924259e-08, + "loss": 1.4254, + "mean_token_accuracy": 0.6129616498947144, + "num_tokens": 4307540.0, + "step": 167 + }, + { + "epoch": 0.018449374039095103, + "grad_norm": 5.372011661529541, + "learning_rate": 9.165751920965971e-08, + "loss": 1.28, + "mean_token_accuracy": 0.6400719285011292, + "num_tokens": 4327338.0, + "step": 168 + }, + { + "epoch": 0.018559191741708763, + "grad_norm": 4.700902938842773, + "learning_rate": 9.220636663007683e-08, + "loss": 1.3767, + "mean_token_accuracy": 0.6194136142730713, + "num_tokens": 4349359.0, + "step": 169 + }, + { + "epoch": 0.018669009444322426, + "grad_norm": 6.162675857543945, + "learning_rate": 9.275521405049396e-08, + "loss": 1.2648, + "mean_token_accuracy": 0.6511425375938416, + "num_tokens": 4367535.0, + "step": 170 + }, + { + "epoch": 0.018778827146936086, + "grad_norm": 5.458215236663818, + "learning_rate": 9.330406147091107e-08, + "loss": 1.334, + "mean_token_accuracy": 0.636917233467102, + "num_tokens": 4386241.0, + "step": 171 + }, + { + "epoch": 0.018888644849549746, + "grad_norm": 4.320774078369141, + "learning_rate": 9.38529088913282e-08, + "loss": 1.3713, + "mean_token_accuracy": 0.6191563606262207, + "num_tokens": 4415298.0, + "step": 172 + }, + { + "epoch": 0.01899846255216341, + "grad_norm": 4.249129295349121, + "learning_rate": 9.440175631174534e-08, + "loss": 1.3856, + "mean_token_accuracy": 0.6195068359375, + "num_tokens": 4442068.0, + "step": 173 + }, + { + "epoch": 0.01910828025477707, + "grad_norm": 4.011239528656006, + "learning_rate": 9.495060373216246e-08, + "loss": 1.2679, + "mean_token_accuracy": 0.6479382514953613, + "num_tokens": 4471746.0, + "step": 174 + }, + { + "epoch": 0.019218097957390733, + "grad_norm": 4.626778602600098, + "learning_rate": 9.549945115257959e-08, + "loss": 1.3159, + "mean_token_accuracy": 0.6401896476745605, + "num_tokens": 4492563.0, + "step": 175 + }, + { + "epoch": 0.019327915660004392, + "grad_norm": 3.883632183074951, + "learning_rate": 9.60482985729967e-08, + "loss": 1.3351, + "mean_token_accuracy": 0.6377078890800476, + "num_tokens": 4521774.0, + "step": 176 + }, + { + "epoch": 0.019437733362618052, + "grad_norm": 4.481681823730469, + "learning_rate": 9.659714599341383e-08, + "loss": 1.3297, + "mean_token_accuracy": 0.6364642977714539, + "num_tokens": 4544999.0, + "step": 177 + }, + { + "epoch": 0.019547551065231716, + "grad_norm": 3.7309019565582275, + "learning_rate": 9.714599341383095e-08, + "loss": 1.3356, + "mean_token_accuracy": 0.635826826095581, + "num_tokens": 4574179.0, + "step": 178 + }, + { + "epoch": 0.019657368767845376, + "grad_norm": 3.9440622329711914, + "learning_rate": 9.769484083424807e-08, + "loss": 1.2956, + "mean_token_accuracy": 0.6459222435951233, + "num_tokens": 4602947.0, + "step": 179 + }, + { + "epoch": 0.01976718647045904, + "grad_norm": 4.452601432800293, + "learning_rate": 9.82436882546652e-08, + "loss": 1.3589, + "mean_token_accuracy": 0.6250158548355103, + "num_tokens": 4625393.0, + "step": 180 + }, + { + "epoch": 0.0198770041730727, + "grad_norm": 3.8140506744384766, + "learning_rate": 9.879253567508232e-08, + "loss": 1.2991, + "mean_token_accuracy": 0.6462627649307251, + "num_tokens": 4657005.0, + "step": 181 + }, + { + "epoch": 0.019986821875686362, + "grad_norm": 4.07537317276001, + "learning_rate": 9.934138309549945e-08, + "loss": 1.2853, + "mean_token_accuracy": 0.6439248919487, + "num_tokens": 4687945.0, + "step": 182 + }, + { + "epoch": 0.020096639578300022, + "grad_norm": 4.36903715133667, + "learning_rate": 9.989023051591656e-08, + "loss": 1.3164, + "mean_token_accuracy": 0.633952260017395, + "num_tokens": 4717524.0, + "step": 183 + }, + { + "epoch": 0.020206457280913682, + "grad_norm": 4.217543125152588, + "learning_rate": 1.0043907793633369e-07, + "loss": 1.3242, + "mean_token_accuracy": 0.6362271904945374, + "num_tokens": 4748321.0, + "step": 184 + }, + { + "epoch": 0.020316274983527345, + "grad_norm": 4.097868919372559, + "learning_rate": 1.0098792535675082e-07, + "loss": 1.2896, + "mean_token_accuracy": 0.6455677151679993, + "num_tokens": 4772602.0, + "step": 185 + }, + { + "epoch": 0.020426092686141005, + "grad_norm": 5.652535915374756, + "learning_rate": 1.0153677277716795e-07, + "loss": 1.2941, + "mean_token_accuracy": 0.6463623642921448, + "num_tokens": 4789752.0, + "step": 186 + }, + { + "epoch": 0.02053591038875467, + "grad_norm": 4.489101886749268, + "learning_rate": 1.0208562019758508e-07, + "loss": 1.2963, + "mean_token_accuracy": 0.6426504850387573, + "num_tokens": 4814008.0, + "step": 187 + }, + { + "epoch": 0.020645728091368328, + "grad_norm": 4.50019645690918, + "learning_rate": 1.0263446761800219e-07, + "loss": 1.3132, + "mean_token_accuracy": 0.6382004022598267, + "num_tokens": 4838460.0, + "step": 188 + }, + { + "epoch": 0.02075554579398199, + "grad_norm": 4.90104341506958, + "learning_rate": 1.0318331503841932e-07, + "loss": 1.3067, + "mean_token_accuracy": 0.641255259513855, + "num_tokens": 4864657.0, + "step": 189 + }, + { + "epoch": 0.02086536349659565, + "grad_norm": 4.555761337280273, + "learning_rate": 1.0373216245883643e-07, + "loss": 1.4593, + "mean_token_accuracy": 0.6112907528877258, + "num_tokens": 4888225.0, + "step": 190 + }, + { + "epoch": 0.02097518119920931, + "grad_norm": 3.6093130111694336, + "learning_rate": 1.0428100987925356e-07, + "loss": 1.3107, + "mean_token_accuracy": 0.6366462707519531, + "num_tokens": 4919280.0, + "step": 191 + }, + { + "epoch": 0.021084998901822975, + "grad_norm": 3.9960079193115234, + "learning_rate": 1.0482985729967068e-07, + "loss": 1.3309, + "mean_token_accuracy": 0.6317435503005981, + "num_tokens": 4947590.0, + "step": 192 + }, + { + "epoch": 0.021194816604436634, + "grad_norm": 3.7334022521972656, + "learning_rate": 1.0537870472008781e-07, + "loss": 1.3458, + "mean_token_accuracy": 0.6311653256416321, + "num_tokens": 4979902.0, + "step": 193 + }, + { + "epoch": 0.021304634307050298, + "grad_norm": 4.218454837799072, + "learning_rate": 1.0592755214050494e-07, + "loss": 1.4156, + "mean_token_accuracy": 0.6142665147781372, + "num_tokens": 5003312.0, + "step": 194 + }, + { + "epoch": 0.021414452009663958, + "grad_norm": 4.728745460510254, + "learning_rate": 1.0647639956092205e-07, + "loss": 1.2817, + "mean_token_accuracy": 0.6446450352668762, + "num_tokens": 5025043.0, + "step": 195 + }, + { + "epoch": 0.021524269712277617, + "grad_norm": 4.458101272583008, + "learning_rate": 1.0702524698133918e-07, + "loss": 1.3084, + "mean_token_accuracy": 0.6437536478042603, + "num_tokens": 5048012.0, + "step": 196 + }, + { + "epoch": 0.02163408741489128, + "grad_norm": 5.092242240905762, + "learning_rate": 1.0757409440175631e-07, + "loss": 1.286, + "mean_token_accuracy": 0.643911600112915, + "num_tokens": 5075320.0, + "step": 197 + }, + { + "epoch": 0.02174390511750494, + "grad_norm": 4.848008632659912, + "learning_rate": 1.0812294182217344e-07, + "loss": 1.2908, + "mean_token_accuracy": 0.6432056427001953, + "num_tokens": 5099676.0, + "step": 198 + }, + { + "epoch": 0.021853722820118604, + "grad_norm": 6.06216287612915, + "learning_rate": 1.0867178924259056e-07, + "loss": 1.221, + "mean_token_accuracy": 0.6526443958282471, + "num_tokens": 5116120.0, + "step": 199 + }, + { + "epoch": 0.021963540522732264, + "grad_norm": 4.8727288246154785, + "learning_rate": 1.0922063666300768e-07, + "loss": 1.3389, + "mean_token_accuracy": 0.6262339353561401, + "num_tokens": 5143339.0, + "step": 200 + }, + { + "epoch": 0.022073358225345927, + "grad_norm": 4.435262680053711, + "learning_rate": 1.0976948408342481e-07, + "loss": 1.3286, + "mean_token_accuracy": 0.6269840002059937, + "num_tokens": 5169825.0, + "step": 201 + }, + { + "epoch": 0.022183175927959587, + "grad_norm": 4.898035526275635, + "learning_rate": 1.1031833150384192e-07, + "loss": 1.373, + "mean_token_accuracy": 0.6183314323425293, + "num_tokens": 5199228.0, + "step": 202 + }, + { + "epoch": 0.022292993630573247, + "grad_norm": 4.360095977783203, + "learning_rate": 1.1086717892425905e-07, + "loss": 1.1454, + "mean_token_accuracy": 0.6683323383331299, + "num_tokens": 5223504.0, + "step": 203 + }, + { + "epoch": 0.02240281133318691, + "grad_norm": 5.363147258758545, + "learning_rate": 1.1141602634467617e-07, + "loss": 1.2837, + "mean_token_accuracy": 0.6414899826049805, + "num_tokens": 5243898.0, + "step": 204 + }, + { + "epoch": 0.02251262903580057, + "grad_norm": 4.210151195526123, + "learning_rate": 1.119648737650933e-07, + "loss": 1.341, + "mean_token_accuracy": 0.6327681541442871, + "num_tokens": 5271421.0, + "step": 205 + }, + { + "epoch": 0.022622446738414233, + "grad_norm": 4.277538776397705, + "learning_rate": 1.1251372118551042e-07, + "loss": 1.3093, + "mean_token_accuracy": 0.6344760656356812, + "num_tokens": 5298353.0, + "step": 206 + }, + { + "epoch": 0.022732264441027893, + "grad_norm": 4.981099605560303, + "learning_rate": 1.1306256860592754e-07, + "loss": 1.3256, + "mean_token_accuracy": 0.6370819807052612, + "num_tokens": 5324309.0, + "step": 207 + }, + { + "epoch": 0.022842082143641557, + "grad_norm": 4.576216220855713, + "learning_rate": 1.1361141602634467e-07, + "loss": 1.3206, + "mean_token_accuracy": 0.6366074681282043, + "num_tokens": 5351694.0, + "step": 208 + }, + { + "epoch": 0.022951899846255217, + "grad_norm": 4.815907001495361, + "learning_rate": 1.141602634467618e-07, + "loss": 1.314, + "mean_token_accuracy": 0.6379421949386597, + "num_tokens": 5379942.0, + "step": 209 + }, + { + "epoch": 0.023061717548868876, + "grad_norm": 4.647455215454102, + "learning_rate": 1.1470911086717892e-07, + "loss": 1.3732, + "mean_token_accuracy": 0.6267041563987732, + "num_tokens": 5410895.0, + "step": 210 + }, + { + "epoch": 0.02317153525148254, + "grad_norm": 4.634090900421143, + "learning_rate": 1.1525795828759604e-07, + "loss": 1.2368, + "mean_token_accuracy": 0.6540911197662354, + "num_tokens": 5437105.0, + "step": 211 + }, + { + "epoch": 0.0232813529540962, + "grad_norm": 5.983651161193848, + "learning_rate": 1.1580680570801317e-07, + "loss": 1.3607, + "mean_token_accuracy": 0.6213585138320923, + "num_tokens": 5469077.0, + "step": 212 + }, + { + "epoch": 0.023391170656709863, + "grad_norm": 4.671662330627441, + "learning_rate": 1.163556531284303e-07, + "loss": 1.3571, + "mean_token_accuracy": 0.6379916667938232, + "num_tokens": 5497064.0, + "step": 213 + }, + { + "epoch": 0.023500988359323523, + "grad_norm": 5.698843002319336, + "learning_rate": 1.1690450054884741e-07, + "loss": 1.2419, + "mean_token_accuracy": 0.6544818878173828, + "num_tokens": 5519268.0, + "step": 214 + }, + { + "epoch": 0.023610806061937183, + "grad_norm": 4.423768043518066, + "learning_rate": 1.1745334796926454e-07, + "loss": 1.284, + "mean_token_accuracy": 0.6446283459663391, + "num_tokens": 5549938.0, + "step": 215 + }, + { + "epoch": 0.023720623764550846, + "grad_norm": 6.146470546722412, + "learning_rate": 1.1800219538968166e-07, + "loss": 1.231, + "mean_token_accuracy": 0.6541129350662231, + "num_tokens": 5569781.0, + "step": 216 + }, + { + "epoch": 0.023830441467164506, + "grad_norm": 4.466990947723389, + "learning_rate": 1.1855104281009878e-07, + "loss": 1.3494, + "mean_token_accuracy": 0.6321786046028137, + "num_tokens": 5601634.0, + "step": 217 + }, + { + "epoch": 0.02394025916977817, + "grad_norm": 4.352505683898926, + "learning_rate": 1.1909989023051591e-07, + "loss": 1.3023, + "mean_token_accuracy": 0.6435118913650513, + "num_tokens": 5630707.0, + "step": 218 + }, + { + "epoch": 0.02405007687239183, + "grad_norm": 5.206647872924805, + "learning_rate": 1.1964873765093303e-07, + "loss": 1.2851, + "mean_token_accuracy": 0.6430725455284119, + "num_tokens": 5658683.0, + "step": 219 + }, + { + "epoch": 0.024159894575005492, + "grad_norm": 5.744487285614014, + "learning_rate": 1.2019758507135017e-07, + "loss": 1.2754, + "mean_token_accuracy": 0.6434608697891235, + "num_tokens": 5681721.0, + "step": 220 + }, + { + "epoch": 0.024269712277619152, + "grad_norm": 5.298373222351074, + "learning_rate": 1.2074643249176729e-07, + "loss": 1.2476, + "mean_token_accuracy": 0.6532942056655884, + "num_tokens": 5713100.0, + "step": 221 + }, + { + "epoch": 0.024379529980232812, + "grad_norm": 7.067374229431152, + "learning_rate": 1.212952799121844e-07, + "loss": 1.283, + "mean_token_accuracy": 0.6502127647399902, + "num_tokens": 5738493.0, + "step": 222 + }, + { + "epoch": 0.024489347682846475, + "grad_norm": 6.15041971206665, + "learning_rate": 1.2184412733260154e-07, + "loss": 1.2724, + "mean_token_accuracy": 0.6465886831283569, + "num_tokens": 5760019.0, + "step": 223 + }, + { + "epoch": 0.024599165385460135, + "grad_norm": 5.239870071411133, + "learning_rate": 1.2239297475301866e-07, + "loss": 1.1585, + "mean_token_accuracy": 0.6658585071563721, + "num_tokens": 5781879.0, + "step": 224 + }, + { + "epoch": 0.0247089830880738, + "grad_norm": 7.486237049102783, + "learning_rate": 1.2294182217343577e-07, + "loss": 1.236, + "mean_token_accuracy": 0.6460365056991577, + "num_tokens": 5802252.0, + "step": 225 + }, + { + "epoch": 0.02481880079068746, + "grad_norm": 6.074740886688232, + "learning_rate": 1.2349066959385291e-07, + "loss": 1.2328, + "mean_token_accuracy": 0.6529794931411743, + "num_tokens": 5830077.0, + "step": 226 + }, + { + "epoch": 0.024928618493301122, + "grad_norm": 5.0144362449646, + "learning_rate": 1.2403951701427003e-07, + "loss": 1.3152, + "mean_token_accuracy": 0.6211760640144348, + "num_tokens": 5858607.0, + "step": 227 + }, + { + "epoch": 0.02503843619591478, + "grad_norm": 5.98635196685791, + "learning_rate": 1.2458836443468714e-07, + "loss": 1.2156, + "mean_token_accuracy": 0.6558916568756104, + "num_tokens": 5881866.0, + "step": 228 + }, + { + "epoch": 0.02514825389852844, + "grad_norm": 5.451389312744141, + "learning_rate": 1.2513721185510429e-07, + "loss": 1.1741, + "mean_token_accuracy": 0.6717559695243835, + "num_tokens": 5908367.0, + "step": 229 + }, + { + "epoch": 0.025258071601142105, + "grad_norm": 4.761264324188232, + "learning_rate": 1.256860592755214e-07, + "loss": 1.2176, + "mean_token_accuracy": 0.6615591049194336, + "num_tokens": 5937021.0, + "step": 230 + }, + { + "epoch": 0.025367889303755765, + "grad_norm": 4.658876419067383, + "learning_rate": 1.2623490669593854e-07, + "loss": 1.2583, + "mean_token_accuracy": 0.6505842208862305, + "num_tokens": 5965943.0, + "step": 231 + }, + { + "epoch": 0.025477707006369428, + "grad_norm": 6.7758097648620605, + "learning_rate": 1.2678375411635563e-07, + "loss": 1.3371, + "mean_token_accuracy": 0.6423532962799072, + "num_tokens": 5987770.0, + "step": 232 + }, + { + "epoch": 0.025587524708983088, + "grad_norm": 6.025655746459961, + "learning_rate": 1.2733260153677277e-07, + "loss": 1.3109, + "mean_token_accuracy": 0.6290343403816223, + "num_tokens": 6009631.0, + "step": 233 + }, + { + "epoch": 0.025697342411596748, + "grad_norm": 6.3221564292907715, + "learning_rate": 1.278814489571899e-07, + "loss": 1.2952, + "mean_token_accuracy": 0.6361500024795532, + "num_tokens": 6032628.0, + "step": 234 + }, + { + "epoch": 0.02580716011421041, + "grad_norm": 5.132676124572754, + "learning_rate": 1.2843029637760703e-07, + "loss": 1.2035, + "mean_token_accuracy": 0.67081618309021, + "num_tokens": 6063196.0, + "step": 235 + }, + { + "epoch": 0.02591697781682407, + "grad_norm": 5.549376964569092, + "learning_rate": 1.2897914379802412e-07, + "loss": 1.2407, + "mean_token_accuracy": 0.6494331359863281, + "num_tokens": 6088918.0, + "step": 236 + }, + { + "epoch": 0.026026795519437734, + "grad_norm": 5.824056148529053, + "learning_rate": 1.2952799121844126e-07, + "loss": 1.1479, + "mean_token_accuracy": 0.6774070262908936, + "num_tokens": 6114554.0, + "step": 237 + }, + { + "epoch": 0.026136613222051394, + "grad_norm": 5.822218894958496, + "learning_rate": 1.300768386388584e-07, + "loss": 1.1994, + "mean_token_accuracy": 0.6645663380622864, + "num_tokens": 6140547.0, + "step": 238 + }, + { + "epoch": 0.026246430924665057, + "grad_norm": 6.2488861083984375, + "learning_rate": 1.3062568605927552e-07, + "loss": 1.2755, + "mean_token_accuracy": 0.6453362107276917, + "num_tokens": 6167294.0, + "step": 239 + }, + { + "epoch": 0.026356248627278717, + "grad_norm": 5.4994730949401855, + "learning_rate": 1.3117453347969266e-07, + "loss": 1.2399, + "mean_token_accuracy": 0.6491696834564209, + "num_tokens": 6190001.0, + "step": 240 + }, + { + "epoch": 0.026466066329892377, + "grad_norm": 6.886782646179199, + "learning_rate": 1.3172338090010975e-07, + "loss": 1.2226, + "mean_token_accuracy": 0.6587322950363159, + "num_tokens": 6211107.0, + "step": 241 + }, + { + "epoch": 0.02657588403250604, + "grad_norm": 5.109975337982178, + "learning_rate": 1.322722283205269e-07, + "loss": 1.3227, + "mean_token_accuracy": 0.6288667917251587, + "num_tokens": 6236860.0, + "step": 242 + }, + { + "epoch": 0.0266857017351197, + "grad_norm": 6.138133525848389, + "learning_rate": 1.32821075740944e-07, + "loss": 1.2421, + "mean_token_accuracy": 0.6554150581359863, + "num_tokens": 6266773.0, + "step": 243 + }, + { + "epoch": 0.026795519437733364, + "grad_norm": 5.4781012535095215, + "learning_rate": 1.3336992316136115e-07, + "loss": 1.3819, + "mean_token_accuracy": 0.6152228116989136, + "num_tokens": 6293107.0, + "step": 244 + }, + { + "epoch": 0.026905337140347024, + "grad_norm": 5.5572710037231445, + "learning_rate": 1.3391877058177826e-07, + "loss": 1.2166, + "mean_token_accuracy": 0.677168607711792, + "num_tokens": 6317175.0, + "step": 245 + }, + { + "epoch": 0.027015154842960687, + "grad_norm": 6.219461917877197, + "learning_rate": 1.3446761800219538e-07, + "loss": 1.3515, + "mean_token_accuracy": 0.6290022134780884, + "num_tokens": 6341521.0, + "step": 246 + }, + { + "epoch": 0.027124972545574347, + "grad_norm": 7.0836873054504395, + "learning_rate": 1.350164654226125e-07, + "loss": 1.2224, + "mean_token_accuracy": 0.6513441801071167, + "num_tokens": 6368580.0, + "step": 247 + }, + { + "epoch": 0.027234790248188007, + "grad_norm": 6.493309020996094, + "learning_rate": 1.3556531284302963e-07, + "loss": 1.1954, + "mean_token_accuracy": 0.6601212024688721, + "num_tokens": 6389366.0, + "step": 248 + }, + { + "epoch": 0.02734460795080167, + "grad_norm": 5.601600170135498, + "learning_rate": 1.3611416026344675e-07, + "loss": 1.2053, + "mean_token_accuracy": 0.6565067768096924, + "num_tokens": 6412533.0, + "step": 249 + }, + { + "epoch": 0.02745442565341533, + "grad_norm": 6.254323959350586, + "learning_rate": 1.366630076838639e-07, + "loss": 1.3011, + "mean_token_accuracy": 0.6403510570526123, + "num_tokens": 6440760.0, + "step": 250 + }, + { + "epoch": 0.027564243356028993, + "grad_norm": 5.175616264343262, + "learning_rate": 1.37211855104281e-07, + "loss": 1.2002, + "mean_token_accuracy": 0.6569468379020691, + "num_tokens": 6464465.0, + "step": 251 + }, + { + "epoch": 0.027674061058642653, + "grad_norm": 5.275678634643555, + "learning_rate": 1.3776070252469812e-07, + "loss": 1.2217, + "mean_token_accuracy": 0.6492674946784973, + "num_tokens": 6487767.0, + "step": 252 + }, + { + "epoch": 0.027783878761256313, + "grad_norm": 5.8388776779174805, + "learning_rate": 1.3830954994511526e-07, + "loss": 1.2733, + "mean_token_accuracy": 0.6388698220252991, + "num_tokens": 6509454.0, + "step": 253 + }, + { + "epoch": 0.027893696463869976, + "grad_norm": 7.513961315155029, + "learning_rate": 1.3885839736553238e-07, + "loss": 1.1192, + "mean_token_accuracy": 0.6787033081054688, + "num_tokens": 6530315.0, + "step": 254 + }, + { + "epoch": 0.028003514166483636, + "grad_norm": 4.780346870422363, + "learning_rate": 1.394072447859495e-07, + "loss": 1.2205, + "mean_token_accuracy": 0.6531215906143188, + "num_tokens": 6557715.0, + "step": 255 + }, + { + "epoch": 0.0281133318690973, + "grad_norm": 5.276611328125, + "learning_rate": 1.399560922063666e-07, + "loss": 1.193, + "mean_token_accuracy": 0.6642729043960571, + "num_tokens": 6580370.0, + "step": 256 + }, + { + "epoch": 0.02822314957171096, + "grad_norm": 5.5757060050964355, + "learning_rate": 1.4050493962678375e-07, + "loss": 1.299, + "mean_token_accuracy": 0.6378278732299805, + "num_tokens": 6606511.0, + "step": 257 + }, + { + "epoch": 0.028332967274324623, + "grad_norm": 4.912115573883057, + "learning_rate": 1.4105378704720087e-07, + "loss": 1.2252, + "mean_token_accuracy": 0.6614087820053101, + "num_tokens": 6631579.0, + "step": 258 + }, + { + "epoch": 0.028442784976938282, + "grad_norm": 6.1575164794921875, + "learning_rate": 1.41602634467618e-07, + "loss": 1.257, + "mean_token_accuracy": 0.6631867289543152, + "num_tokens": 6653945.0, + "step": 259 + }, + { + "epoch": 0.028552602679551942, + "grad_norm": 6.1414618492126465, + "learning_rate": 1.421514818880351e-07, + "loss": 1.2589, + "mean_token_accuracy": 0.6489673852920532, + "num_tokens": 6676326.0, + "step": 260 + }, + { + "epoch": 0.028662420382165606, + "grad_norm": 5.455733776092529, + "learning_rate": 1.4270032930845224e-07, + "loss": 1.2097, + "mean_token_accuracy": 0.6586052775382996, + "num_tokens": 6706674.0, + "step": 261 + }, + { + "epoch": 0.028772238084779266, + "grad_norm": 5.172214984893799, + "learning_rate": 1.4324917672886938e-07, + "loss": 1.2781, + "mean_token_accuracy": 0.639980673789978, + "num_tokens": 6731802.0, + "step": 262 + }, + { + "epoch": 0.02888205578739293, + "grad_norm": 5.551685333251953, + "learning_rate": 1.437980241492865e-07, + "loss": 1.1878, + "mean_token_accuracy": 0.662988543510437, + "num_tokens": 6756907.0, + "step": 263 + }, + { + "epoch": 0.02899187349000659, + "grad_norm": 6.414708614349365, + "learning_rate": 1.4434687156970364e-07, + "loss": 1.2689, + "mean_token_accuracy": 0.6421482563018799, + "num_tokens": 6782127.0, + "step": 264 + }, + { + "epoch": 0.029101691192620252, + "grad_norm": 6.088720321655273, + "learning_rate": 1.4489571899012073e-07, + "loss": 1.3193, + "mean_token_accuracy": 0.6342250108718872, + "num_tokens": 6803864.0, + "step": 265 + }, + { + "epoch": 0.029211508895233912, + "grad_norm": 6.5865912437438965, + "learning_rate": 1.4544456641053787e-07, + "loss": 1.1845, + "mean_token_accuracy": 0.6677565574645996, + "num_tokens": 6830156.0, + "step": 266 + }, + { + "epoch": 0.029321326597847572, + "grad_norm": 6.042848587036133, + "learning_rate": 1.4599341383095498e-07, + "loss": 1.3256, + "mean_token_accuracy": 0.6403924226760864, + "num_tokens": 6855566.0, + "step": 267 + }, + { + "epoch": 0.029431144300461235, + "grad_norm": 5.499892711639404, + "learning_rate": 1.4654226125137212e-07, + "loss": 1.2469, + "mean_token_accuracy": 0.6536927223205566, + "num_tokens": 6882686.0, + "step": 268 + }, + { + "epoch": 0.029540962003074895, + "grad_norm": 5.411120891571045, + "learning_rate": 1.4709110867178924e-07, + "loss": 1.2286, + "mean_token_accuracy": 0.654003918170929, + "num_tokens": 6909814.0, + "step": 269 + }, + { + "epoch": 0.02965077970568856, + "grad_norm": 5.656541347503662, + "learning_rate": 1.4763995609220636e-07, + "loss": 1.2665, + "mean_token_accuracy": 0.6471383571624756, + "num_tokens": 6940032.0, + "step": 270 + }, + { + "epoch": 0.029760597408302218, + "grad_norm": 5.608570098876953, + "learning_rate": 1.4818880351262347e-07, + "loss": 1.2895, + "mean_token_accuracy": 0.6420177221298218, + "num_tokens": 6966911.0, + "step": 271 + }, + { + "epoch": 0.029870415110915878, + "grad_norm": 5.659779071807861, + "learning_rate": 1.487376509330406e-07, + "loss": 1.1734, + "mean_token_accuracy": 0.6731195449829102, + "num_tokens": 6991133.0, + "step": 272 + }, + { + "epoch": 0.02998023281352954, + "grad_norm": 6.318817138671875, + "learning_rate": 1.4928649835345773e-07, + "loss": 1.3125, + "mean_token_accuracy": 0.644629955291748, + "num_tokens": 7017490.0, + "step": 273 + }, + { + "epoch": 0.0300900505161432, + "grad_norm": 4.527459621429443, + "learning_rate": 1.4983534577387484e-07, + "loss": 1.1875, + "mean_token_accuracy": 0.6676506996154785, + "num_tokens": 7044227.0, + "step": 274 + }, + { + "epoch": 0.030199868218756865, + "grad_norm": 4.777981281280518, + "learning_rate": 1.5038419319429198e-07, + "loss": 1.2157, + "mean_token_accuracy": 0.6462973356246948, + "num_tokens": 7069938.0, + "step": 275 + }, + { + "epoch": 0.030309685921370524, + "grad_norm": 5.3882527351379395, + "learning_rate": 1.509330406147091e-07, + "loss": 1.1078, + "mean_token_accuracy": 0.6789013147354126, + "num_tokens": 7093980.0, + "step": 276 + }, + { + "epoch": 0.030419503623984188, + "grad_norm": 5.812991619110107, + "learning_rate": 1.5148188803512624e-07, + "loss": 1.2645, + "mean_token_accuracy": 0.6461018919944763, + "num_tokens": 7118991.0, + "step": 277 + }, + { + "epoch": 0.030529321326597848, + "grad_norm": 5.083370208740234, + "learning_rate": 1.5203073545554336e-07, + "loss": 1.2499, + "mean_token_accuracy": 0.6506079435348511, + "num_tokens": 7145223.0, + "step": 278 + }, + { + "epoch": 0.030639139029211507, + "grad_norm": 6.661009311676025, + "learning_rate": 1.5257958287596047e-07, + "loss": 1.1967, + "mean_token_accuracy": 0.6619991660118103, + "num_tokens": 7168875.0, + "step": 279 + }, + { + "epoch": 0.03074895673182517, + "grad_norm": 6.076034069061279, + "learning_rate": 1.531284302963776e-07, + "loss": 1.1871, + "mean_token_accuracy": 0.6669723987579346, + "num_tokens": 7194079.0, + "step": 280 + }, + { + "epoch": 0.03085877443443883, + "grad_norm": 5.371271133422852, + "learning_rate": 1.5367727771679473e-07, + "loss": 1.3186, + "mean_token_accuracy": 0.6299723982810974, + "num_tokens": 7224994.0, + "step": 281 + }, + { + "epoch": 0.030968592137052494, + "grad_norm": 6.095330715179443, + "learning_rate": 1.5422612513721184e-07, + "loss": 1.2635, + "mean_token_accuracy": 0.6472434401512146, + "num_tokens": 7254627.0, + "step": 282 + }, + { + "epoch": 0.031078409839666154, + "grad_norm": 4.764731407165527, + "learning_rate": 1.5477497255762899e-07, + "loss": 1.217, + "mean_token_accuracy": 0.6555144190788269, + "num_tokens": 7284071.0, + "step": 283 + }, + { + "epoch": 0.031188227542279817, + "grad_norm": 4.360300064086914, + "learning_rate": 1.5532381997804607e-07, + "loss": 1.2589, + "mean_token_accuracy": 0.6478551626205444, + "num_tokens": 7310382.0, + "step": 284 + }, + { + "epoch": 0.03129804524489348, + "grad_norm": 6.163324356079102, + "learning_rate": 1.5587266739846322e-07, + "loss": 1.3088, + "mean_token_accuracy": 0.6459208726882935, + "num_tokens": 7333864.0, + "step": 285 + }, + { + "epoch": 0.03140786294750714, + "grad_norm": 6.4567718505859375, + "learning_rate": 1.5642151481888036e-07, + "loss": 1.1348, + "mean_token_accuracy": 0.6727630496025085, + "num_tokens": 7353697.0, + "step": 286 + }, + { + "epoch": 0.0315176806501208, + "grad_norm": 6.401641368865967, + "learning_rate": 1.5697036223929747e-07, + "loss": 1.1886, + "mean_token_accuracy": 0.6628429889678955, + "num_tokens": 7374155.0, + "step": 287 + }, + { + "epoch": 0.031627498352734464, + "grad_norm": 5.109067440032959, + "learning_rate": 1.5751920965971461e-07, + "loss": 1.2452, + "mean_token_accuracy": 0.6523860692977905, + "num_tokens": 7407210.0, + "step": 288 + }, + { + "epoch": 0.03173731605534812, + "grad_norm": 5.587200164794922, + "learning_rate": 1.580680570801317e-07, + "loss": 1.1912, + "mean_token_accuracy": 0.6494753360748291, + "num_tokens": 7430625.0, + "step": 289 + }, + { + "epoch": 0.03184713375796178, + "grad_norm": 5.286588191986084, + "learning_rate": 1.5861690450054885e-07, + "loss": 1.1854, + "mean_token_accuracy": 0.6625566482543945, + "num_tokens": 7459863.0, + "step": 290 + }, + { + "epoch": 0.03195695146057544, + "grad_norm": 7.990634441375732, + "learning_rate": 1.5916575192096596e-07, + "loss": 1.2599, + "mean_token_accuracy": 0.6501679420471191, + "num_tokens": 7484423.0, + "step": 291 + }, + { + "epoch": 0.0320667691631891, + "grad_norm": 5.217491149902344, + "learning_rate": 1.597145993413831e-07, + "loss": 1.3341, + "mean_token_accuracy": 0.6268227100372314, + "num_tokens": 7516925.0, + "step": 292 + }, + { + "epoch": 0.03217658686580277, + "grad_norm": 4.668872356414795, + "learning_rate": 1.602634467618002e-07, + "loss": 1.278, + "mean_token_accuracy": 0.642392635345459, + "num_tokens": 7546204.0, + "step": 293 + }, + { + "epoch": 0.03228640456841643, + "grad_norm": 4.973870277404785, + "learning_rate": 1.6081229418221733e-07, + "loss": 1.27, + "mean_token_accuracy": 0.6380777359008789, + "num_tokens": 7574782.0, + "step": 294 + }, + { + "epoch": 0.03239622227103009, + "grad_norm": 4.144391059875488, + "learning_rate": 1.6136114160263445e-07, + "loss": 1.1683, + "mean_token_accuracy": 0.6653579473495483, + "num_tokens": 7602063.0, + "step": 295 + }, + { + "epoch": 0.03250603997364375, + "grad_norm": 4.723386764526367, + "learning_rate": 1.619099890230516e-07, + "loss": 1.1654, + "mean_token_accuracy": 0.6618049144744873, + "num_tokens": 7630606.0, + "step": 296 + }, + { + "epoch": 0.03261585767625741, + "grad_norm": 5.313791275024414, + "learning_rate": 1.6245883644346873e-07, + "loss": 1.2503, + "mean_token_accuracy": 0.6427723169326782, + "num_tokens": 7656343.0, + "step": 297 + }, + { + "epoch": 0.032725675378871076, + "grad_norm": 6.092991828918457, + "learning_rate": 1.6300768386388582e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.6792199611663818, + "num_tokens": 7679920.0, + "step": 298 + }, + { + "epoch": 0.032835493081484736, + "grad_norm": 4.506723880767822, + "learning_rate": 1.6355653128430296e-07, + "loss": 1.2855, + "mean_token_accuracy": 0.646541953086853, + "num_tokens": 7702233.0, + "step": 299 + }, + { + "epoch": 0.032945310784098396, + "grad_norm": 4.468372344970703, + "learning_rate": 1.6410537870472008e-07, + "loss": 1.1807, + "mean_token_accuracy": 0.663381814956665, + "num_tokens": 7732401.0, + "step": 300 + }, + { + "epoch": 0.033055128486712056, + "grad_norm": 5.628025054931641, + "learning_rate": 1.6465422612513722e-07, + "loss": 1.1538, + "mean_token_accuracy": 0.6668460965156555, + "num_tokens": 7754480.0, + "step": 301 + }, + { + "epoch": 0.03316494618932572, + "grad_norm": 7.499962329864502, + "learning_rate": 1.6520307354555433e-07, + "loss": 1.1793, + "mean_token_accuracy": 0.6570918560028076, + "num_tokens": 7774396.0, + "step": 302 + }, + { + "epoch": 0.03327476389193938, + "grad_norm": 4.740800857543945, + "learning_rate": 1.6575192096597145e-07, + "loss": 1.2723, + "mean_token_accuracy": 0.656136691570282, + "num_tokens": 7799820.0, + "step": 303 + }, + { + "epoch": 0.03338458159455304, + "grad_norm": 6.121516227722168, + "learning_rate": 1.6630076838638856e-07, + "loss": 1.2318, + "mean_token_accuracy": 0.6530343890190125, + "num_tokens": 7832822.0, + "step": 304 + }, + { + "epoch": 0.0334943992971667, + "grad_norm": 4.668183326721191, + "learning_rate": 1.668496158068057e-07, + "loss": 1.2124, + "mean_token_accuracy": 0.6524491310119629, + "num_tokens": 7859077.0, + "step": 305 + }, + { + "epoch": 0.03360421699978036, + "grad_norm": 5.132410049438477, + "learning_rate": 1.6739846322722282e-07, + "loss": 1.154, + "mean_token_accuracy": 0.6714551448822021, + "num_tokens": 7888814.0, + "step": 306 + }, + { + "epoch": 0.03371403470239403, + "grad_norm": 4.165431499481201, + "learning_rate": 1.6794731064763996e-07, + "loss": 1.2554, + "mean_token_accuracy": 0.6491538882255554, + "num_tokens": 7915888.0, + "step": 307 + }, + { + "epoch": 0.03382385240500769, + "grad_norm": 5.083107948303223, + "learning_rate": 1.6849615806805705e-07, + "loss": 1.2415, + "mean_token_accuracy": 0.6477325558662415, + "num_tokens": 7939529.0, + "step": 308 + }, + { + "epoch": 0.03393367010762135, + "grad_norm": 5.561104774475098, + "learning_rate": 1.690450054884742e-07, + "loss": 1.1928, + "mean_token_accuracy": 0.6630145311355591, + "num_tokens": 7963250.0, + "step": 309 + }, + { + "epoch": 0.03404348781023501, + "grad_norm": 4.89998197555542, + "learning_rate": 1.6959385290889134e-07, + "loss": 1.2409, + "mean_token_accuracy": 0.6536746025085449, + "num_tokens": 7994264.0, + "step": 310 + }, + { + "epoch": 0.03415330551284867, + "grad_norm": 4.325306415557861, + "learning_rate": 1.7014270032930845e-07, + "loss": 1.2463, + "mean_token_accuracy": 0.647392749786377, + "num_tokens": 8029569.0, + "step": 311 + }, + { + "epoch": 0.034263123215462335, + "grad_norm": 4.090392112731934, + "learning_rate": 1.7069154774972557e-07, + "loss": 1.255, + "mean_token_accuracy": 0.6421408653259277, + "num_tokens": 8057989.0, + "step": 312 + }, + { + "epoch": 0.034372940918075995, + "grad_norm": 5.595335006713867, + "learning_rate": 1.7124039517014268e-07, + "loss": 1.1473, + "mean_token_accuracy": 0.6793884038925171, + "num_tokens": 8082097.0, + "step": 313 + }, + { + "epoch": 0.034482758620689655, + "grad_norm": 4.584921360015869, + "learning_rate": 1.7178924259055982e-07, + "loss": 1.2444, + "mean_token_accuracy": 0.649608850479126, + "num_tokens": 8110586.0, + "step": 314 + }, + { + "epoch": 0.034592576323303315, + "grad_norm": 5.178337574005127, + "learning_rate": 1.7233809001097694e-07, + "loss": 1.1398, + "mean_token_accuracy": 0.6760333776473999, + "num_tokens": 8134681.0, + "step": 315 + }, + { + "epoch": 0.034702394025916974, + "grad_norm": 5.251002788543701, + "learning_rate": 1.7288693743139408e-07, + "loss": 1.1208, + "mean_token_accuracy": 0.6751429438591003, + "num_tokens": 8160130.0, + "step": 316 + }, + { + "epoch": 0.03481221172853064, + "grad_norm": 4.368530750274658, + "learning_rate": 1.7343578485181117e-07, + "loss": 1.1587, + "mean_token_accuracy": 0.6649571061134338, + "num_tokens": 8182447.0, + "step": 317 + }, + { + "epoch": 0.0349220294311443, + "grad_norm": 5.062530517578125, + "learning_rate": 1.739846322722283e-07, + "loss": 1.2018, + "mean_token_accuracy": 0.6613837480545044, + "num_tokens": 8212506.0, + "step": 318 + }, + { + "epoch": 0.03503184713375796, + "grad_norm": 4.334657669067383, + "learning_rate": 1.7453347969264543e-07, + "loss": 1.1136, + "mean_token_accuracy": 0.6757320165634155, + "num_tokens": 8237998.0, + "step": 319 + }, + { + "epoch": 0.03514166483637162, + "grad_norm": 4.821310043334961, + "learning_rate": 1.7508232711306257e-07, + "loss": 1.1743, + "mean_token_accuracy": 0.6684830188751221, + "num_tokens": 8258937.0, + "step": 320 + }, + { + "epoch": 0.03525148253898529, + "grad_norm": 5.300125598907471, + "learning_rate": 1.756311745334797e-07, + "loss": 1.205, + "mean_token_accuracy": 0.6603655219078064, + "num_tokens": 8279044.0, + "step": 321 + }, + { + "epoch": 0.03536130024159895, + "grad_norm": 4.003087043762207, + "learning_rate": 1.761800219538968e-07, + "loss": 1.2057, + "mean_token_accuracy": 0.6646608114242554, + "num_tokens": 8306937.0, + "step": 322 + }, + { + "epoch": 0.03547111794421261, + "grad_norm": 5.617528438568115, + "learning_rate": 1.7672886937431394e-07, + "loss": 1.1314, + "mean_token_accuracy": 0.6646061539649963, + "num_tokens": 8324439.0, + "step": 323 + }, + { + "epoch": 0.03558093564682627, + "grad_norm": 4.731512069702148, + "learning_rate": 1.7727771679473105e-07, + "loss": 1.279, + "mean_token_accuracy": 0.6340307593345642, + "num_tokens": 8351402.0, + "step": 324 + }, + { + "epoch": 0.03569075334943993, + "grad_norm": 5.251392364501953, + "learning_rate": 1.778265642151482e-07, + "loss": 1.2897, + "mean_token_accuracy": 0.6420137882232666, + "num_tokens": 8376015.0, + "step": 325 + }, + { + "epoch": 0.035800571052053594, + "grad_norm": 5.854483127593994, + "learning_rate": 1.7837541163556529e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6760533452033997, + "num_tokens": 8397959.0, + "step": 326 + }, + { + "epoch": 0.035910388754667254, + "grad_norm": 3.9541258811950684, + "learning_rate": 1.7892425905598243e-07, + "loss": 1.2657, + "mean_token_accuracy": 0.6451447606086731, + "num_tokens": 8425355.0, + "step": 327 + }, + { + "epoch": 0.036020206457280914, + "grad_norm": 4.76937198638916, + "learning_rate": 1.7947310647639954e-07, + "loss": 1.2121, + "mean_token_accuracy": 0.667957067489624, + "num_tokens": 8454718.0, + "step": 328 + }, + { + "epoch": 0.03613002415989457, + "grad_norm": 3.8782975673675537, + "learning_rate": 1.8002195389681668e-07, + "loss": 1.278, + "mean_token_accuracy": 0.6374715566635132, + "num_tokens": 8482731.0, + "step": 329 + }, + { + "epoch": 0.03623984186250823, + "grad_norm": 4.517838954925537, + "learning_rate": 1.805708013172338e-07, + "loss": 1.2973, + "mean_token_accuracy": 0.645124614238739, + "num_tokens": 8510542.0, + "step": 330 + }, + { + "epoch": 0.0363496595651219, + "grad_norm": 5.221327304840088, + "learning_rate": 1.8111964873765091e-07, + "loss": 1.3076, + "mean_token_accuracy": 0.6321308016777039, + "num_tokens": 8535351.0, + "step": 331 + }, + { + "epoch": 0.03645947726773556, + "grad_norm": 6.113970756530762, + "learning_rate": 1.8166849615806803e-07, + "loss": 1.2273, + "mean_token_accuracy": 0.6472039222717285, + "num_tokens": 8560597.0, + "step": 332 + }, + { + "epoch": 0.03656929497034922, + "grad_norm": 4.356454372406006, + "learning_rate": 1.8221734357848517e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6734216213226318, + "num_tokens": 8584033.0, + "step": 333 + }, + { + "epoch": 0.03667911267296288, + "grad_norm": 3.6676008701324463, + "learning_rate": 1.827661909989023e-07, + "loss": 1.2162, + "mean_token_accuracy": 0.6573660373687744, + "num_tokens": 8615330.0, + "step": 334 + }, + { + "epoch": 0.03678893037557654, + "grad_norm": 5.906844615936279, + "learning_rate": 1.8331503841931943e-07, + "loss": 1.129, + "mean_token_accuracy": 0.677981972694397, + "num_tokens": 8635928.0, + "step": 335 + }, + { + "epoch": 0.036898748078190206, + "grad_norm": 4.388279914855957, + "learning_rate": 1.8386388583973654e-07, + "loss": 1.1615, + "mean_token_accuracy": 0.6642815470695496, + "num_tokens": 8656058.0, + "step": 336 + }, + { + "epoch": 0.037008565780803866, + "grad_norm": 4.213094234466553, + "learning_rate": 1.8441273326015366e-07, + "loss": 1.1581, + "mean_token_accuracy": 0.6718294620513916, + "num_tokens": 8679586.0, + "step": 337 + }, + { + "epoch": 0.037118383483417526, + "grad_norm": 4.554157257080078, + "learning_rate": 1.849615806805708e-07, + "loss": 1.1613, + "mean_token_accuracy": 0.66303950548172, + "num_tokens": 8705418.0, + "step": 338 + }, + { + "epoch": 0.037228201186031186, + "grad_norm": 4.46209716796875, + "learning_rate": 1.8551042810098792e-07, + "loss": 1.1489, + "mean_token_accuracy": 0.670622706413269, + "num_tokens": 8729934.0, + "step": 339 + }, + { + "epoch": 0.03733801888864485, + "grad_norm": 4.651026725769043, + "learning_rate": 1.8605927552140506e-07, + "loss": 1.1357, + "mean_token_accuracy": 0.6770144701004028, + "num_tokens": 8749616.0, + "step": 340 + }, + { + "epoch": 0.03744783659125851, + "grad_norm": 4.691188812255859, + "learning_rate": 1.8660812294182215e-07, + "loss": 1.2407, + "mean_token_accuracy": 0.6546655893325806, + "num_tokens": 8774823.0, + "step": 341 + }, + { + "epoch": 0.03755765429387217, + "grad_norm": 4.09330940246582, + "learning_rate": 1.871569703622393e-07, + "loss": 1.2625, + "mean_token_accuracy": 0.6400835514068604, + "num_tokens": 8804662.0, + "step": 342 + }, + { + "epoch": 0.03766747199648583, + "grad_norm": 4.624171257019043, + "learning_rate": 1.877058177826564e-07, + "loss": 1.1735, + "mean_token_accuracy": 0.662428081035614, + "num_tokens": 8830593.0, + "step": 343 + }, + { + "epoch": 0.03777728969909949, + "grad_norm": 4.930492877960205, + "learning_rate": 1.8825466520307354e-07, + "loss": 1.2442, + "mean_token_accuracy": 0.6476290225982666, + "num_tokens": 8854903.0, + "step": 344 + }, + { + "epoch": 0.03788710740171316, + "grad_norm": 5.586031436920166, + "learning_rate": 1.8880351262349069e-07, + "loss": 1.1147, + "mean_token_accuracy": 0.6762106418609619, + "num_tokens": 8873829.0, + "step": 345 + }, + { + "epoch": 0.03799692510432682, + "grad_norm": 6.300300121307373, + "learning_rate": 1.8935236004390778e-07, + "loss": 1.265, + "mean_token_accuracy": 0.6413426995277405, + "num_tokens": 8902961.0, + "step": 346 + }, + { + "epoch": 0.03810674280694048, + "grad_norm": 4.849955081939697, + "learning_rate": 1.8990120746432492e-07, + "loss": 1.2698, + "mean_token_accuracy": 0.6376584768295288, + "num_tokens": 8932175.0, + "step": 347 + }, + { + "epoch": 0.03821656050955414, + "grad_norm": 3.8476083278656006, + "learning_rate": 1.9045005488474203e-07, + "loss": 1.2315, + "mean_token_accuracy": 0.6537219882011414, + "num_tokens": 8964226.0, + "step": 348 + }, + { + "epoch": 0.0383263782121678, + "grad_norm": 4.916658401489258, + "learning_rate": 1.9099890230515917e-07, + "loss": 1.171, + "mean_token_accuracy": 0.659798264503479, + "num_tokens": 8986845.0, + "step": 349 + }, + { + "epoch": 0.038436195914781465, + "grad_norm": 5.231420040130615, + "learning_rate": 1.9154774972557626e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6761865019798279, + "num_tokens": 9005700.0, + "step": 350 + }, + { + "epoch": 0.038546013617395125, + "grad_norm": 5.430691242218018, + "learning_rate": 1.920965971459934e-07, + "loss": 1.214, + "mean_token_accuracy": 0.6688086986541748, + "num_tokens": 9030388.0, + "step": 351 + }, + { + "epoch": 0.038655831320008785, + "grad_norm": 4.800854206085205, + "learning_rate": 1.9264544456641052e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6757997274398804, + "num_tokens": 9053634.0, + "step": 352 + }, + { + "epoch": 0.038765649022622445, + "grad_norm": 4.250025749206543, + "learning_rate": 1.9319429198682766e-07, + "loss": 1.2978, + "mean_token_accuracy": 0.638129711151123, + "num_tokens": 9083026.0, + "step": 353 + }, + { + "epoch": 0.038875466725236105, + "grad_norm": 4.537811279296875, + "learning_rate": 1.9374313940724478e-07, + "loss": 1.1733, + "mean_token_accuracy": 0.6636962890625, + "num_tokens": 9105390.0, + "step": 354 + }, + { + "epoch": 0.03898528442784977, + "grad_norm": 4.09757661819458, + "learning_rate": 1.942919868276619e-07, + "loss": 1.2389, + "mean_token_accuracy": 0.6486430764198303, + "num_tokens": 9127341.0, + "step": 355 + }, + { + "epoch": 0.03909510213046343, + "grad_norm": 4.7298173904418945, + "learning_rate": 1.9484083424807903e-07, + "loss": 1.1766, + "mean_token_accuracy": 0.6650265455245972, + "num_tokens": 9157903.0, + "step": 356 + }, + { + "epoch": 0.03920491983307709, + "grad_norm": 5.23850154876709, + "learning_rate": 1.9538968166849615e-07, + "loss": 1.263, + "mean_token_accuracy": 0.6514431238174438, + "num_tokens": 9183933.0, + "step": 357 + }, + { + "epoch": 0.03931473753569075, + "grad_norm": 4.3384833335876465, + "learning_rate": 1.959385290889133e-07, + "loss": 1.1654, + "mean_token_accuracy": 0.6685746908187866, + "num_tokens": 9208081.0, + "step": 358 + }, + { + "epoch": 0.03942455523830442, + "grad_norm": 4.554821491241455, + "learning_rate": 1.964873765093304e-07, + "loss": 1.1952, + "mean_token_accuracy": 0.6550672650337219, + "num_tokens": 9235143.0, + "step": 359 + }, + { + "epoch": 0.03953437294091808, + "grad_norm": 4.358982563018799, + "learning_rate": 1.9703622392974752e-07, + "loss": 1.1822, + "mean_token_accuracy": 0.6717649698257446, + "num_tokens": 9257423.0, + "step": 360 + }, + { + "epoch": 0.03964419064353174, + "grad_norm": 4.049408435821533, + "learning_rate": 1.9758507135016464e-07, + "loss": 1.1891, + "mean_token_accuracy": 0.6565467119216919, + "num_tokens": 9282250.0, + "step": 361 + }, + { + "epoch": 0.0397540083461454, + "grad_norm": 3.971503734588623, + "learning_rate": 1.9813391877058178e-07, + "loss": 1.2023, + "mean_token_accuracy": 0.6541351675987244, + "num_tokens": 9310493.0, + "step": 362 + }, + { + "epoch": 0.03986382604875906, + "grad_norm": 3.8229126930236816, + "learning_rate": 1.986827661909989e-07, + "loss": 1.2103, + "mean_token_accuracy": 0.6550173759460449, + "num_tokens": 9340578.0, + "step": 363 + }, + { + "epoch": 0.039973643751372724, + "grad_norm": 4.347021579742432, + "learning_rate": 1.9923161361141603e-07, + "loss": 1.2213, + "mean_token_accuracy": 0.6648902893066406, + "num_tokens": 9365740.0, + "step": 364 + }, + { + "epoch": 0.040083461453986384, + "grad_norm": 3.7338833808898926, + "learning_rate": 1.9978046103183312e-07, + "loss": 1.2615, + "mean_token_accuracy": 0.6432459354400635, + "num_tokens": 9395500.0, + "step": 365 + }, + { + "epoch": 0.040193279156600044, + "grad_norm": 5.05778694152832, + "learning_rate": 2.0032930845225027e-07, + "loss": 1.2259, + "mean_token_accuracy": 0.6446888446807861, + "num_tokens": 9416407.0, + "step": 366 + }, + { + "epoch": 0.040303096859213704, + "grad_norm": 3.3753747940063477, + "learning_rate": 2.0087815587266738e-07, + "loss": 1.1682, + "mean_token_accuracy": 0.6743870377540588, + "num_tokens": 9447462.0, + "step": 367 + }, + { + "epoch": 0.040412914561827364, + "grad_norm": 3.9459831714630127, + "learning_rate": 2.0142700329308452e-07, + "loss": 1.1659, + "mean_token_accuracy": 0.6647689938545227, + "num_tokens": 9474937.0, + "step": 368 + }, + { + "epoch": 0.04052273226444103, + "grad_norm": 4.594443321228027, + "learning_rate": 2.0197585071350164e-07, + "loss": 1.1934, + "mean_token_accuracy": 0.6818909049034119, + "num_tokens": 9499351.0, + "step": 369 + }, + { + "epoch": 0.04063254996705469, + "grad_norm": 4.214311599731445, + "learning_rate": 2.0252469813391875e-07, + "loss": 1.0672, + "mean_token_accuracy": 0.6954911351203918, + "num_tokens": 9523476.0, + "step": 370 + }, + { + "epoch": 0.04074236766966835, + "grad_norm": 4.3880510330200195, + "learning_rate": 2.030735455543359e-07, + "loss": 1.1238, + "mean_token_accuracy": 0.6930744051933289, + "num_tokens": 9547266.0, + "step": 371 + }, + { + "epoch": 0.04085218537228201, + "grad_norm": 3.2013046741485596, + "learning_rate": 2.03622392974753e-07, + "loss": 1.1517, + "mean_token_accuracy": 0.6661689281463623, + "num_tokens": 9579120.0, + "step": 372 + }, + { + "epoch": 0.04096200307489567, + "grad_norm": 4.17141056060791, + "learning_rate": 2.0417124039517015e-07, + "loss": 1.1863, + "mean_token_accuracy": 0.6674935817718506, + "num_tokens": 9602564.0, + "step": 373 + }, + { + "epoch": 0.04107182077750934, + "grad_norm": 3.186497926712036, + "learning_rate": 2.0472008781558724e-07, + "loss": 1.1748, + "mean_token_accuracy": 0.6661578416824341, + "num_tokens": 9638866.0, + "step": 374 + }, + { + "epoch": 0.041181638480122996, + "grad_norm": 3.4257359504699707, + "learning_rate": 2.0526893523600438e-07, + "loss": 1.2539, + "mean_token_accuracy": 0.643177330493927, + "num_tokens": 9671288.0, + "step": 375 + }, + { + "epoch": 0.041291456182736656, + "grad_norm": 3.624720811843872, + "learning_rate": 2.058177826564215e-07, + "loss": 1.2102, + "mean_token_accuracy": 0.658899188041687, + "num_tokens": 9701020.0, + "step": 376 + }, + { + "epoch": 0.041401273885350316, + "grad_norm": 4.995517253875732, + "learning_rate": 2.0636663007683864e-07, + "loss": 1.1952, + "mean_token_accuracy": 0.656171977519989, + "num_tokens": 9722737.0, + "step": 377 + }, + { + "epoch": 0.04151109158796398, + "grad_norm": 4.856442928314209, + "learning_rate": 2.0691547749725575e-07, + "loss": 1.2318, + "mean_token_accuracy": 0.6471506953239441, + "num_tokens": 9748228.0, + "step": 378 + }, + { + "epoch": 0.04162090929057764, + "grad_norm": 3.3617160320281982, + "learning_rate": 2.0746432491767287e-07, + "loss": 1.2657, + "mean_token_accuracy": 0.6347624063491821, + "num_tokens": 9777238.0, + "step": 379 + }, + { + "epoch": 0.0417307269931913, + "grad_norm": 3.530855417251587, + "learning_rate": 2.0801317233809e-07, + "loss": 1.2247, + "mean_token_accuracy": 0.654370129108429, + "num_tokens": 9805776.0, + "step": 380 + }, + { + "epoch": 0.04184054469580496, + "grad_norm": 3.797964096069336, + "learning_rate": 2.0856201975850713e-07, + "loss": 1.1957, + "mean_token_accuracy": 0.6639024019241333, + "num_tokens": 9837273.0, + "step": 381 + }, + { + "epoch": 0.04195036239841862, + "grad_norm": 3.8349180221557617, + "learning_rate": 2.0911086717892427e-07, + "loss": 1.1493, + "mean_token_accuracy": 0.6659078598022461, + "num_tokens": 9862594.0, + "step": 382 + }, + { + "epoch": 0.04206018010103229, + "grad_norm": 4.016434192657471, + "learning_rate": 2.0965971459934136e-07, + "loss": 1.2663, + "mean_token_accuracy": 0.6444213390350342, + "num_tokens": 9890072.0, + "step": 383 + }, + { + "epoch": 0.04216999780364595, + "grad_norm": 3.638173818588257, + "learning_rate": 2.102085620197585e-07, + "loss": 1.1563, + "mean_token_accuracy": 0.6676033735275269, + "num_tokens": 9916323.0, + "step": 384 + }, + { + "epoch": 0.04227981550625961, + "grad_norm": 4.391975402832031, + "learning_rate": 2.1075740944017561e-07, + "loss": 1.1866, + "mean_token_accuracy": 0.661217212677002, + "num_tokens": 9943394.0, + "step": 385 + }, + { + "epoch": 0.04238963320887327, + "grad_norm": 4.0686187744140625, + "learning_rate": 2.1130625686059276e-07, + "loss": 1.1926, + "mean_token_accuracy": 0.6579450368881226, + "num_tokens": 9968254.0, + "step": 386 + }, + { + "epoch": 0.04249945091148693, + "grad_norm": 3.1034035682678223, + "learning_rate": 2.1185510428100987e-07, + "loss": 1.198, + "mean_token_accuracy": 0.6596943736076355, + "num_tokens": 9995987.0, + "step": 387 + }, + { + "epoch": 0.042609268614100596, + "grad_norm": 3.9975998401641846, + "learning_rate": 2.1240395170142699e-07, + "loss": 1.1586, + "mean_token_accuracy": 0.6676046848297119, + "num_tokens": 10023481.0, + "step": 388 + }, + { + "epoch": 0.042719086316714255, + "grad_norm": 3.9905354976654053, + "learning_rate": 2.129527991218441e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6756174564361572, + "num_tokens": 10046035.0, + "step": 389 + }, + { + "epoch": 0.042828904019327915, + "grad_norm": 4.950873374938965, + "learning_rate": 2.1350164654226124e-07, + "loss": 1.1506, + "mean_token_accuracy": 0.663886308670044, + "num_tokens": 10065100.0, + "step": 390 + }, + { + "epoch": 0.042938721721941575, + "grad_norm": 4.323631286621094, + "learning_rate": 2.1405049396267836e-07, + "loss": 1.1489, + "mean_token_accuracy": 0.6689159870147705, + "num_tokens": 10092431.0, + "step": 391 + }, + { + "epoch": 0.043048539424555235, + "grad_norm": 4.425236701965332, + "learning_rate": 2.145993413830955e-07, + "loss": 1.1246, + "mean_token_accuracy": 0.673577070236206, + "num_tokens": 10117817.0, + "step": 392 + }, + { + "epoch": 0.0431583571271689, + "grad_norm": 3.7834856510162354, + "learning_rate": 2.1514818880351262e-07, + "loss": 1.1188, + "mean_token_accuracy": 0.6800717711448669, + "num_tokens": 10140092.0, + "step": 393 + }, + { + "epoch": 0.04326817482978256, + "grad_norm": 3.459441661834717, + "learning_rate": 2.1569703622392973e-07, + "loss": 1.2346, + "mean_token_accuracy": 0.6465470194816589, + "num_tokens": 10165105.0, + "step": 394 + }, + { + "epoch": 0.04337799253239622, + "grad_norm": 3.411724805831909, + "learning_rate": 2.1624588364434687e-07, + "loss": 1.2283, + "mean_token_accuracy": 0.6590484976768494, + "num_tokens": 10194859.0, + "step": 395 + }, + { + "epoch": 0.04348781023500988, + "grad_norm": 3.5442428588867188, + "learning_rate": 2.16794731064764e-07, + "loss": 1.2795, + "mean_token_accuracy": 0.649748682975769, + "num_tokens": 10220998.0, + "step": 396 + }, + { + "epoch": 0.04359762793762355, + "grad_norm": 5.505099773406982, + "learning_rate": 2.1734357848518113e-07, + "loss": 1.1399, + "mean_token_accuracy": 0.6704980134963989, + "num_tokens": 10238463.0, + "step": 397 + }, + { + "epoch": 0.04370744564023721, + "grad_norm": 3.210703134536743, + "learning_rate": 2.1789242590559822e-07, + "loss": 1.2512, + "mean_token_accuracy": 0.6431623697280884, + "num_tokens": 10266445.0, + "step": 398 + }, + { + "epoch": 0.04381726334285087, + "grad_norm": 2.9881396293640137, + "learning_rate": 2.1844127332601536e-07, + "loss": 1.2185, + "mean_token_accuracy": 0.6503974795341492, + "num_tokens": 10299262.0, + "step": 399 + }, + { + "epoch": 0.04392708104546453, + "grad_norm": 3.2044312953948975, + "learning_rate": 2.1899012074643247e-07, + "loss": 1.1318, + "mean_token_accuracy": 0.6728498935699463, + "num_tokens": 10328648.0, + "step": 400 + }, + { + "epoch": 0.04403689874807819, + "grad_norm": 4.173638343811035, + "learning_rate": 2.1953896816684962e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.688075602054596, + "num_tokens": 10351635.0, + "step": 401 + }, + { + "epoch": 0.044146716450691854, + "grad_norm": 3.3236331939697266, + "learning_rate": 2.200878155872667e-07, + "loss": 1.1973, + "mean_token_accuracy": 0.6603402495384216, + "num_tokens": 10383035.0, + "step": 402 + }, + { + "epoch": 0.044256534153305514, + "grad_norm": 3.6645901203155518, + "learning_rate": 2.2063666300768385e-07, + "loss": 1.2382, + "mean_token_accuracy": 0.6437113285064697, + "num_tokens": 10408967.0, + "step": 403 + }, + { + "epoch": 0.044366351855919174, + "grad_norm": 3.9541242122650146, + "learning_rate": 2.21185510428101e-07, + "loss": 1.3156, + "mean_token_accuracy": 0.6374862790107727, + "num_tokens": 10438425.0, + "step": 404 + }, + { + "epoch": 0.044476169558532834, + "grad_norm": 3.4626095294952393, + "learning_rate": 2.217343578485181e-07, + "loss": 1.2222, + "mean_token_accuracy": 0.6526035070419312, + "num_tokens": 10468435.0, + "step": 405 + }, + { + "epoch": 0.044585987261146494, + "grad_norm": 4.042892932891846, + "learning_rate": 2.2228320526893525e-07, + "loss": 1.2456, + "mean_token_accuracy": 0.646306574344635, + "num_tokens": 10495240.0, + "step": 406 + }, + { + "epoch": 0.04469580496376016, + "grad_norm": 2.97159743309021, + "learning_rate": 2.2283205268935233e-07, + "loss": 1.1729, + "mean_token_accuracy": 0.672167956829071, + "num_tokens": 10523243.0, + "step": 407 + }, + { + "epoch": 0.04480562266637382, + "grad_norm": 3.8398990631103516, + "learning_rate": 2.2338090010976948e-07, + "loss": 1.0972, + "mean_token_accuracy": 0.6824924945831299, + "num_tokens": 10547410.0, + "step": 408 + }, + { + "epoch": 0.04491544036898748, + "grad_norm": 3.4058640003204346, + "learning_rate": 2.239297475301866e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6721813678741455, + "num_tokens": 10572581.0, + "step": 409 + }, + { + "epoch": 0.04502525807160114, + "grad_norm": 3.6843559741973877, + "learning_rate": 2.2447859495060373e-07, + "loss": 1.1911, + "mean_token_accuracy": 0.6680945158004761, + "num_tokens": 10597594.0, + "step": 410 + }, + { + "epoch": 0.0451350757742148, + "grad_norm": 3.6686201095581055, + "learning_rate": 2.2502744237102085e-07, + "loss": 1.1863, + "mean_token_accuracy": 0.6663823127746582, + "num_tokens": 10625727.0, + "step": 411 + }, + { + "epoch": 0.04524489347682847, + "grad_norm": 3.6907105445861816, + "learning_rate": 2.2557628979143796e-07, + "loss": 1.1606, + "mean_token_accuracy": 0.6772780418395996, + "num_tokens": 10648028.0, + "step": 412 + }, + { + "epoch": 0.04535471117944213, + "grad_norm": 3.8406972885131836, + "learning_rate": 2.2612513721185508e-07, + "loss": 1.1565, + "mean_token_accuracy": 0.6628037691116333, + "num_tokens": 10673000.0, + "step": 413 + }, + { + "epoch": 0.04546452888205579, + "grad_norm": 4.416815280914307, + "learning_rate": 2.2667398463227222e-07, + "loss": 1.1376, + "mean_token_accuracy": 0.6762301921844482, + "num_tokens": 10692459.0, + "step": 414 + }, + { + "epoch": 0.045574346584669446, + "grad_norm": 3.3040874004364014, + "learning_rate": 2.2722283205268934e-07, + "loss": 1.1877, + "mean_token_accuracy": 0.656684160232544, + "num_tokens": 10718617.0, + "step": 415 + }, + { + "epoch": 0.04568416428728311, + "grad_norm": 3.465055465698242, + "learning_rate": 2.2777167947310648e-07, + "loss": 1.1941, + "mean_token_accuracy": 0.6600798964500427, + "num_tokens": 10748871.0, + "step": 416 + }, + { + "epoch": 0.04579398198989677, + "grad_norm": 4.1894049644470215, + "learning_rate": 2.283205268935236e-07, + "loss": 1.1594, + "mean_token_accuracy": 0.6659698486328125, + "num_tokens": 10771895.0, + "step": 417 + }, + { + "epoch": 0.04590379969251043, + "grad_norm": 3.346766471862793, + "learning_rate": 2.288693743139407e-07, + "loss": 1.1813, + "mean_token_accuracy": 0.6657622456550598, + "num_tokens": 10802730.0, + "step": 418 + }, + { + "epoch": 0.04601361739512409, + "grad_norm": 3.6176064014434814, + "learning_rate": 2.2941822173435785e-07, + "loss": 1.1564, + "mean_token_accuracy": 0.667605459690094, + "num_tokens": 10830506.0, + "step": 419 + }, + { + "epoch": 0.04612343509773775, + "grad_norm": 3.4723894596099854, + "learning_rate": 2.2996706915477496e-07, + "loss": 1.1452, + "mean_token_accuracy": 0.6716974973678589, + "num_tokens": 10852935.0, + "step": 420 + }, + { + "epoch": 0.04623325280035142, + "grad_norm": 3.9134390354156494, + "learning_rate": 2.3051591657519208e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6764343976974487, + "num_tokens": 10877466.0, + "step": 421 + }, + { + "epoch": 0.04634307050296508, + "grad_norm": 4.559454917907715, + "learning_rate": 2.310647639956092e-07, + "loss": 1.29, + "mean_token_accuracy": 0.6326936483383179, + "num_tokens": 10901539.0, + "step": 422 + }, + { + "epoch": 0.04645288820557874, + "grad_norm": 3.5964908599853516, + "learning_rate": 2.3161361141602634e-07, + "loss": 1.2008, + "mean_token_accuracy": 0.6613451838493347, + "num_tokens": 10925952.0, + "step": 423 + }, + { + "epoch": 0.0465627059081924, + "grad_norm": 3.7000949382781982, + "learning_rate": 2.3216245883644345e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6779930591583252, + "num_tokens": 10946093.0, + "step": 424 + }, + { + "epoch": 0.04667252361080606, + "grad_norm": 4.0119218826293945, + "learning_rate": 2.327113062568606e-07, + "loss": 1.1886, + "mean_token_accuracy": 0.6574459671974182, + "num_tokens": 10967992.0, + "step": 425 + }, + { + "epoch": 0.046782341313419726, + "grad_norm": 3.5525336265563965, + "learning_rate": 2.3326015367727768e-07, + "loss": 1.0902, + "mean_token_accuracy": 0.686151921749115, + "num_tokens": 10992553.0, + "step": 426 + }, + { + "epoch": 0.046892159016033386, + "grad_norm": 3.4995951652526855, + "learning_rate": 2.3380900109769482e-07, + "loss": 1.2351, + "mean_token_accuracy": 0.6583859920501709, + "num_tokens": 11018069.0, + "step": 427 + }, + { + "epoch": 0.047001976718647046, + "grad_norm": 3.4146761894226074, + "learning_rate": 2.3435784851811197e-07, + "loss": 1.1273, + "mean_token_accuracy": 0.6695701479911804, + "num_tokens": 11044694.0, + "step": 428 + }, + { + "epoch": 0.047111794421260705, + "grad_norm": 2.943099021911621, + "learning_rate": 2.3490669593852908e-07, + "loss": 1.143, + "mean_token_accuracy": 0.6687943935394287, + "num_tokens": 11072904.0, + "step": 429 + }, + { + "epoch": 0.047221612123874365, + "grad_norm": 3.042056083679199, + "learning_rate": 2.3545554335894622e-07, + "loss": 1.26, + "mean_token_accuracy": 0.6402145028114319, + "num_tokens": 11104640.0, + "step": 430 + }, + { + "epoch": 0.04733142982648803, + "grad_norm": 3.3478755950927734, + "learning_rate": 2.360043907793633e-07, + "loss": 1.1445, + "mean_token_accuracy": 0.6730588674545288, + "num_tokens": 11128921.0, + "step": 431 + }, + { + "epoch": 0.04744124752910169, + "grad_norm": 2.8656439781188965, + "learning_rate": 2.3655323819978045e-07, + "loss": 1.1964, + "mean_token_accuracy": 0.6522825360298157, + "num_tokens": 11156744.0, + "step": 432 + }, + { + "epoch": 0.04755106523171535, + "grad_norm": 3.029297351837158, + "learning_rate": 2.3710208562019757e-07, + "loss": 1.1377, + "mean_token_accuracy": 0.674307644367218, + "num_tokens": 11182143.0, + "step": 433 + }, + { + "epoch": 0.04766088293432901, + "grad_norm": 2.6743409633636475, + "learning_rate": 2.376509330406147e-07, + "loss": 1.2226, + "mean_token_accuracy": 0.6486610770225525, + "num_tokens": 11213456.0, + "step": 434 + }, + { + "epoch": 0.04777070063694268, + "grad_norm": 3.3818349838256836, + "learning_rate": 2.3819978046103183e-07, + "loss": 1.164, + "mean_token_accuracy": 0.6608444452285767, + "num_tokens": 11233864.0, + "step": 435 + }, + { + "epoch": 0.04788051833955634, + "grad_norm": 3.562032461166382, + "learning_rate": 2.3874862788144894e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6778846979141235, + "num_tokens": 11254175.0, + "step": 436 + }, + { + "epoch": 0.04799033604217, + "grad_norm": 2.960205554962158, + "learning_rate": 2.3929747530186606e-07, + "loss": 1.0946, + "mean_token_accuracy": 0.6820549368858337, + "num_tokens": 11285370.0, + "step": 437 + }, + { + "epoch": 0.04810015374478366, + "grad_norm": 3.8660454750061035, + "learning_rate": 2.3984632272228317e-07, + "loss": 1.208, + "mean_token_accuracy": 0.653372049331665, + "num_tokens": 11305800.0, + "step": 438 + }, + { + "epoch": 0.04820997144739732, + "grad_norm": 3.0140485763549805, + "learning_rate": 2.4039517014270034e-07, + "loss": 1.1684, + "mean_token_accuracy": 0.6575211882591248, + "num_tokens": 11334569.0, + "step": 439 + }, + { + "epoch": 0.048319789150010985, + "grad_norm": 3.671295642852783, + "learning_rate": 2.4094401756311745e-07, + "loss": 1.2338, + "mean_token_accuracy": 0.6461405754089355, + "num_tokens": 11357289.0, + "step": 440 + }, + { + "epoch": 0.048429606852624645, + "grad_norm": 3.4250588417053223, + "learning_rate": 2.4149286498353457e-07, + "loss": 1.1823, + "mean_token_accuracy": 0.6565593481063843, + "num_tokens": 11386832.0, + "step": 441 + }, + { + "epoch": 0.048539424555238304, + "grad_norm": 3.5711143016815186, + "learning_rate": 2.420417124039517e-07, + "loss": 1.1205, + "mean_token_accuracy": 0.6736968755722046, + "num_tokens": 11408615.0, + "step": 442 + }, + { + "epoch": 0.048649242257851964, + "grad_norm": 3.0084526538848877, + "learning_rate": 2.425905598243688e-07, + "loss": 1.2256, + "mean_token_accuracy": 0.6493858695030212, + "num_tokens": 11439222.0, + "step": 443 + }, + { + "epoch": 0.048759059960465624, + "grad_norm": 3.0286712646484375, + "learning_rate": 2.431394072447859e-07, + "loss": 1.144, + "mean_token_accuracy": 0.6752204895019531, + "num_tokens": 11467048.0, + "step": 444 + }, + { + "epoch": 0.04886887766307929, + "grad_norm": 3.327585458755493, + "learning_rate": 2.436882546652031e-07, + "loss": 1.148, + "mean_token_accuracy": 0.6721030473709106, + "num_tokens": 11490485.0, + "step": 445 + }, + { + "epoch": 0.04897869536569295, + "grad_norm": 3.2981514930725098, + "learning_rate": 2.442371020856202e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6884114742279053, + "num_tokens": 11513558.0, + "step": 446 + }, + { + "epoch": 0.04908851306830661, + "grad_norm": 3.6579713821411133, + "learning_rate": 2.447859495060373e-07, + "loss": 1.1674, + "mean_token_accuracy": 0.6618920564651489, + "num_tokens": 11533306.0, + "step": 447 + }, + { + "epoch": 0.04919833077092027, + "grad_norm": 3.745177745819092, + "learning_rate": 2.4533479692645443e-07, + "loss": 1.1921, + "mean_token_accuracy": 0.6515558958053589, + "num_tokens": 11554827.0, + "step": 448 + }, + { + "epoch": 0.04930814847353393, + "grad_norm": 3.153454065322876, + "learning_rate": 2.4588364434687154e-07, + "loss": 1.1432, + "mean_token_accuracy": 0.6703471541404724, + "num_tokens": 11579570.0, + "step": 449 + }, + { + "epoch": 0.0494179661761476, + "grad_norm": 3.474409818649292, + "learning_rate": 2.4643249176728866e-07, + "loss": 1.1326, + "mean_token_accuracy": 0.6783491373062134, + "num_tokens": 11606270.0, + "step": 450 + }, + { + "epoch": 0.04952778387876126, + "grad_norm": 4.031963348388672, + "learning_rate": 2.4698133918770583e-07, + "loss": 1.1987, + "mean_token_accuracy": 0.6561270952224731, + "num_tokens": 11628741.0, + "step": 451 + }, + { + "epoch": 0.04963760158137492, + "grad_norm": 3.3450841903686523, + "learning_rate": 2.4753018660812294e-07, + "loss": 1.133, + "mean_token_accuracy": 0.6694998741149902, + "num_tokens": 11652363.0, + "step": 452 + }, + { + "epoch": 0.04974741928398858, + "grad_norm": 3.231982707977295, + "learning_rate": 2.4807903402854006e-07, + "loss": 1.1896, + "mean_token_accuracy": 0.653766393661499, + "num_tokens": 11677044.0, + "step": 453 + }, + { + "epoch": 0.049857236986602244, + "grad_norm": 3.4852402210235596, + "learning_rate": 2.486278814489572e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6596474647521973, + "num_tokens": 11699168.0, + "step": 454 + }, + { + "epoch": 0.0499670546892159, + "grad_norm": 3.678738832473755, + "learning_rate": 2.491767288693743e-07, + "loss": 1.1123, + "mean_token_accuracy": 0.6680972576141357, + "num_tokens": 11721023.0, + "step": 455 + }, + { + "epoch": 0.05007687239182956, + "grad_norm": 3.4164507389068604, + "learning_rate": 2.4972557628979146e-07, + "loss": 1.2498, + "mean_token_accuracy": 0.6424973607063293, + "num_tokens": 11751028.0, + "step": 456 + }, + { + "epoch": 0.05018669009444322, + "grad_norm": 2.8839380741119385, + "learning_rate": 2.5027442371020857e-07, + "loss": 1.1424, + "mean_token_accuracy": 0.6860448718070984, + "num_tokens": 11778316.0, + "step": 457 + }, + { + "epoch": 0.05029650779705688, + "grad_norm": 3.1898818016052246, + "learning_rate": 2.508232711306257e-07, + "loss": 1.1852, + "mean_token_accuracy": 0.659544825553894, + "num_tokens": 11803463.0, + "step": 458 + }, + { + "epoch": 0.05040632549967055, + "grad_norm": 2.8101646900177, + "learning_rate": 2.513721185510428e-07, + "loss": 1.1677, + "mean_token_accuracy": 0.6700890064239502, + "num_tokens": 11831037.0, + "step": 459 + }, + { + "epoch": 0.05051614320228421, + "grad_norm": 2.5827176570892334, + "learning_rate": 2.519209659714599e-07, + "loss": 1.2155, + "mean_token_accuracy": 0.6503406763076782, + "num_tokens": 11865676.0, + "step": 460 + }, + { + "epoch": 0.05062596090489787, + "grad_norm": 2.866119384765625, + "learning_rate": 2.524698133918771e-07, + "loss": 1.1066, + "mean_token_accuracy": 0.6757497787475586, + "num_tokens": 11894887.0, + "step": 461 + }, + { + "epoch": 0.05073577860751153, + "grad_norm": 3.785996437072754, + "learning_rate": 2.5301866081229415e-07, + "loss": 1.183, + "mean_token_accuracy": 0.6599652767181396, + "num_tokens": 11921529.0, + "step": 462 + }, + { + "epoch": 0.05084559631012519, + "grad_norm": 2.9783010482788086, + "learning_rate": 2.5356750823271126e-07, + "loss": 1.1623, + "mean_token_accuracy": 0.6676862239837646, + "num_tokens": 11952307.0, + "step": 463 + }, + { + "epoch": 0.050955414012738856, + "grad_norm": 3.0202603340148926, + "learning_rate": 2.5411635565312843e-07, + "loss": 1.1563, + "mean_token_accuracy": 0.6639326214790344, + "num_tokens": 11976730.0, + "step": 464 + }, + { + "epoch": 0.051065231715352516, + "grad_norm": 2.827364444732666, + "learning_rate": 2.5466520307354555e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6809078454971313, + "num_tokens": 12007588.0, + "step": 465 + }, + { + "epoch": 0.051175049417966176, + "grad_norm": 3.2901723384857178, + "learning_rate": 2.5521405049396266e-07, + "loss": 1.1199, + "mean_token_accuracy": 0.676410436630249, + "num_tokens": 12031043.0, + "step": 466 + }, + { + "epoch": 0.051284867120579836, + "grad_norm": 3.465571641921997, + "learning_rate": 2.557628979143798e-07, + "loss": 1.1238, + "mean_token_accuracy": 0.6753427386283875, + "num_tokens": 12053798.0, + "step": 467 + }, + { + "epoch": 0.051394684823193496, + "grad_norm": 3.2409327030181885, + "learning_rate": 2.5631174533479695e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6965352296829224, + "num_tokens": 12077464.0, + "step": 468 + }, + { + "epoch": 0.05150450252580716, + "grad_norm": 3.0137853622436523, + "learning_rate": 2.5686059275521406e-07, + "loss": 1.1666, + "mean_token_accuracy": 0.6669148802757263, + "num_tokens": 12102114.0, + "step": 469 + }, + { + "epoch": 0.05161432022842082, + "grad_norm": 2.673858165740967, + "learning_rate": 2.574094401756312e-07, + "loss": 1.1691, + "mean_token_accuracy": 0.6678037643432617, + "num_tokens": 12128040.0, + "step": 470 + }, + { + "epoch": 0.05172413793103448, + "grad_norm": 3.053999185562134, + "learning_rate": 2.5795828759604824e-07, + "loss": 1.3122, + "mean_token_accuracy": 0.6301594972610474, + "num_tokens": 12158522.0, + "step": 471 + }, + { + "epoch": 0.05183395563364814, + "grad_norm": 3.1981189250946045, + "learning_rate": 2.585071350164654e-07, + "loss": 1.1458, + "mean_token_accuracy": 0.6637744307518005, + "num_tokens": 12181097.0, + "step": 472 + }, + { + "epoch": 0.05194377333626181, + "grad_norm": 2.856210231781006, + "learning_rate": 2.590559824368825e-07, + "loss": 1.098, + "mean_token_accuracy": 0.6804942488670349, + "num_tokens": 12206552.0, + "step": 473 + }, + { + "epoch": 0.05205359103887547, + "grad_norm": 3.0908830165863037, + "learning_rate": 2.5960482985729964e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.6770902872085571, + "num_tokens": 12228780.0, + "step": 474 + }, + { + "epoch": 0.05216340874148913, + "grad_norm": 2.834484100341797, + "learning_rate": 2.601536772777168e-07, + "loss": 1.1513, + "mean_token_accuracy": 0.6743897199630737, + "num_tokens": 12257403.0, + "step": 475 + }, + { + "epoch": 0.05227322644410279, + "grad_norm": 3.0580620765686035, + "learning_rate": 2.607025246981339e-07, + "loss": 1.1632, + "mean_token_accuracy": 0.6666932702064514, + "num_tokens": 12284277.0, + "step": 476 + }, + { + "epoch": 0.05238304414671645, + "grad_norm": 2.871431350708008, + "learning_rate": 2.6125137211855104e-07, + "loss": 1.2175, + "mean_token_accuracy": 0.6579543352127075, + "num_tokens": 12310660.0, + "step": 477 + }, + { + "epoch": 0.052492861849330115, + "grad_norm": 3.2635364532470703, + "learning_rate": 2.6180021953896815e-07, + "loss": 1.1001, + "mean_token_accuracy": 0.6826508045196533, + "num_tokens": 12332237.0, + "step": 478 + }, + { + "epoch": 0.052602679551943775, + "grad_norm": 2.980973243713379, + "learning_rate": 2.623490669593853e-07, + "loss": 1.2342, + "mean_token_accuracy": 0.6467565894126892, + "num_tokens": 12356800.0, + "step": 479 + }, + { + "epoch": 0.052712497254557435, + "grad_norm": 2.830005645751953, + "learning_rate": 2.6289791437980244e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6887964010238647, + "num_tokens": 12381275.0, + "step": 480 + }, + { + "epoch": 0.052822314957171095, + "grad_norm": 2.7090609073638916, + "learning_rate": 2.634467618002195e-07, + "loss": 1.1616, + "mean_token_accuracy": 0.66317218542099, + "num_tokens": 12410928.0, + "step": 481 + }, + { + "epoch": 0.052932132659784754, + "grad_norm": 2.8453125953674316, + "learning_rate": 2.639956092206366e-07, + "loss": 1.0387, + "mean_token_accuracy": 0.6935009956359863, + "num_tokens": 12437477.0, + "step": 482 + }, + { + "epoch": 0.05304195036239842, + "grad_norm": 2.5217244625091553, + "learning_rate": 2.645444566410538e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6847454905509949, + "num_tokens": 12469156.0, + "step": 483 + }, + { + "epoch": 0.05315176806501208, + "grad_norm": 2.7423031330108643, + "learning_rate": 2.650933040614709e-07, + "loss": 1.2793, + "mean_token_accuracy": 0.6412578821182251, + "num_tokens": 12498934.0, + "step": 484 + }, + { + "epoch": 0.05326158576762574, + "grad_norm": 3.570826768875122, + "learning_rate": 2.65642151481888e-07, + "loss": 1.1712, + "mean_token_accuracy": 0.6618643999099731, + "num_tokens": 12519009.0, + "step": 485 + }, + { + "epoch": 0.0533714034702394, + "grad_norm": 3.377336263656616, + "learning_rate": 2.6619099890230513e-07, + "loss": 1.2354, + "mean_token_accuracy": 0.6502575874328613, + "num_tokens": 12543850.0, + "step": 486 + }, + { + "epoch": 0.05348122117285306, + "grad_norm": 3.2009291648864746, + "learning_rate": 2.667398463227223e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.683011531829834, + "num_tokens": 12566432.0, + "step": 487 + }, + { + "epoch": 0.05359103887546673, + "grad_norm": 2.694277286529541, + "learning_rate": 2.672886937431394e-07, + "loss": 1.1204, + "mean_token_accuracy": 0.6786506772041321, + "num_tokens": 12595531.0, + "step": 488 + }, + { + "epoch": 0.05370085657808039, + "grad_norm": 2.9749841690063477, + "learning_rate": 2.678375411635565e-07, + "loss": 1.0909, + "mean_token_accuracy": 0.6771296262741089, + "num_tokens": 12619489.0, + "step": 489 + }, + { + "epoch": 0.05381067428069405, + "grad_norm": 3.04616117477417, + "learning_rate": 2.6838638858397364e-07, + "loss": 1.102, + "mean_token_accuracy": 0.6756287813186646, + "num_tokens": 12641791.0, + "step": 490 + }, + { + "epoch": 0.05392049198330771, + "grad_norm": 3.0211732387542725, + "learning_rate": 2.6893523600439076e-07, + "loss": 1.145, + "mean_token_accuracy": 0.669708251953125, + "num_tokens": 12666416.0, + "step": 491 + }, + { + "epoch": 0.054030309685921374, + "grad_norm": 3.3244619369506836, + "learning_rate": 2.6948408342480787e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6717710494995117, + "num_tokens": 12689605.0, + "step": 492 + }, + { + "epoch": 0.054140127388535034, + "grad_norm": 3.063728094100952, + "learning_rate": 2.70032930845225e-07, + "loss": 1.0853, + "mean_token_accuracy": 0.6799708008766174, + "num_tokens": 12716370.0, + "step": 493 + }, + { + "epoch": 0.054249945091148694, + "grad_norm": 2.986290693283081, + "learning_rate": 2.7058177826564215e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6642839312553406, + "num_tokens": 12743338.0, + "step": 494 + }, + { + "epoch": 0.05435976279376235, + "grad_norm": 3.2496609687805176, + "learning_rate": 2.7113062568605927e-07, + "loss": 1.1154, + "mean_token_accuracy": 0.6766266822814941, + "num_tokens": 12766185.0, + "step": 495 + }, + { + "epoch": 0.05446958049637601, + "grad_norm": 2.6840555667877197, + "learning_rate": 2.716794731064764e-07, + "loss": 1.1578, + "mean_token_accuracy": 0.6669902801513672, + "num_tokens": 12793395.0, + "step": 496 + }, + { + "epoch": 0.05457939819898968, + "grad_norm": 3.149812936782837, + "learning_rate": 2.722283205268935e-07, + "loss": 1.0646, + "mean_token_accuracy": 0.6899845600128174, + "num_tokens": 12814977.0, + "step": 497 + }, + { + "epoch": 0.05468921590160334, + "grad_norm": 2.662902355194092, + "learning_rate": 2.7277716794731067e-07, + "loss": 1.2384, + "mean_token_accuracy": 0.6456203460693359, + "num_tokens": 12843719.0, + "step": 498 + }, + { + "epoch": 0.054799033604217, + "grad_norm": 2.84843373298645, + "learning_rate": 2.733260153677278e-07, + "loss": 1.1821, + "mean_token_accuracy": 0.6577695608139038, + "num_tokens": 12872710.0, + "step": 499 + }, + { + "epoch": 0.05490885130683066, + "grad_norm": 2.93611741065979, + "learning_rate": 2.7387486278814485e-07, + "loss": 1.0986, + "mean_token_accuracy": 0.675508975982666, + "num_tokens": 12895933.0, + "step": 500 + }, + { + "epoch": 0.05501866900944432, + "grad_norm": 2.689037322998047, + "learning_rate": 2.74423710208562e-07, + "loss": 1.116, + "mean_token_accuracy": 0.6754105091094971, + "num_tokens": 12920512.0, + "step": 501 + }, + { + "epoch": 0.055128486712057986, + "grad_norm": 2.4194495677948, + "learning_rate": 2.7497255762897913e-07, + "loss": 1.2185, + "mean_token_accuracy": 0.6502492427825928, + "num_tokens": 12951148.0, + "step": 502 + }, + { + "epoch": 0.055238304414671646, + "grad_norm": 3.2149345874786377, + "learning_rate": 2.7552140504939624e-07, + "loss": 1.1686, + "mean_token_accuracy": 0.654903769493103, + "num_tokens": 12972261.0, + "step": 503 + }, + { + "epoch": 0.055348122117285306, + "grad_norm": 2.569957971572876, + "learning_rate": 2.7607025246981336e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.6776187419891357, + "num_tokens": 13000322.0, + "step": 504 + }, + { + "epoch": 0.055457939819898966, + "grad_norm": 3.2056877613067627, + "learning_rate": 2.7661909989023053e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6960643529891968, + "num_tokens": 13021532.0, + "step": 505 + }, + { + "epoch": 0.055567757522512626, + "grad_norm": 2.8401591777801514, + "learning_rate": 2.7716794731064764e-07, + "loss": 1.1837, + "mean_token_accuracy": 0.664267361164093, + "num_tokens": 13049005.0, + "step": 506 + }, + { + "epoch": 0.05567757522512629, + "grad_norm": 3.0210423469543457, + "learning_rate": 2.7771679473106476e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.6827904582023621, + "num_tokens": 13072138.0, + "step": 507 + }, + { + "epoch": 0.05578739292773995, + "grad_norm": 3.0226352214813232, + "learning_rate": 2.7826564215148187e-07, + "loss": 1.1519, + "mean_token_accuracy": 0.6677789688110352, + "num_tokens": 13093520.0, + "step": 508 + }, + { + "epoch": 0.05589721063035361, + "grad_norm": 2.85076904296875, + "learning_rate": 2.78814489571899e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6913706064224243, + "num_tokens": 13114981.0, + "step": 509 + }, + { + "epoch": 0.05600702833296727, + "grad_norm": 2.952091693878174, + "learning_rate": 2.793633369923161e-07, + "loss": 1.0402, + "mean_token_accuracy": 0.6978510618209839, + "num_tokens": 13134781.0, + "step": 510 + }, + { + "epoch": 0.05611684603558094, + "grad_norm": 2.8876659870147705, + "learning_rate": 2.799121844127332e-07, + "loss": 1.1718, + "mean_token_accuracy": 0.6635299921035767, + "num_tokens": 13160229.0, + "step": 511 + }, + { + "epoch": 0.0562266637381946, + "grad_norm": 2.9895405769348145, + "learning_rate": 2.804610318331504e-07, + "loss": 1.1261, + "mean_token_accuracy": 0.6755766272544861, + "num_tokens": 13182425.0, + "step": 512 + }, + { + "epoch": 0.05633648144080826, + "grad_norm": 3.119256019592285, + "learning_rate": 2.810098792535675e-07, + "loss": 1.05, + "mean_token_accuracy": 0.6923587322235107, + "num_tokens": 13203632.0, + "step": 513 + }, + { + "epoch": 0.05644629914342192, + "grad_norm": 3.5291671752929688, + "learning_rate": 2.815587266739846e-07, + "loss": 1.0988, + "mean_token_accuracy": 0.6766510009765625, + "num_tokens": 13221146.0, + "step": 514 + }, + { + "epoch": 0.05655611684603558, + "grad_norm": 2.909834861755371, + "learning_rate": 2.8210757409440173e-07, + "loss": 1.1195, + "mean_token_accuracy": 0.673413872718811, + "num_tokens": 13244989.0, + "step": 515 + }, + { + "epoch": 0.056665934548649245, + "grad_norm": 2.5210793018341064, + "learning_rate": 2.826564215148189e-07, + "loss": 1.1854, + "mean_token_accuracy": 0.6528964638710022, + "num_tokens": 13274213.0, + "step": 516 + }, + { + "epoch": 0.056775752251262905, + "grad_norm": 2.809049606323242, + "learning_rate": 2.83205268935236e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.6954584717750549, + "num_tokens": 13296987.0, + "step": 517 + }, + { + "epoch": 0.056885569953876565, + "grad_norm": 2.4500348567962646, + "learning_rate": 2.8375411635565313e-07, + "loss": 1.12, + "mean_token_accuracy": 0.6764472723007202, + "num_tokens": 13328383.0, + "step": 518 + }, + { + "epoch": 0.056995387656490225, + "grad_norm": 2.663135051727295, + "learning_rate": 2.843029637760702e-07, + "loss": 1.1912, + "mean_token_accuracy": 0.6588306427001953, + "num_tokens": 13355845.0, + "step": 519 + }, + { + "epoch": 0.057105205359103885, + "grad_norm": 2.7464592456817627, + "learning_rate": 2.8485181119648736e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6979736089706421, + "num_tokens": 13382234.0, + "step": 520 + }, + { + "epoch": 0.05721502306171755, + "grad_norm": 2.6682233810424805, + "learning_rate": 2.854006586169045e-07, + "loss": 1.1137, + "mean_token_accuracy": 0.6795953512191772, + "num_tokens": 13408746.0, + "step": 521 + }, + { + "epoch": 0.05732484076433121, + "grad_norm": 2.42197322845459, + "learning_rate": 2.859495060373216e-07, + "loss": 1.2081, + "mean_token_accuracy": 0.6592180132865906, + "num_tokens": 13438876.0, + "step": 522 + }, + { + "epoch": 0.05743465846694487, + "grad_norm": 2.6381659507751465, + "learning_rate": 2.8649835345773876e-07, + "loss": 1.2045, + "mean_token_accuracy": 0.6625485420227051, + "num_tokens": 13463355.0, + "step": 523 + }, + { + "epoch": 0.05754447616955853, + "grad_norm": 2.8247807025909424, + "learning_rate": 2.870472008781559e-07, + "loss": 1.1592, + "mean_token_accuracy": 0.6596149206161499, + "num_tokens": 13485004.0, + "step": 524 + }, + { + "epoch": 0.05765429387217219, + "grad_norm": 2.7841758728027344, + "learning_rate": 2.87596048298573e-07, + "loss": 1.1803, + "mean_token_accuracy": 0.6549053192138672, + "num_tokens": 13510087.0, + "step": 525 + }, + { + "epoch": 0.05776411157478586, + "grad_norm": 2.9794812202453613, + "learning_rate": 2.881448957189901e-07, + "loss": 1.084, + "mean_token_accuracy": 0.6721004247665405, + "num_tokens": 13532478.0, + "step": 526 + }, + { + "epoch": 0.05787392927739952, + "grad_norm": 2.854224681854248, + "learning_rate": 2.886937431394073e-07, + "loss": 1.134, + "mean_token_accuracy": 0.6670573949813843, + "num_tokens": 13555674.0, + "step": 527 + }, + { + "epoch": 0.05798374698001318, + "grad_norm": 2.6651015281677246, + "learning_rate": 2.8924259055982434e-07, + "loss": 1.1717, + "mean_token_accuracy": 0.6716020703315735, + "num_tokens": 13582274.0, + "step": 528 + }, + { + "epoch": 0.05809356468262684, + "grad_norm": 3.068920850753784, + "learning_rate": 2.8979143798024145e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6692076325416565, + "num_tokens": 13602530.0, + "step": 529 + }, + { + "epoch": 0.058203382385240504, + "grad_norm": 2.727147102355957, + "learning_rate": 2.9034028540065857e-07, + "loss": 1.1136, + "mean_token_accuracy": 0.671244204044342, + "num_tokens": 13627061.0, + "step": 530 + }, + { + "epoch": 0.058313200087854164, + "grad_norm": 2.7410836219787598, + "learning_rate": 2.9088913282107574e-07, + "loss": 1.0953, + "mean_token_accuracy": 0.6786634922027588, + "num_tokens": 13653827.0, + "step": 531 + }, + { + "epoch": 0.058423017790467824, + "grad_norm": 2.69441556930542, + "learning_rate": 2.9143798024149285e-07, + "loss": 1.1431, + "mean_token_accuracy": 0.6739728450775146, + "num_tokens": 13680269.0, + "step": 532 + }, + { + "epoch": 0.058532835493081484, + "grad_norm": 2.6261723041534424, + "learning_rate": 2.9198682766190997e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6855077147483826, + "num_tokens": 13706421.0, + "step": 533 + }, + { + "epoch": 0.058642653195695144, + "grad_norm": 2.9758963584899902, + "learning_rate": 2.9253567508232713e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.7045199275016785, + "num_tokens": 13726048.0, + "step": 534 + }, + { + "epoch": 0.05875247089830881, + "grad_norm": 3.296292543411255, + "learning_rate": 2.9308452250274425e-07, + "loss": 1.1224, + "mean_token_accuracy": 0.6694005131721497, + "num_tokens": 13744429.0, + "step": 535 + }, + { + "epoch": 0.05886228860092247, + "grad_norm": 2.822077751159668, + "learning_rate": 2.9363336992316136e-07, + "loss": 1.111, + "mean_token_accuracy": 0.6787686347961426, + "num_tokens": 13768669.0, + "step": 536 + }, + { + "epoch": 0.05897210630353613, + "grad_norm": 2.7185966968536377, + "learning_rate": 2.941822173435785e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6798032522201538, + "num_tokens": 13792792.0, + "step": 537 + }, + { + "epoch": 0.05908192400614979, + "grad_norm": 2.3440029621124268, + "learning_rate": 2.947310647639956e-07, + "loss": 1.1106, + "mean_token_accuracy": 0.6809860467910767, + "num_tokens": 13824798.0, + "step": 538 + }, + { + "epoch": 0.05919174170876345, + "grad_norm": 2.348109245300293, + "learning_rate": 2.952799121844127e-07, + "loss": 1.1771, + "mean_token_accuracy": 0.6560312509536743, + "num_tokens": 13855798.0, + "step": 539 + }, + { + "epoch": 0.05930155941137712, + "grad_norm": 2.5476489067077637, + "learning_rate": 2.958287596048298e-07, + "loss": 1.2054, + "mean_token_accuracy": 0.6528757214546204, + "num_tokens": 13884838.0, + "step": 540 + }, + { + "epoch": 0.059411377113990776, + "grad_norm": 2.816110134124756, + "learning_rate": 2.9637760702524694e-07, + "loss": 1.1831, + "mean_token_accuracy": 0.6609179973602295, + "num_tokens": 13908538.0, + "step": 541 + }, + { + "epoch": 0.059521194816604436, + "grad_norm": 2.4698245525360107, + "learning_rate": 2.969264544456641e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6836147308349609, + "num_tokens": 13935660.0, + "step": 542 + }, + { + "epoch": 0.059631012519218096, + "grad_norm": 2.59395432472229, + "learning_rate": 2.974753018660812e-07, + "loss": 1.1318, + "mean_token_accuracy": 0.6818431615829468, + "num_tokens": 13962258.0, + "step": 543 + }, + { + "epoch": 0.059740830221831756, + "grad_norm": 2.429302453994751, + "learning_rate": 2.9802414928649834e-07, + "loss": 1.1786, + "mean_token_accuracy": 0.6600596904754639, + "num_tokens": 13991987.0, + "step": 544 + }, + { + "epoch": 0.05985064792444542, + "grad_norm": 2.6945748329162598, + "learning_rate": 2.9857299670691546e-07, + "loss": 1.1892, + "mean_token_accuracy": 0.6570158004760742, + "num_tokens": 14017394.0, + "step": 545 + }, + { + "epoch": 0.05996046562705908, + "grad_norm": 2.6865479946136475, + "learning_rate": 2.991218441273326e-07, + "loss": 1.1014, + "mean_token_accuracy": 0.6768734455108643, + "num_tokens": 14046663.0, + "step": 546 + }, + { + "epoch": 0.06007028332967274, + "grad_norm": 2.4721672534942627, + "learning_rate": 2.996706915477497e-07, + "loss": 1.117, + "mean_token_accuracy": 0.674640417098999, + "num_tokens": 14075696.0, + "step": 547 + }, + { + "epoch": 0.0601801010322864, + "grad_norm": 2.2165067195892334, + "learning_rate": 3.002195389681668e-07, + "loss": 1.2451, + "mean_token_accuracy": 0.6390098333358765, + "num_tokens": 14112763.0, + "step": 548 + }, + { + "epoch": 0.06028991873490007, + "grad_norm": 2.4320411682128906, + "learning_rate": 3.0076838638858397e-07, + "loss": 1.1052, + "mean_token_accuracy": 0.6746840476989746, + "num_tokens": 14141115.0, + "step": 549 + }, + { + "epoch": 0.06039973643751373, + "grad_norm": 2.5588414669036865, + "learning_rate": 3.013172338090011e-07, + "loss": 1.1477, + "mean_token_accuracy": 0.6680233478546143, + "num_tokens": 14170499.0, + "step": 550 + }, + { + "epoch": 0.06050955414012739, + "grad_norm": 2.4618091583251953, + "learning_rate": 3.018660812294182e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6817662715911865, + "num_tokens": 14199124.0, + "step": 551 + }, + { + "epoch": 0.06061937184274105, + "grad_norm": 2.660133123397827, + "learning_rate": 3.024149286498353e-07, + "loss": 1.181, + "mean_token_accuracy": 0.6578633785247803, + "num_tokens": 14224526.0, + "step": 552 + }, + { + "epoch": 0.06072918954535471, + "grad_norm": 2.4507503509521484, + "learning_rate": 3.029637760702525e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.6822704672813416, + "num_tokens": 14255630.0, + "step": 553 + }, + { + "epoch": 0.060839007247968376, + "grad_norm": 2.3329403400421143, + "learning_rate": 3.035126234906696e-07, + "loss": 1.2275, + "mean_token_accuracy": 0.6507984399795532, + "num_tokens": 14288954.0, + "step": 554 + }, + { + "epoch": 0.060948824950582035, + "grad_norm": 2.7817652225494385, + "learning_rate": 3.040614709110867e-07, + "loss": 1.1605, + "mean_token_accuracy": 0.6636850833892822, + "num_tokens": 14314570.0, + "step": 555 + }, + { + "epoch": 0.061058642653195695, + "grad_norm": 3.0154857635498047, + "learning_rate": 3.046103183315038e-07, + "loss": 1.1547, + "mean_token_accuracy": 0.671459972858429, + "num_tokens": 14334555.0, + "step": 556 + }, + { + "epoch": 0.061168460355809355, + "grad_norm": 2.744490623474121, + "learning_rate": 3.0515916575192094e-07, + "loss": 1.1196, + "mean_token_accuracy": 0.677564263343811, + "num_tokens": 14357512.0, + "step": 557 + }, + { + "epoch": 0.061278278058423015, + "grad_norm": 2.603560209274292, + "learning_rate": 3.0570801317233806e-07, + "loss": 1.1413, + "mean_token_accuracy": 0.6710929870605469, + "num_tokens": 14385797.0, + "step": 558 + }, + { + "epoch": 0.06138809576103668, + "grad_norm": 2.476578712463379, + "learning_rate": 3.062568605927552e-07, + "loss": 1.0613, + "mean_token_accuracy": 0.6881267428398132, + "num_tokens": 14413278.0, + "step": 559 + }, + { + "epoch": 0.06149791346365034, + "grad_norm": 3.0184855461120605, + "learning_rate": 3.0680570801317234e-07, + "loss": 1.1245, + "mean_token_accuracy": 0.6719378232955933, + "num_tokens": 14437703.0, + "step": 560 + }, + { + "epoch": 0.061607731166264, + "grad_norm": 2.6417784690856934, + "learning_rate": 3.0735455543358946e-07, + "loss": 1.2005, + "mean_token_accuracy": 0.6665399670600891, + "num_tokens": 14462598.0, + "step": 561 + }, + { + "epoch": 0.06171754886887766, + "grad_norm": 2.1721231937408447, + "learning_rate": 3.0790340285400657e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6819611191749573, + "num_tokens": 14499426.0, + "step": 562 + }, + { + "epoch": 0.06182736657149132, + "grad_norm": 2.749406337738037, + "learning_rate": 3.084522502744237e-07, + "loss": 1.0918, + "mean_token_accuracy": 0.6790647506713867, + "num_tokens": 14522657.0, + "step": 563 + }, + { + "epoch": 0.06193718427410499, + "grad_norm": 2.2048325538635254, + "learning_rate": 3.0900109769484086e-07, + "loss": 1.2432, + "mean_token_accuracy": 0.642806887626648, + "num_tokens": 14557858.0, + "step": 564 + }, + { + "epoch": 0.06204700197671865, + "grad_norm": 2.794625759124756, + "learning_rate": 3.0954994511525797e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6718685626983643, + "num_tokens": 14582920.0, + "step": 565 + }, + { + "epoch": 0.06215681967933231, + "grad_norm": 2.8614532947540283, + "learning_rate": 3.1009879253567503e-07, + "loss": 1.1521, + "mean_token_accuracy": 0.667252779006958, + "num_tokens": 14605034.0, + "step": 566 + }, + { + "epoch": 0.06226663738194597, + "grad_norm": 2.2233941555023193, + "learning_rate": 3.1064763995609215e-07, + "loss": 1.1618, + "mean_token_accuracy": 0.6687324047088623, + "num_tokens": 14640901.0, + "step": 567 + }, + { + "epoch": 0.062376455084559634, + "grad_norm": 2.3480963706970215, + "learning_rate": 3.111964873765093e-07, + "loss": 1.1542, + "mean_token_accuracy": 0.6647657752037048, + "num_tokens": 14671232.0, + "step": 568 + }, + { + "epoch": 0.062486272787173294, + "grad_norm": 2.75262451171875, + "learning_rate": 3.1174533479692643e-07, + "loss": 1.0996, + "mean_token_accuracy": 0.681022584438324, + "num_tokens": 14694067.0, + "step": 569 + }, + { + "epoch": 0.06259609048978695, + "grad_norm": 2.2894465923309326, + "learning_rate": 3.1229418221734355e-07, + "loss": 1.219, + "mean_token_accuracy": 0.6521063446998596, + "num_tokens": 14728469.0, + "step": 570 + }, + { + "epoch": 0.06270590819240061, + "grad_norm": 3.1783764362335205, + "learning_rate": 3.128430296377607e-07, + "loss": 1.0316, + "mean_token_accuracy": 0.6969385147094727, + "num_tokens": 14748378.0, + "step": 571 + }, + { + "epoch": 0.06281572589501427, + "grad_norm": 2.6995503902435303, + "learning_rate": 3.1339187705817783e-07, + "loss": 1.1545, + "mean_token_accuracy": 0.6532487869262695, + "num_tokens": 14774095.0, + "step": 572 + }, + { + "epoch": 0.06292554359762793, + "grad_norm": 2.815106153488159, + "learning_rate": 3.1394072447859495e-07, + "loss": 1.08, + "mean_token_accuracy": 0.6823896169662476, + "num_tokens": 14797438.0, + "step": 573 + }, + { + "epoch": 0.0630353613002416, + "grad_norm": 2.493232011795044, + "learning_rate": 3.1448957189901206e-07, + "loss": 1.113, + "mean_token_accuracy": 0.6788531541824341, + "num_tokens": 14826412.0, + "step": 574 + }, + { + "epoch": 0.06314517900285525, + "grad_norm": 2.517422914505005, + "learning_rate": 3.1503841931942923e-07, + "loss": 1.134, + "mean_token_accuracy": 0.6695951819419861, + "num_tokens": 14855382.0, + "step": 575 + }, + { + "epoch": 0.06325499670546893, + "grad_norm": 2.898879289627075, + "learning_rate": 3.155872667398463e-07, + "loss": 1.1002, + "mean_token_accuracy": 0.6782481670379639, + "num_tokens": 14877884.0, + "step": 576 + }, + { + "epoch": 0.06336481440808259, + "grad_norm": 2.1317689418792725, + "learning_rate": 3.161361141602634e-07, + "loss": 1.2996, + "mean_token_accuracy": 0.627497136592865, + "num_tokens": 14914401.0, + "step": 577 + }, + { + "epoch": 0.06347463211069625, + "grad_norm": 2.3872475624084473, + "learning_rate": 3.166849615806805e-07, + "loss": 1.1084, + "mean_token_accuracy": 0.6758533120155334, + "num_tokens": 14944549.0, + "step": 578 + }, + { + "epoch": 0.0635844498133099, + "grad_norm": 2.595644235610962, + "learning_rate": 3.172338090010977e-07, + "loss": 1.1828, + "mean_token_accuracy": 0.6626525521278381, + "num_tokens": 14972122.0, + "step": 579 + }, + { + "epoch": 0.06369426751592357, + "grad_norm": 2.654045343399048, + "learning_rate": 3.177826564215148e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6833177208900452, + "num_tokens": 14998340.0, + "step": 580 + }, + { + "epoch": 0.06380408521853723, + "grad_norm": 2.837050676345825, + "learning_rate": 3.183315038419319e-07, + "loss": 1.0961, + "mean_token_accuracy": 0.6784970164299011, + "num_tokens": 15020696.0, + "step": 581 + }, + { + "epoch": 0.06391390292115089, + "grad_norm": 2.63215708732605, + "learning_rate": 3.188803512623491e-07, + "loss": 1.227, + "mean_token_accuracy": 0.6422814130783081, + "num_tokens": 15047343.0, + "step": 582 + }, + { + "epoch": 0.06402372062376455, + "grad_norm": 2.697110414505005, + "learning_rate": 3.194291986827662e-07, + "loss": 1.1199, + "mean_token_accuracy": 0.6730882525444031, + "num_tokens": 15070899.0, + "step": 583 + }, + { + "epoch": 0.0641335383263782, + "grad_norm": 3.021003484725952, + "learning_rate": 3.199780461031833e-07, + "loss": 1.1615, + "mean_token_accuracy": 0.6743926405906677, + "num_tokens": 15093104.0, + "step": 584 + }, + { + "epoch": 0.06424335602899188, + "grad_norm": 2.5849671363830566, + "learning_rate": 3.205268935236004e-07, + "loss": 1.214, + "mean_token_accuracy": 0.6442333459854126, + "num_tokens": 15120011.0, + "step": 585 + }, + { + "epoch": 0.06435317373160554, + "grad_norm": 3.014357328414917, + "learning_rate": 3.2107574094401755e-07, + "loss": 1.1661, + "mean_token_accuracy": 0.6668482422828674, + "num_tokens": 15139932.0, + "step": 586 + }, + { + "epoch": 0.0644629914342192, + "grad_norm": 2.490893840789795, + "learning_rate": 3.2162458836443467e-07, + "loss": 1.03, + "mean_token_accuracy": 0.7025210857391357, + "num_tokens": 15166788.0, + "step": 587 + }, + { + "epoch": 0.06457280913683286, + "grad_norm": 2.508673906326294, + "learning_rate": 3.221734357848518e-07, + "loss": 1.1904, + "mean_token_accuracy": 0.6645998954772949, + "num_tokens": 15194734.0, + "step": 588 + }, + { + "epoch": 0.06468262683944652, + "grad_norm": 2.658155679702759, + "learning_rate": 3.227222832052689e-07, + "loss": 1.1263, + "mean_token_accuracy": 0.6748684048652649, + "num_tokens": 15220931.0, + "step": 589 + }, + { + "epoch": 0.06479244454206018, + "grad_norm": 2.2423362731933594, + "learning_rate": 3.2327113062568606e-07, + "loss": 1.0861, + "mean_token_accuracy": 0.690997838973999, + "num_tokens": 15251542.0, + "step": 590 + }, + { + "epoch": 0.06490226224467384, + "grad_norm": 2.8958446979522705, + "learning_rate": 3.238199780461032e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6867309808731079, + "num_tokens": 15272372.0, + "step": 591 + }, + { + "epoch": 0.0650120799472875, + "grad_norm": 2.4720616340637207, + "learning_rate": 3.243688254665203e-07, + "loss": 1.106, + "mean_token_accuracy": 0.6776013374328613, + "num_tokens": 15299778.0, + "step": 592 + }, + { + "epoch": 0.06512189764990116, + "grad_norm": 3.0516979694366455, + "learning_rate": 3.2491767288693746e-07, + "loss": 0.9214, + "mean_token_accuracy": 0.7211612462997437, + "num_tokens": 15317294.0, + "step": 593 + }, + { + "epoch": 0.06523171535251482, + "grad_norm": 2.445366382598877, + "learning_rate": 3.254665203073546e-07, + "loss": 1.0049, + "mean_token_accuracy": 0.7026647329330444, + "num_tokens": 15343675.0, + "step": 594 + }, + { + "epoch": 0.06534153305512849, + "grad_norm": 2.708918333053589, + "learning_rate": 3.2601536772777164e-07, + "loss": 1.109, + "mean_token_accuracy": 0.679847776889801, + "num_tokens": 15367489.0, + "step": 595 + }, + { + "epoch": 0.06545135075774215, + "grad_norm": 2.3389573097229004, + "learning_rate": 3.2656421514818876e-07, + "loss": 1.0696, + "mean_token_accuracy": 0.6910394430160522, + "num_tokens": 15399494.0, + "step": 596 + }, + { + "epoch": 0.06556116846035581, + "grad_norm": 2.4477720260620117, + "learning_rate": 3.271130625686059e-07, + "loss": 1.1383, + "mean_token_accuracy": 0.6717137694358826, + "num_tokens": 15426657.0, + "step": 597 + }, + { + "epoch": 0.06567098616296947, + "grad_norm": 2.3603289127349854, + "learning_rate": 3.2766190998902304e-07, + "loss": 1.1113, + "mean_token_accuracy": 0.6773684620857239, + "num_tokens": 15457420.0, + "step": 598 + }, + { + "epoch": 0.06578080386558313, + "grad_norm": 2.7708637714385986, + "learning_rate": 3.2821075740944015e-07, + "loss": 1.1372, + "mean_token_accuracy": 0.6756311655044556, + "num_tokens": 15479564.0, + "step": 599 + }, + { + "epoch": 0.06589062156819679, + "grad_norm": 2.7398102283477783, + "learning_rate": 3.2875960482985727e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6810094714164734, + "num_tokens": 15501520.0, + "step": 600 + }, + { + "epoch": 0.06600043927081045, + "grad_norm": 2.3736495971679688, + "learning_rate": 3.2930845225027444e-07, + "loss": 1.2161, + "mean_token_accuracy": 0.6546581983566284, + "num_tokens": 15532339.0, + "step": 601 + }, + { + "epoch": 0.06611025697342411, + "grad_norm": 2.488430976867676, + "learning_rate": 3.2985729967069155e-07, + "loss": 1.0361, + "mean_token_accuracy": 0.6978837251663208, + "num_tokens": 15557250.0, + "step": 602 + }, + { + "epoch": 0.06622007467603777, + "grad_norm": 2.580273389816284, + "learning_rate": 3.3040614709110867e-07, + "loss": 1.1351, + "mean_token_accuracy": 0.6713882684707642, + "num_tokens": 15581414.0, + "step": 603 + }, + { + "epoch": 0.06632989237865144, + "grad_norm": 2.1972763538360596, + "learning_rate": 3.3095499451152573e-07, + "loss": 1.1158, + "mean_token_accuracy": 0.6752565503120422, + "num_tokens": 15614274.0, + "step": 604 + }, + { + "epoch": 0.0664397100812651, + "grad_norm": 2.8436644077301025, + "learning_rate": 3.315038419319429e-07, + "loss": 1.0752, + "mean_token_accuracy": 0.6804741024971008, + "num_tokens": 15633905.0, + "step": 605 + }, + { + "epoch": 0.06654952778387876, + "grad_norm": 2.489518642425537, + "learning_rate": 3.3205268935236e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6984946131706238, + "num_tokens": 15657974.0, + "step": 606 + }, + { + "epoch": 0.06665934548649242, + "grad_norm": 2.320643186569214, + "learning_rate": 3.3260153677277713e-07, + "loss": 1.1714, + "mean_token_accuracy": 0.6546472907066345, + "num_tokens": 15688091.0, + "step": 607 + }, + { + "epoch": 0.06676916318910608, + "grad_norm": 2.746554136276245, + "learning_rate": 3.331503841931943e-07, + "loss": 1.1829, + "mean_token_accuracy": 0.6598870754241943, + "num_tokens": 15710030.0, + "step": 608 + }, + { + "epoch": 0.06687898089171974, + "grad_norm": 2.720459222793579, + "learning_rate": 3.336992316136114e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6872288584709167, + "num_tokens": 15734356.0, + "step": 609 + }, + { + "epoch": 0.0669887985943334, + "grad_norm": 2.238973617553711, + "learning_rate": 3.3424807903402853e-07, + "loss": 1.1335, + "mean_token_accuracy": 0.6645327806472778, + "num_tokens": 15765554.0, + "step": 610 + }, + { + "epoch": 0.06709861629694706, + "grad_norm": 2.559215784072876, + "learning_rate": 3.3479692645444564e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6958437561988831, + "num_tokens": 15791543.0, + "step": 611 + }, + { + "epoch": 0.06720843399956072, + "grad_norm": 2.6604833602905273, + "learning_rate": 3.353457738748628e-07, + "loss": 1.0468, + "mean_token_accuracy": 0.6915310621261597, + "num_tokens": 15814345.0, + "step": 612 + }, + { + "epoch": 0.06731825170217438, + "grad_norm": 2.6394217014312744, + "learning_rate": 3.3589462129527993e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.6750473380088806, + "num_tokens": 15839038.0, + "step": 613 + }, + { + "epoch": 0.06742806940478806, + "grad_norm": 2.2986292839050293, + "learning_rate": 3.36443468715697e-07, + "loss": 1.0854, + "mean_token_accuracy": 0.6926525831222534, + "num_tokens": 15869277.0, + "step": 614 + }, + { + "epoch": 0.06753788710740172, + "grad_norm": 2.871350049972534, + "learning_rate": 3.369923161361141e-07, + "loss": 1.1255, + "mean_token_accuracy": 0.6796224117279053, + "num_tokens": 15890091.0, + "step": 615 + }, + { + "epoch": 0.06764770481001538, + "grad_norm": 2.469928503036499, + "learning_rate": 3.3754116355653127e-07, + "loss": 1.0761, + "mean_token_accuracy": 0.6899073123931885, + "num_tokens": 15914804.0, + "step": 616 + }, + { + "epoch": 0.06775752251262904, + "grad_norm": 2.739485502243042, + "learning_rate": 3.380900109769484e-07, + "loss": 1.1739, + "mean_token_accuracy": 0.6594994068145752, + "num_tokens": 15937391.0, + "step": 617 + }, + { + "epoch": 0.0678673402152427, + "grad_norm": 2.5423166751861572, + "learning_rate": 3.386388583973655e-07, + "loss": 1.0941, + "mean_token_accuracy": 0.6730822324752808, + "num_tokens": 15961243.0, + "step": 618 + }, + { + "epoch": 0.06797715791785636, + "grad_norm": 2.2108168601989746, + "learning_rate": 3.3918770581778267e-07, + "loss": 1.1218, + "mean_token_accuracy": 0.6668039560317993, + "num_tokens": 15993867.0, + "step": 619 + }, + { + "epoch": 0.06808697562047002, + "grad_norm": 2.551823616027832, + "learning_rate": 3.397365532381998e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6805397272109985, + "num_tokens": 16020634.0, + "step": 620 + }, + { + "epoch": 0.06819679332308368, + "grad_norm": 2.456037998199463, + "learning_rate": 3.402854006586169e-07, + "loss": 1.1792, + "mean_token_accuracy": 0.6523274779319763, + "num_tokens": 16049229.0, + "step": 621 + }, + { + "epoch": 0.06830661102569734, + "grad_norm": 2.6045498847961426, + "learning_rate": 3.40834248079034e-07, + "loss": 1.1503, + "mean_token_accuracy": 0.663906455039978, + "num_tokens": 16073330.0, + "step": 622 + }, + { + "epoch": 0.06841642872831101, + "grad_norm": 2.723710536956787, + "learning_rate": 3.4138309549945113e-07, + "loss": 1.0792, + "mean_token_accuracy": 0.6831985712051392, + "num_tokens": 16096241.0, + "step": 623 + }, + { + "epoch": 0.06852624643092467, + "grad_norm": 2.804187297821045, + "learning_rate": 3.4193194291986825e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.6825786828994751, + "num_tokens": 16117923.0, + "step": 624 + }, + { + "epoch": 0.06863606413353833, + "grad_norm": 2.6598682403564453, + "learning_rate": 3.4248079034028536e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.689720094203949, + "num_tokens": 16139935.0, + "step": 625 + }, + { + "epoch": 0.06874588183615199, + "grad_norm": 2.585725784301758, + "learning_rate": 3.430296377607025e-07, + "loss": 1.1137, + "mean_token_accuracy": 0.6776716113090515, + "num_tokens": 16164806.0, + "step": 626 + }, + { + "epoch": 0.06885569953876565, + "grad_norm": 2.564257860183716, + "learning_rate": 3.4357848518111965e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6885042190551758, + "num_tokens": 16189621.0, + "step": 627 + }, + { + "epoch": 0.06896551724137931, + "grad_norm": 2.768909215927124, + "learning_rate": 3.4412733260153676e-07, + "loss": 1.1102, + "mean_token_accuracy": 0.6801679134368896, + "num_tokens": 16211332.0, + "step": 628 + }, + { + "epoch": 0.06907533494399297, + "grad_norm": 2.160489320755005, + "learning_rate": 3.446761800219539e-07, + "loss": 1.1009, + "mean_token_accuracy": 0.6882023811340332, + "num_tokens": 16242539.0, + "step": 629 + }, + { + "epoch": 0.06918515264660663, + "grad_norm": 2.808720588684082, + "learning_rate": 3.4522502744237104e-07, + "loss": 1.0317, + "mean_token_accuracy": 0.7024773359298706, + "num_tokens": 16263866.0, + "step": 630 + }, + { + "epoch": 0.06929497034922029, + "grad_norm": 2.1077558994293213, + "learning_rate": 3.4577387486278816e-07, + "loss": 1.118, + "mean_token_accuracy": 0.675731897354126, + "num_tokens": 16298406.0, + "step": 631 + }, + { + "epoch": 0.06940478805183395, + "grad_norm": 2.333911180496216, + "learning_rate": 3.463227222832053e-07, + "loss": 1.0963, + "mean_token_accuracy": 0.6762607097625732, + "num_tokens": 16326483.0, + "step": 632 + }, + { + "epoch": 0.06951460575444762, + "grad_norm": 2.3963263034820557, + "learning_rate": 3.4687156970362234e-07, + "loss": 1.0387, + "mean_token_accuracy": 0.693041980266571, + "num_tokens": 16352026.0, + "step": 633 + }, + { + "epoch": 0.06962442345706128, + "grad_norm": 2.9177911281585693, + "learning_rate": 3.474204171240395e-07, + "loss": 1.2429, + "mean_token_accuracy": 0.6611645817756653, + "num_tokens": 16372806.0, + "step": 634 + }, + { + "epoch": 0.06973424115967494, + "grad_norm": 2.7347359657287598, + "learning_rate": 3.479692645444566e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6872312426567078, + "num_tokens": 16396930.0, + "step": 635 + }, + { + "epoch": 0.0698440588622886, + "grad_norm": 2.525460720062256, + "learning_rate": 3.4851811196487374e-07, + "loss": 1.1846, + "mean_token_accuracy": 0.6596242785453796, + "num_tokens": 16422334.0, + "step": 636 + }, + { + "epoch": 0.06995387656490226, + "grad_norm": 2.822389841079712, + "learning_rate": 3.4906695938529085e-07, + "loss": 1.1773, + "mean_token_accuracy": 0.6575550436973572, + "num_tokens": 16446183.0, + "step": 637 + }, + { + "epoch": 0.07006369426751592, + "grad_norm": 2.264967441558838, + "learning_rate": 3.49615806805708e-07, + "loss": 1.0588, + "mean_token_accuracy": 0.6910548806190491, + "num_tokens": 16473804.0, + "step": 638 + }, + { + "epoch": 0.07017351197012958, + "grad_norm": 2.8432531356811523, + "learning_rate": 3.5016465422612513e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6927735209465027, + "num_tokens": 16494715.0, + "step": 639 + }, + { + "epoch": 0.07028332967274324, + "grad_norm": 3.102787733078003, + "learning_rate": 3.5071350164654225e-07, + "loss": 1.0076, + "mean_token_accuracy": 0.7017433643341064, + "num_tokens": 16514778.0, + "step": 640 + }, + { + "epoch": 0.0703931473753569, + "grad_norm": 2.2419304847717285, + "learning_rate": 3.512623490669594e-07, + "loss": 1.1873, + "mean_token_accuracy": 0.6559606790542603, + "num_tokens": 16544203.0, + "step": 641 + }, + { + "epoch": 0.07050296507797058, + "grad_norm": 2.725580930709839, + "learning_rate": 3.518111964873765e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.6832288503646851, + "num_tokens": 16564859.0, + "step": 642 + }, + { + "epoch": 0.07061278278058424, + "grad_norm": 2.9876914024353027, + "learning_rate": 3.523600439077936e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.6827144622802734, + "num_tokens": 16584235.0, + "step": 643 + }, + { + "epoch": 0.0707226004831979, + "grad_norm": 2.43678617477417, + "learning_rate": 3.529088913282107e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.6701573133468628, + "num_tokens": 16609823.0, + "step": 644 + }, + { + "epoch": 0.07083241818581155, + "grad_norm": 2.515495777130127, + "learning_rate": 3.534577387486279e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.6753917336463928, + "num_tokens": 16633684.0, + "step": 645 + }, + { + "epoch": 0.07094223588842521, + "grad_norm": 2.2846784591674805, + "learning_rate": 3.54006586169045e-07, + "loss": 1.1791, + "mean_token_accuracy": 0.6568848490715027, + "num_tokens": 16663348.0, + "step": 646 + }, + { + "epoch": 0.07105205359103887, + "grad_norm": 2.375922203063965, + "learning_rate": 3.545554335894621e-07, + "loss": 1.2038, + "mean_token_accuracy": 0.6597906351089478, + "num_tokens": 16693019.0, + "step": 647 + }, + { + "epoch": 0.07116187129365253, + "grad_norm": 2.665825843811035, + "learning_rate": 3.551042810098792e-07, + "loss": 1.1805, + "mean_token_accuracy": 0.6683536767959595, + "num_tokens": 16717017.0, + "step": 648 + }, + { + "epoch": 0.0712716889962662, + "grad_norm": 2.5122010707855225, + "learning_rate": 3.556531284302964e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.7004982233047485, + "num_tokens": 16740747.0, + "step": 649 + }, + { + "epoch": 0.07138150669887985, + "grad_norm": 2.4978456497192383, + "learning_rate": 3.562019758507135e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.684415340423584, + "num_tokens": 16765947.0, + "step": 650 + }, + { + "epoch": 0.07149132440149351, + "grad_norm": 2.8762593269348145, + "learning_rate": 3.5675082327113057e-07, + "loss": 1.1499, + "mean_token_accuracy": 0.6703606247901917, + "num_tokens": 16785492.0, + "step": 651 + }, + { + "epoch": 0.07160114210410719, + "grad_norm": 2.2868258953094482, + "learning_rate": 3.5729967069154774e-07, + "loss": 1.069, + "mean_token_accuracy": 0.6842790842056274, + "num_tokens": 16814230.0, + "step": 652 + }, + { + "epoch": 0.07171095980672085, + "grad_norm": 2.3168997764587402, + "learning_rate": 3.5784851811196485e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6788545846939087, + "num_tokens": 16842705.0, + "step": 653 + }, + { + "epoch": 0.07182077750933451, + "grad_norm": 2.5029420852661133, + "learning_rate": 3.5839736553238197e-07, + "loss": 1.1302, + "mean_token_accuracy": 0.6684993505477905, + "num_tokens": 16868509.0, + "step": 654 + }, + { + "epoch": 0.07193059521194817, + "grad_norm": 2.839768886566162, + "learning_rate": 3.589462129527991e-07, + "loss": 1.027, + "mean_token_accuracy": 0.6924635171890259, + "num_tokens": 16890252.0, + "step": 655 + }, + { + "epoch": 0.07204041291456183, + "grad_norm": 2.538461923599243, + "learning_rate": 3.5949506037321625e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6823430061340332, + "num_tokens": 16915941.0, + "step": 656 + }, + { + "epoch": 0.07215023061717549, + "grad_norm": 2.654898166656494, + "learning_rate": 3.6004390779363337e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6542962193489075, + "num_tokens": 16939581.0, + "step": 657 + }, + { + "epoch": 0.07226004831978915, + "grad_norm": 2.2875027656555176, + "learning_rate": 3.605927552140505e-07, + "loss": 1.169, + "mean_token_accuracy": 0.664458692073822, + "num_tokens": 16969441.0, + "step": 658 + }, + { + "epoch": 0.0723698660224028, + "grad_norm": 2.738332986831665, + "learning_rate": 3.611416026344676e-07, + "loss": 1.1304, + "mean_token_accuracy": 0.6687246561050415, + "num_tokens": 16991497.0, + "step": 659 + }, + { + "epoch": 0.07247968372501647, + "grad_norm": 2.501424551010132, + "learning_rate": 3.6169045005488477e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6774454116821289, + "num_tokens": 17015919.0, + "step": 660 + }, + { + "epoch": 0.07258950142763014, + "grad_norm": 2.3603992462158203, + "learning_rate": 3.6223929747530183e-07, + "loss": 1.097, + "mean_token_accuracy": 0.6861397624015808, + "num_tokens": 17046007.0, + "step": 661 + }, + { + "epoch": 0.0726993191302438, + "grad_norm": 2.3654024600982666, + "learning_rate": 3.6278814489571894e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6725050210952759, + "num_tokens": 17073163.0, + "step": 662 + }, + { + "epoch": 0.07280913683285746, + "grad_norm": 2.443650722503662, + "learning_rate": 3.6333699231613606e-07, + "loss": 1.1828, + "mean_token_accuracy": 0.6547626256942749, + "num_tokens": 17101877.0, + "step": 663 + }, + { + "epoch": 0.07291895453547112, + "grad_norm": 2.4066381454467773, + "learning_rate": 3.6388583973655323e-07, + "loss": 1.1541, + "mean_token_accuracy": 0.6621950268745422, + "num_tokens": 17128962.0, + "step": 664 + }, + { + "epoch": 0.07302877223808478, + "grad_norm": 2.4221489429473877, + "learning_rate": 3.6443468715697034e-07, + "loss": 1.088, + "mean_token_accuracy": 0.6829213500022888, + "num_tokens": 17156670.0, + "step": 665 + }, + { + "epoch": 0.07313858994069844, + "grad_norm": 2.3297955989837646, + "learning_rate": 3.6498353457738746e-07, + "loss": 1.0918, + "mean_token_accuracy": 0.6781599521636963, + "num_tokens": 17183420.0, + "step": 666 + }, + { + "epoch": 0.0732484076433121, + "grad_norm": 2.4257616996765137, + "learning_rate": 3.655323819978046e-07, + "loss": 1.0651, + "mean_token_accuracy": 0.681668758392334, + "num_tokens": 17208903.0, + "step": 667 + }, + { + "epoch": 0.07335822534592576, + "grad_norm": 2.6271581649780273, + "learning_rate": 3.6608122941822174e-07, + "loss": 1.1764, + "mean_token_accuracy": 0.6693969964981079, + "num_tokens": 17232742.0, + "step": 668 + }, + { + "epoch": 0.07346804304853942, + "grad_norm": 2.7195241451263428, + "learning_rate": 3.6663007683863886e-07, + "loss": 1.0053, + "mean_token_accuracy": 0.7030537128448486, + "num_tokens": 17253493.0, + "step": 669 + }, + { + "epoch": 0.07357786075115308, + "grad_norm": 2.4222970008850098, + "learning_rate": 3.671789242590559e-07, + "loss": 1.1164, + "mean_token_accuracy": 0.6780850291252136, + "num_tokens": 17279044.0, + "step": 670 + }, + { + "epoch": 0.07368767845376675, + "grad_norm": 2.242434501647949, + "learning_rate": 3.677277716794731e-07, + "loss": 1.1065, + "mean_token_accuracy": 0.6817284822463989, + "num_tokens": 17307545.0, + "step": 671 + }, + { + "epoch": 0.07379749615638041, + "grad_norm": 2.4573147296905518, + "learning_rate": 3.682766190998902e-07, + "loss": 1.1303, + "mean_token_accuracy": 0.6742412447929382, + "num_tokens": 17333668.0, + "step": 672 + }, + { + "epoch": 0.07390731385899407, + "grad_norm": 2.5360729694366455, + "learning_rate": 3.688254665203073e-07, + "loss": 1.0077, + "mean_token_accuracy": 0.704404354095459, + "num_tokens": 17357406.0, + "step": 673 + }, + { + "epoch": 0.07401713156160773, + "grad_norm": 2.3463590145111084, + "learning_rate": 3.6937431394072443e-07, + "loss": 1.153, + "mean_token_accuracy": 0.6604936718940735, + "num_tokens": 17384340.0, + "step": 674 + }, + { + "epoch": 0.07412694926422139, + "grad_norm": 2.1853997707366943, + "learning_rate": 3.699231613611416e-07, + "loss": 1.1441, + "mean_token_accuracy": 0.6682497262954712, + "num_tokens": 17418975.0, + "step": 675 + }, + { + "epoch": 0.07423676696683505, + "grad_norm": 2.7751708030700684, + "learning_rate": 3.704720087815587e-07, + "loss": 1.0099, + "mean_token_accuracy": 0.7035788893699646, + "num_tokens": 17439679.0, + "step": 676 + }, + { + "epoch": 0.07434658466944871, + "grad_norm": 2.8604438304901123, + "learning_rate": 3.7102085620197583e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.659514844417572, + "num_tokens": 17460999.0, + "step": 677 + }, + { + "epoch": 0.07445640237206237, + "grad_norm": 3.19266939163208, + "learning_rate": 3.71569703622393e-07, + "loss": 1.0946, + "mean_token_accuracy": 0.676821768283844, + "num_tokens": 17481153.0, + "step": 678 + }, + { + "epoch": 0.07456622007467603, + "grad_norm": 2.4343223571777344, + "learning_rate": 3.721185510428101e-07, + "loss": 1.0774, + "mean_token_accuracy": 0.6769063472747803, + "num_tokens": 17506353.0, + "step": 679 + }, + { + "epoch": 0.0746760377772897, + "grad_norm": 2.7144153118133545, + "learning_rate": 3.726673984632272e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.702724277973175, + "num_tokens": 17528864.0, + "step": 680 + }, + { + "epoch": 0.07478585547990337, + "grad_norm": 2.492001533508301, + "learning_rate": 3.732162458836443e-07, + "loss": 1.1213, + "mean_token_accuracy": 0.6700570583343506, + "num_tokens": 17553819.0, + "step": 681 + }, + { + "epoch": 0.07489567318251703, + "grad_norm": 2.425929069519043, + "learning_rate": 3.7376509330406146e-07, + "loss": 1.1898, + "mean_token_accuracy": 0.6534858345985413, + "num_tokens": 17581561.0, + "step": 682 + }, + { + "epoch": 0.07500549088513069, + "grad_norm": 2.536684513092041, + "learning_rate": 3.743139407244786e-07, + "loss": 1.0778, + "mean_token_accuracy": 0.6814733743667603, + "num_tokens": 17606395.0, + "step": 683 + }, + { + "epoch": 0.07511530858774434, + "grad_norm": 2.427321434020996, + "learning_rate": 3.748627881448957e-07, + "loss": 1.1055, + "mean_token_accuracy": 0.670585572719574, + "num_tokens": 17633034.0, + "step": 684 + }, + { + "epoch": 0.075225126290358, + "grad_norm": 2.575789451599121, + "learning_rate": 3.754116355653128e-07, + "loss": 1.063, + "mean_token_accuracy": 0.6836634278297424, + "num_tokens": 17656972.0, + "step": 685 + }, + { + "epoch": 0.07533494399297166, + "grad_norm": 2.4511404037475586, + "learning_rate": 3.7596048298573e-07, + "loss": 1.2329, + "mean_token_accuracy": 0.6470005512237549, + "num_tokens": 17684081.0, + "step": 686 + }, + { + "epoch": 0.07544476169558532, + "grad_norm": 2.5932202339172363, + "learning_rate": 3.765093304061471e-07, + "loss": 1.1405, + "mean_token_accuracy": 0.6737632155418396, + "num_tokens": 17707942.0, + "step": 687 + }, + { + "epoch": 0.07555457939819898, + "grad_norm": 2.3022654056549072, + "learning_rate": 3.770581778265642e-07, + "loss": 1.0308, + "mean_token_accuracy": 0.6971641182899475, + "num_tokens": 17734237.0, + "step": 688 + }, + { + "epoch": 0.07566439710081264, + "grad_norm": 2.5114359855651855, + "learning_rate": 3.7760702524698137e-07, + "loss": 1.0334, + "mean_token_accuracy": 0.6918670535087585, + "num_tokens": 17758403.0, + "step": 689 + }, + { + "epoch": 0.07577421480342632, + "grad_norm": 2.227494239807129, + "learning_rate": 3.7815587266739844e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6828498244285583, + "num_tokens": 17787106.0, + "step": 690 + }, + { + "epoch": 0.07588403250603998, + "grad_norm": 2.454333543777466, + "learning_rate": 3.7870472008781555e-07, + "loss": 1.1268, + "mean_token_accuracy": 0.6707912683486938, + "num_tokens": 17811763.0, + "step": 691 + }, + { + "epoch": 0.07599385020865364, + "grad_norm": 2.517322301864624, + "learning_rate": 3.7925356750823267e-07, + "loss": 1.0675, + "mean_token_accuracy": 0.6819335222244263, + "num_tokens": 17834829.0, + "step": 692 + }, + { + "epoch": 0.0761036679112673, + "grad_norm": 2.081631898880005, + "learning_rate": 3.7980241492864983e-07, + "loss": 1.1966, + "mean_token_accuracy": 0.6536966562271118, + "num_tokens": 17871339.0, + "step": 693 + }, + { + "epoch": 0.07621348561388096, + "grad_norm": 2.369682550430298, + "learning_rate": 3.8035126234906695e-07, + "loss": 1.0825, + "mean_token_accuracy": 0.6853011846542358, + "num_tokens": 17901308.0, + "step": 694 + }, + { + "epoch": 0.07632330331649462, + "grad_norm": 3.0704689025878906, + "learning_rate": 3.8090010976948406e-07, + "loss": 0.9837, + "mean_token_accuracy": 0.7003518342971802, + "num_tokens": 17918975.0, + "step": 695 + }, + { + "epoch": 0.07643312101910828, + "grad_norm": 2.3771345615386963, + "learning_rate": 3.814489571899012e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.6747715473175049, + "num_tokens": 17945691.0, + "step": 696 + }, + { + "epoch": 0.07654293872172194, + "grad_norm": 2.561981201171875, + "learning_rate": 3.8199780461031835e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.6766238212585449, + "num_tokens": 17968948.0, + "step": 697 + }, + { + "epoch": 0.0766527564243356, + "grad_norm": 2.5573439598083496, + "learning_rate": 3.8254665203073546e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.6746814250946045, + "num_tokens": 17992830.0, + "step": 698 + }, + { + "epoch": 0.07676257412694927, + "grad_norm": 2.2108237743377686, + "learning_rate": 3.830954994511525e-07, + "loss": 1.168, + "mean_token_accuracy": 0.6617380976676941, + "num_tokens": 18023763.0, + "step": 699 + }, + { + "epoch": 0.07687239182956293, + "grad_norm": 2.761277198791504, + "learning_rate": 3.836443468715697e-07, + "loss": 1.1139, + "mean_token_accuracy": 0.6758972406387329, + "num_tokens": 18044852.0, + "step": 700 + }, + { + "epoch": 0.07698220953217659, + "grad_norm": 2.755082607269287, + "learning_rate": 3.841931942919868e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7066501379013062, + "num_tokens": 18063893.0, + "step": 701 + }, + { + "epoch": 0.07709202723479025, + "grad_norm": 3.0217907428741455, + "learning_rate": 3.847420417124039e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.6986186504364014, + "num_tokens": 18080834.0, + "step": 702 + }, + { + "epoch": 0.07720184493740391, + "grad_norm": 2.601775646209717, + "learning_rate": 3.8529088913282104e-07, + "loss": 0.9688, + "mean_token_accuracy": 0.7085288763046265, + "num_tokens": 18102241.0, + "step": 703 + }, + { + "epoch": 0.07731166264001757, + "grad_norm": 2.9648475646972656, + "learning_rate": 3.858397365532382e-07, + "loss": 0.9543, + "mean_token_accuracy": 0.7040352821350098, + "num_tokens": 18119766.0, + "step": 704 + }, + { + "epoch": 0.07742148034263123, + "grad_norm": 2.4023125171661377, + "learning_rate": 3.863885839736553e-07, + "loss": 1.1564, + "mean_token_accuracy": 0.6575770974159241, + "num_tokens": 18147156.0, + "step": 705 + }, + { + "epoch": 0.07753129804524489, + "grad_norm": 2.5329196453094482, + "learning_rate": 3.8693743139407244e-07, + "loss": 1.0744, + "mean_token_accuracy": 0.6871561408042908, + "num_tokens": 18171715.0, + "step": 706 + }, + { + "epoch": 0.07764111574785855, + "grad_norm": 2.258131742477417, + "learning_rate": 3.8748627881448955e-07, + "loss": 1.14, + "mean_token_accuracy": 0.6657870411872864, + "num_tokens": 18204077.0, + "step": 707 + }, + { + "epoch": 0.07775093345047221, + "grad_norm": 2.4869225025177, + "learning_rate": 3.880351262349067e-07, + "loss": 1.0244, + "mean_token_accuracy": 0.6956592798233032, + "num_tokens": 18227656.0, + "step": 708 + }, + { + "epoch": 0.07786075115308588, + "grad_norm": 2.568171262741089, + "learning_rate": 3.885839736553238e-07, + "loss": 1.0213, + "mean_token_accuracy": 0.6928159594535828, + "num_tokens": 18249416.0, + "step": 709 + }, + { + "epoch": 0.07797056885569954, + "grad_norm": 2.57285737991333, + "learning_rate": 3.891328210757409e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.6857011318206787, + "num_tokens": 18273717.0, + "step": 710 + }, + { + "epoch": 0.0780803865583132, + "grad_norm": 2.336723566055298, + "learning_rate": 3.8968166849615807e-07, + "loss": 1.1198, + "mean_token_accuracy": 0.6753630638122559, + "num_tokens": 18302745.0, + "step": 711 + }, + { + "epoch": 0.07819020426092686, + "grad_norm": 2.155951738357544, + "learning_rate": 3.902305159165752e-07, + "loss": 1.1344, + "mean_token_accuracy": 0.6758033037185669, + "num_tokens": 18334011.0, + "step": 712 + }, + { + "epoch": 0.07830002196354052, + "grad_norm": 2.61865496635437, + "learning_rate": 3.907793633369923e-07, + "loss": 1.0659, + "mean_token_accuracy": 0.6884974241256714, + "num_tokens": 18356457.0, + "step": 713 + }, + { + "epoch": 0.07840983966615418, + "grad_norm": 2.5096514225006104, + "learning_rate": 3.913282107574094e-07, + "loss": 1.0595, + "mean_token_accuracy": 0.6825203895568848, + "num_tokens": 18377482.0, + "step": 714 + }, + { + "epoch": 0.07851965736876784, + "grad_norm": 2.4933485984802246, + "learning_rate": 3.918770581778266e-07, + "loss": 1.0979, + "mean_token_accuracy": 0.6765769720077515, + "num_tokens": 18401783.0, + "step": 715 + }, + { + "epoch": 0.0786294750713815, + "grad_norm": 2.441286325454712, + "learning_rate": 3.924259055982437e-07, + "loss": 1.0456, + "mean_token_accuracy": 0.7137014865875244, + "num_tokens": 18428349.0, + "step": 716 + }, + { + "epoch": 0.07873929277399516, + "grad_norm": 2.5801737308502197, + "learning_rate": 3.929747530186608e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.6901133060455322, + "num_tokens": 18451181.0, + "step": 717 + }, + { + "epoch": 0.07884911047660884, + "grad_norm": 2.2876546382904053, + "learning_rate": 3.935236004390779e-07, + "loss": 1.0811, + "mean_token_accuracy": 0.6859476566314697, + "num_tokens": 18482471.0, + "step": 718 + }, + { + "epoch": 0.0789589281792225, + "grad_norm": 2.201453924179077, + "learning_rate": 3.9407244785949504e-07, + "loss": 1.1714, + "mean_token_accuracy": 0.6805949807167053, + "num_tokens": 18513254.0, + "step": 719 + }, + { + "epoch": 0.07906874588183616, + "grad_norm": 2.3504273891448975, + "learning_rate": 3.9462129527991216e-07, + "loss": 1.0499, + "mean_token_accuracy": 0.6860734224319458, + "num_tokens": 18541355.0, + "step": 720 + }, + { + "epoch": 0.07917856358444982, + "grad_norm": 2.3777973651885986, + "learning_rate": 3.9517014270032927e-07, + "loss": 1.0715, + "mean_token_accuracy": 0.6821449398994446, + "num_tokens": 18568839.0, + "step": 721 + }, + { + "epoch": 0.07928838128706348, + "grad_norm": 2.686328172683716, + "learning_rate": 3.957189901207464e-07, + "loss": 1.127, + "mean_token_accuracy": 0.6680614948272705, + "num_tokens": 18591194.0, + "step": 722 + }, + { + "epoch": 0.07939819898967714, + "grad_norm": 2.0607805252075195, + "learning_rate": 3.9626783754116356e-07, + "loss": 1.0884, + "mean_token_accuracy": 0.6778640747070312, + "num_tokens": 18622182.0, + "step": 723 + }, + { + "epoch": 0.0795080166922908, + "grad_norm": 2.400017738342285, + "learning_rate": 3.9681668496158067e-07, + "loss": 1.0642, + "mean_token_accuracy": 0.6819779872894287, + "num_tokens": 18647306.0, + "step": 724 + }, + { + "epoch": 0.07961783439490445, + "grad_norm": 2.2714498043060303, + "learning_rate": 3.973655323819978e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.7000807523727417, + "num_tokens": 18676145.0, + "step": 725 + }, + { + "epoch": 0.07972765209751811, + "grad_norm": 2.370857000350952, + "learning_rate": 3.9791437980241495e-07, + "loss": 1.0123, + "mean_token_accuracy": 0.6951757073402405, + "num_tokens": 18700850.0, + "step": 726 + }, + { + "epoch": 0.07983746980013177, + "grad_norm": 2.2020773887634277, + "learning_rate": 3.9846322722283207e-07, + "loss": 1.0517, + "mean_token_accuracy": 0.6903795599937439, + "num_tokens": 18733368.0, + "step": 727 + }, + { + "epoch": 0.07994728750274545, + "grad_norm": 2.347209930419922, + "learning_rate": 3.9901207464324913e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.6977893710136414, + "num_tokens": 18760956.0, + "step": 728 + }, + { + "epoch": 0.08005710520535911, + "grad_norm": 2.6287310123443604, + "learning_rate": 3.9956092206366625e-07, + "loss": 1.1191, + "mean_token_accuracy": 0.6716377139091492, + "num_tokens": 18786532.0, + "step": 729 + }, + { + "epoch": 0.08016692290797277, + "grad_norm": 3.1463539600372314, + "learning_rate": 4.001097694840834e-07, + "loss": 1.1584, + "mean_token_accuracy": 0.6836303472518921, + "num_tokens": 18804488.0, + "step": 730 + }, + { + "epoch": 0.08027674061058643, + "grad_norm": 2.7106821537017822, + "learning_rate": 4.0065861690450053e-07, + "loss": 1.0818, + "mean_token_accuracy": 0.6823344230651855, + "num_tokens": 18831152.0, + "step": 731 + }, + { + "epoch": 0.08038655831320009, + "grad_norm": 2.2730324268341064, + "learning_rate": 4.0120746432491765e-07, + "loss": 1.2245, + "mean_token_accuracy": 0.6477319598197937, + "num_tokens": 18860934.0, + "step": 732 + }, + { + "epoch": 0.08049637601581375, + "grad_norm": 2.369501829147339, + "learning_rate": 4.0175631174533476e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.6822363138198853, + "num_tokens": 18890267.0, + "step": 733 + }, + { + "epoch": 0.08060619371842741, + "grad_norm": 2.284292697906494, + "learning_rate": 4.0230515916575193e-07, + "loss": 1.1412, + "mean_token_accuracy": 0.6727917194366455, + "num_tokens": 18916692.0, + "step": 734 + }, + { + "epoch": 0.08071601142104107, + "grad_norm": 2.082228422164917, + "learning_rate": 4.0285400658616904e-07, + "loss": 1.1054, + "mean_token_accuracy": 0.686191737651825, + "num_tokens": 18950377.0, + "step": 735 + }, + { + "epoch": 0.08082582912365473, + "grad_norm": 2.645469903945923, + "learning_rate": 4.0340285400658616e-07, + "loss": 1.0782, + "mean_token_accuracy": 0.6822223663330078, + "num_tokens": 18972745.0, + "step": 736 + }, + { + "epoch": 0.0809356468262684, + "grad_norm": 2.594085931777954, + "learning_rate": 4.039517014270033e-07, + "loss": 0.9743, + "mean_token_accuracy": 0.7112522721290588, + "num_tokens": 18994416.0, + "step": 737 + }, + { + "epoch": 0.08104546452888206, + "grad_norm": 2.5858383178710938, + "learning_rate": 4.045005488474204e-07, + "loss": 1.1116, + "mean_token_accuracy": 0.6776620745658875, + "num_tokens": 19016886.0, + "step": 738 + }, + { + "epoch": 0.08115528223149572, + "grad_norm": 2.4142205715179443, + "learning_rate": 4.050493962678375e-07, + "loss": 1.0758, + "mean_token_accuracy": 0.678093671798706, + "num_tokens": 19042082.0, + "step": 739 + }, + { + "epoch": 0.08126509993410938, + "grad_norm": 2.359337329864502, + "learning_rate": 4.055982436882546e-07, + "loss": 0.9701, + "mean_token_accuracy": 0.7133249044418335, + "num_tokens": 19066251.0, + "step": 740 + }, + { + "epoch": 0.08137491763672304, + "grad_norm": 2.213360548019409, + "learning_rate": 4.061470911086718e-07, + "loss": 1.0108, + "mean_token_accuracy": 0.6985015273094177, + "num_tokens": 19093399.0, + "step": 741 + }, + { + "epoch": 0.0814847353393367, + "grad_norm": 2.4570741653442383, + "learning_rate": 4.066959385290889e-07, + "loss": 1.1194, + "mean_token_accuracy": 0.6711503267288208, + "num_tokens": 19118046.0, + "step": 742 + }, + { + "epoch": 0.08159455304195036, + "grad_norm": 2.2048230171203613, + "learning_rate": 4.07244785949506e-07, + "loss": 1.1235, + "mean_token_accuracy": 0.6746467351913452, + "num_tokens": 19147047.0, + "step": 743 + }, + { + "epoch": 0.08170437074456402, + "grad_norm": 2.2280430793762207, + "learning_rate": 4.0779363336992313e-07, + "loss": 1.0752, + "mean_token_accuracy": 0.6812273859977722, + "num_tokens": 19174554.0, + "step": 744 + }, + { + "epoch": 0.08181418844717768, + "grad_norm": 2.293663501739502, + "learning_rate": 4.083424807903403e-07, + "loss": 1.1115, + "mean_token_accuracy": 0.6728384494781494, + "num_tokens": 19203079.0, + "step": 745 + }, + { + "epoch": 0.08192400614979134, + "grad_norm": 2.2394847869873047, + "learning_rate": 4.0889132821075737e-07, + "loss": 0.9818, + "mean_token_accuracy": 0.710110604763031, + "num_tokens": 19232722.0, + "step": 746 + }, + { + "epoch": 0.08203382385240501, + "grad_norm": 2.101539373397827, + "learning_rate": 4.094401756311745e-07, + "loss": 1.0121, + "mean_token_accuracy": 0.6980043649673462, + "num_tokens": 19263131.0, + "step": 747 + }, + { + "epoch": 0.08214364155501867, + "grad_norm": 2.240805149078369, + "learning_rate": 4.0998902305159165e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.68337082862854, + "num_tokens": 19291570.0, + "step": 748 + }, + { + "epoch": 0.08225345925763233, + "grad_norm": 2.3410887718200684, + "learning_rate": 4.1053787047200876e-07, + "loss": 1.2, + "mean_token_accuracy": 0.6524137854576111, + "num_tokens": 19320151.0, + "step": 749 + }, + { + "epoch": 0.08236327696024599, + "grad_norm": 2.4128782749176025, + "learning_rate": 4.110867178924259e-07, + "loss": 1.1384, + "mean_token_accuracy": 0.6657573580741882, + "num_tokens": 19345286.0, + "step": 750 + }, + { + "epoch": 0.08247309466285965, + "grad_norm": 2.153822660446167, + "learning_rate": 4.11635565312843e-07, + "loss": 1.053, + "mean_token_accuracy": 0.6897947192192078, + "num_tokens": 19376637.0, + "step": 751 + }, + { + "epoch": 0.08258291236547331, + "grad_norm": 2.0679566860198975, + "learning_rate": 4.1218441273326016e-07, + "loss": 1.1567, + "mean_token_accuracy": 0.6665802597999573, + "num_tokens": 19408647.0, + "step": 752 + }, + { + "epoch": 0.08269273006808697, + "grad_norm": 2.315704345703125, + "learning_rate": 4.127332601536773e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.7040307521820068, + "num_tokens": 19434332.0, + "step": 753 + }, + { + "epoch": 0.08280254777070063, + "grad_norm": 2.4247398376464844, + "learning_rate": 4.132821075740944e-07, + "loss": 1.126, + "mean_token_accuracy": 0.6786632537841797, + "num_tokens": 19461743.0, + "step": 754 + }, + { + "epoch": 0.08291236547331429, + "grad_norm": 2.557861328125, + "learning_rate": 4.138309549945115e-07, + "loss": 1.0842, + "mean_token_accuracy": 0.6837880611419678, + "num_tokens": 19485645.0, + "step": 755 + }, + { + "epoch": 0.08302218317592797, + "grad_norm": 2.2087833881378174, + "learning_rate": 4.143798024149286e-07, + "loss": 1.1252, + "mean_token_accuracy": 0.6746777296066284, + "num_tokens": 19514092.0, + "step": 756 + }, + { + "epoch": 0.08313200087854163, + "grad_norm": 2.662203311920166, + "learning_rate": 4.1492864983534574e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.672903299331665, + "num_tokens": 19535173.0, + "step": 757 + }, + { + "epoch": 0.08324181858115529, + "grad_norm": 2.615247964859009, + "learning_rate": 4.1547749725576285e-07, + "loss": 1.028, + "mean_token_accuracy": 0.6959903240203857, + "num_tokens": 19555749.0, + "step": 758 + }, + { + "epoch": 0.08335163628376895, + "grad_norm": 2.2556421756744385, + "learning_rate": 4.1602634467618e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.6870807409286499, + "num_tokens": 19582018.0, + "step": 759 + }, + { + "epoch": 0.0834614539863826, + "grad_norm": 2.4417293071746826, + "learning_rate": 4.1657519209659714e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.7051514983177185, + "num_tokens": 19605844.0, + "step": 760 + }, + { + "epoch": 0.08357127168899627, + "grad_norm": 2.498873472213745, + "learning_rate": 4.1712403951701425e-07, + "loss": 1.0082, + "mean_token_accuracy": 0.7012791633605957, + "num_tokens": 19629462.0, + "step": 761 + }, + { + "epoch": 0.08368108939160993, + "grad_norm": 2.1984574794769287, + "learning_rate": 4.1767288693743137e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.6897039413452148, + "num_tokens": 19660696.0, + "step": 762 + }, + { + "epoch": 0.08379090709422359, + "grad_norm": 2.508716583251953, + "learning_rate": 4.1822173435784854e-07, + "loss": 1.091, + "mean_token_accuracy": 0.6872658729553223, + "num_tokens": 19686576.0, + "step": 763 + }, + { + "epoch": 0.08390072479683724, + "grad_norm": 1.9087271690368652, + "learning_rate": 4.1877058177826565e-07, + "loss": 1.0825, + "mean_token_accuracy": 0.6777503490447998, + "num_tokens": 19722185.0, + "step": 764 + }, + { + "epoch": 0.0840105424994509, + "grad_norm": 2.0783560276031494, + "learning_rate": 4.193194291986827e-07, + "loss": 1.1065, + "mean_token_accuracy": 0.6764228940010071, + "num_tokens": 19755008.0, + "step": 765 + }, + { + "epoch": 0.08412036020206458, + "grad_norm": 2.3965742588043213, + "learning_rate": 4.1986827661909983e-07, + "loss": 1.0327, + "mean_token_accuracy": 0.6923460960388184, + "num_tokens": 19781846.0, + "step": 766 + }, + { + "epoch": 0.08423017790467824, + "grad_norm": 2.295806646347046, + "learning_rate": 4.20417124039517e-07, + "loss": 1.1316, + "mean_token_accuracy": 0.6683974862098694, + "num_tokens": 19809331.0, + "step": 767 + }, + { + "epoch": 0.0843399956072919, + "grad_norm": 2.2168028354644775, + "learning_rate": 4.209659714599341e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6681755781173706, + "num_tokens": 19839000.0, + "step": 768 + }, + { + "epoch": 0.08444981330990556, + "grad_norm": 2.2559800148010254, + "learning_rate": 4.2151481888035123e-07, + "loss": 1.0921, + "mean_token_accuracy": 0.6794140338897705, + "num_tokens": 19867335.0, + "step": 769 + }, + { + "epoch": 0.08455963101251922, + "grad_norm": 2.4286584854125977, + "learning_rate": 4.2206366630076834e-07, + "loss": 1.2007, + "mean_token_accuracy": 0.6554266810417175, + "num_tokens": 19895042.0, + "step": 770 + }, + { + "epoch": 0.08466944871513288, + "grad_norm": 2.2556710243225098, + "learning_rate": 4.226125137211855e-07, + "loss": 1.1639, + "mean_token_accuracy": 0.6703728437423706, + "num_tokens": 19923993.0, + "step": 771 + }, + { + "epoch": 0.08477926641774654, + "grad_norm": 2.983762741088867, + "learning_rate": 4.231613611416026e-07, + "loss": 1.0191, + "mean_token_accuracy": 0.6936074495315552, + "num_tokens": 19942200.0, + "step": 772 + }, + { + "epoch": 0.0848890841203602, + "grad_norm": 2.121439218521118, + "learning_rate": 4.2371020856201974e-07, + "loss": 1.0647, + "mean_token_accuracy": 0.6863154172897339, + "num_tokens": 19975053.0, + "step": 773 + }, + { + "epoch": 0.08499890182297386, + "grad_norm": 2.6010184288024902, + "learning_rate": 4.242590559824369e-07, + "loss": 1.1975, + "mean_token_accuracy": 0.6568882465362549, + "num_tokens": 20001864.0, + "step": 774 + }, + { + "epoch": 0.08510871952558753, + "grad_norm": 2.2142698764801025, + "learning_rate": 4.2480790340285397e-07, + "loss": 1.1971, + "mean_token_accuracy": 0.6590489149093628, + "num_tokens": 20029829.0, + "step": 775 + }, + { + "epoch": 0.08521853722820119, + "grad_norm": 2.548884868621826, + "learning_rate": 4.253567508232711e-07, + "loss": 1.1449, + "mean_token_accuracy": 0.6685471534729004, + "num_tokens": 20053033.0, + "step": 776 + }, + { + "epoch": 0.08532835493081485, + "grad_norm": 2.1688339710235596, + "learning_rate": 4.259055982436882e-07, + "loss": 1.1713, + "mean_token_accuracy": 0.6637248992919922, + "num_tokens": 20083027.0, + "step": 777 + }, + { + "epoch": 0.08543817263342851, + "grad_norm": 2.3440957069396973, + "learning_rate": 4.2645444566410537e-07, + "loss": 1.1092, + "mean_token_accuracy": 0.6804519891738892, + "num_tokens": 20108259.0, + "step": 778 + }, + { + "epoch": 0.08554799033604217, + "grad_norm": 2.1152493953704834, + "learning_rate": 4.270032930845225e-07, + "loss": 1.1272, + "mean_token_accuracy": 0.6697934865951538, + "num_tokens": 20140248.0, + "step": 779 + }, + { + "epoch": 0.08565780803865583, + "grad_norm": 2.456007957458496, + "learning_rate": 4.275521405049396e-07, + "loss": 1.1363, + "mean_token_accuracy": 0.669685959815979, + "num_tokens": 20168636.0, + "step": 780 + }, + { + "epoch": 0.08576762574126949, + "grad_norm": 2.531269073486328, + "learning_rate": 4.281009879253567e-07, + "loss": 1.0138, + "mean_token_accuracy": 0.6984386444091797, + "num_tokens": 20192375.0, + "step": 781 + }, + { + "epoch": 0.08587744344388315, + "grad_norm": 2.310779333114624, + "learning_rate": 4.286498353457739e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6734538674354553, + "num_tokens": 20218275.0, + "step": 782 + }, + { + "epoch": 0.08598726114649681, + "grad_norm": 2.3658828735351562, + "learning_rate": 4.29198682766191e-07, + "loss": 1.0624, + "mean_token_accuracy": 0.6850553154945374, + "num_tokens": 20245100.0, + "step": 783 + }, + { + "epoch": 0.08609707884911047, + "grad_norm": 2.205777168273926, + "learning_rate": 4.2974753018660806e-07, + "loss": 1.1456, + "mean_token_accuracy": 0.6747523546218872, + "num_tokens": 20273207.0, + "step": 784 + }, + { + "epoch": 0.08620689655172414, + "grad_norm": 2.6469364166259766, + "learning_rate": 4.3029637760702523e-07, + "loss": 1.1833, + "mean_token_accuracy": 0.6616923809051514, + "num_tokens": 20298893.0, + "step": 785 + }, + { + "epoch": 0.0863167142543378, + "grad_norm": 2.748594284057617, + "learning_rate": 4.3084522502744235e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.690485954284668, + "num_tokens": 20320844.0, + "step": 786 + }, + { + "epoch": 0.08642653195695146, + "grad_norm": 2.552922248840332, + "learning_rate": 4.3139407244785946e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6893734931945801, + "num_tokens": 20343718.0, + "step": 787 + }, + { + "epoch": 0.08653634965956512, + "grad_norm": 2.5027573108673096, + "learning_rate": 4.319429198682766e-07, + "loss": 1.1502, + "mean_token_accuracy": 0.6603432893753052, + "num_tokens": 20367787.0, + "step": 788 + }, + { + "epoch": 0.08664616736217878, + "grad_norm": 2.546360969543457, + "learning_rate": 4.3249176728869374e-07, + "loss": 1.1319, + "mean_token_accuracy": 0.6647267937660217, + "num_tokens": 20394267.0, + "step": 789 + }, + { + "epoch": 0.08675598506479244, + "grad_norm": 2.331911563873291, + "learning_rate": 4.3304061470911086e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6883370280265808, + "num_tokens": 20420321.0, + "step": 790 + }, + { + "epoch": 0.0868658027674061, + "grad_norm": 2.4064207077026367, + "learning_rate": 4.33589462129528e-07, + "loss": 1.1031, + "mean_token_accuracy": 0.673305869102478, + "num_tokens": 20446390.0, + "step": 791 + }, + { + "epoch": 0.08697562047001976, + "grad_norm": 2.381823778152466, + "learning_rate": 4.341383095499451e-07, + "loss": 1.0828, + "mean_token_accuracy": 0.6799704432487488, + "num_tokens": 20471468.0, + "step": 792 + }, + { + "epoch": 0.08708543817263342, + "grad_norm": 2.620042085647583, + "learning_rate": 4.3468715697036226e-07, + "loss": 1.0531, + "mean_token_accuracy": 0.691545844078064, + "num_tokens": 20494430.0, + "step": 793 + }, + { + "epoch": 0.0871952558752471, + "grad_norm": 2.162052869796753, + "learning_rate": 4.352360043907793e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.6899824142456055, + "num_tokens": 20523929.0, + "step": 794 + }, + { + "epoch": 0.08730507357786076, + "grad_norm": 2.093756675720215, + "learning_rate": 4.3578485181119644e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.6679117679595947, + "num_tokens": 20556458.0, + "step": 795 + }, + { + "epoch": 0.08741489128047442, + "grad_norm": 2.465190887451172, + "learning_rate": 4.363336992316136e-07, + "loss": 1.0621, + "mean_token_accuracy": 0.68617844581604, + "num_tokens": 20579651.0, + "step": 796 + }, + { + "epoch": 0.08752470898308808, + "grad_norm": 2.109280586242676, + "learning_rate": 4.368825466520307e-07, + "loss": 1.1227, + "mean_token_accuracy": 0.6709742546081543, + "num_tokens": 20611256.0, + "step": 797 + }, + { + "epoch": 0.08763452668570174, + "grad_norm": 2.6587963104248047, + "learning_rate": 4.3743139407244783e-07, + "loss": 1.1054, + "mean_token_accuracy": 0.6717488169670105, + "num_tokens": 20634181.0, + "step": 798 + }, + { + "epoch": 0.0877443443883154, + "grad_norm": 2.445899486541748, + "learning_rate": 4.3798024149286495e-07, + "loss": 1.1413, + "mean_token_accuracy": 0.665283203125, + "num_tokens": 20658845.0, + "step": 799 + }, + { + "epoch": 0.08785416209092906, + "grad_norm": 2.7228569984436035, + "learning_rate": 4.385290889132821e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6864800453186035, + "num_tokens": 20681272.0, + "step": 800 + }, + { + "epoch": 0.08796397979354272, + "grad_norm": 2.0703375339508057, + "learning_rate": 4.3907793633369923e-07, + "loss": 1.0103, + "mean_token_accuracy": 0.695285975933075, + "num_tokens": 20712006.0, + "step": 801 + }, + { + "epoch": 0.08807379749615638, + "grad_norm": 2.442122459411621, + "learning_rate": 4.3962678375411635e-07, + "loss": 1.0257, + "mean_token_accuracy": 0.6939612627029419, + "num_tokens": 20737200.0, + "step": 802 + }, + { + "epoch": 0.08818361519877004, + "grad_norm": 2.4671850204467773, + "learning_rate": 4.401756311745334e-07, + "loss": 1.0933, + "mean_token_accuracy": 0.6706172823905945, + "num_tokens": 20762135.0, + "step": 803 + }, + { + "epoch": 0.08829343290138371, + "grad_norm": 2.420522689819336, + "learning_rate": 4.407244785949506e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.6905561685562134, + "num_tokens": 20787431.0, + "step": 804 + }, + { + "epoch": 0.08840325060399737, + "grad_norm": 2.7420458793640137, + "learning_rate": 4.412733260153677e-07, + "loss": 1.1378, + "mean_token_accuracy": 0.6692282557487488, + "num_tokens": 20810968.0, + "step": 805 + }, + { + "epoch": 0.08851306830661103, + "grad_norm": 2.4892032146453857, + "learning_rate": 4.418221734357848e-07, + "loss": 1.1405, + "mean_token_accuracy": 0.6650395393371582, + "num_tokens": 20836125.0, + "step": 806 + }, + { + "epoch": 0.08862288600922469, + "grad_norm": 2.4341213703155518, + "learning_rate": 4.42371020856202e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.6798731088638306, + "num_tokens": 20860587.0, + "step": 807 + }, + { + "epoch": 0.08873270371183835, + "grad_norm": 2.300964593887329, + "learning_rate": 4.429198682766191e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6740900278091431, + "num_tokens": 20887889.0, + "step": 808 + }, + { + "epoch": 0.08884252141445201, + "grad_norm": 2.4227664470672607, + "learning_rate": 4.434687156970362e-07, + "loss": 0.9548, + "mean_token_accuracy": 0.7158953547477722, + "num_tokens": 20913729.0, + "step": 809 + }, + { + "epoch": 0.08895233911706567, + "grad_norm": 2.3398444652557373, + "learning_rate": 4.440175631174533e-07, + "loss": 1.1306, + "mean_token_accuracy": 0.6697219610214233, + "num_tokens": 20940492.0, + "step": 810 + }, + { + "epoch": 0.08906215681967933, + "grad_norm": 2.4528391361236572, + "learning_rate": 4.445664105378705e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6689151525497437, + "num_tokens": 20966678.0, + "step": 811 + }, + { + "epoch": 0.08917197452229299, + "grad_norm": 2.374239921569824, + "learning_rate": 4.451152579582876e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.7043122053146362, + "num_tokens": 20993300.0, + "step": 812 + }, + { + "epoch": 0.08928179222490666, + "grad_norm": 2.43817138671875, + "learning_rate": 4.4566410537870467e-07, + "loss": 1.0158, + "mean_token_accuracy": 0.6991567611694336, + "num_tokens": 21015775.0, + "step": 813 + }, + { + "epoch": 0.08939160992752032, + "grad_norm": 2.615375518798828, + "learning_rate": 4.462129527991218e-07, + "loss": 1.1453, + "mean_token_accuracy": 0.6741111874580383, + "num_tokens": 21037446.0, + "step": 814 + }, + { + "epoch": 0.08950142763013398, + "grad_norm": 2.4114413261413574, + "learning_rate": 4.4676180021953895e-07, + "loss": 1.1458, + "mean_token_accuracy": 0.6644673347473145, + "num_tokens": 21063675.0, + "step": 815 + }, + { + "epoch": 0.08961124533274764, + "grad_norm": 2.586094379425049, + "learning_rate": 4.4731064763995607e-07, + "loss": 0.9606, + "mean_token_accuracy": 0.7072862386703491, + "num_tokens": 21085092.0, + "step": 816 + }, + { + "epoch": 0.0897210630353613, + "grad_norm": 2.5906100273132324, + "learning_rate": 4.478594950603732e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.6970592737197876, + "num_tokens": 21105850.0, + "step": 817 + }, + { + "epoch": 0.08983088073797496, + "grad_norm": 2.267124891281128, + "learning_rate": 4.4840834248079035e-07, + "loss": 1.0958, + "mean_token_accuracy": 0.6850523948669434, + "num_tokens": 21133806.0, + "step": 818 + }, + { + "epoch": 0.08994069844058862, + "grad_norm": 2.493896007537842, + "learning_rate": 4.4895718990120747e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6692390441894531, + "num_tokens": 21157968.0, + "step": 819 + }, + { + "epoch": 0.09005051614320228, + "grad_norm": 2.1538209915161133, + "learning_rate": 4.495060373216246e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.6744027137756348, + "num_tokens": 21187423.0, + "step": 820 + }, + { + "epoch": 0.09016033384581594, + "grad_norm": 2.480609893798828, + "learning_rate": 4.500548847420417e-07, + "loss": 1.151, + "mean_token_accuracy": 0.6681296825408936, + "num_tokens": 21213076.0, + "step": 821 + }, + { + "epoch": 0.0902701515484296, + "grad_norm": 2.4534637928009033, + "learning_rate": 4.5060373216245886e-07, + "loss": 1.0114, + "mean_token_accuracy": 0.6969691514968872, + "num_tokens": 21235265.0, + "step": 822 + }, + { + "epoch": 0.09037996925104327, + "grad_norm": 2.2564456462860107, + "learning_rate": 4.5115257958287593e-07, + "loss": 1.1274, + "mean_token_accuracy": 0.6724420785903931, + "num_tokens": 21264329.0, + "step": 823 + }, + { + "epoch": 0.09048978695365693, + "grad_norm": 2.441992998123169, + "learning_rate": 4.5170142700329304e-07, + "loss": 1.0804, + "mean_token_accuracy": 0.6814984679222107, + "num_tokens": 21289557.0, + "step": 824 + }, + { + "epoch": 0.0905996046562706, + "grad_norm": 2.323160409927368, + "learning_rate": 4.5225027442371016e-07, + "loss": 1.1594, + "mean_token_accuracy": 0.664247989654541, + "num_tokens": 21317393.0, + "step": 825 + }, + { + "epoch": 0.09070942235888425, + "grad_norm": 2.2767422199249268, + "learning_rate": 4.527991218441273e-07, + "loss": 1.0973, + "mean_token_accuracy": 0.6755704879760742, + "num_tokens": 21345238.0, + "step": 826 + }, + { + "epoch": 0.09081924006149791, + "grad_norm": 2.293745517730713, + "learning_rate": 4.5334796926454444e-07, + "loss": 1.1074, + "mean_token_accuracy": 0.6734867095947266, + "num_tokens": 21371225.0, + "step": 827 + }, + { + "epoch": 0.09092905776411157, + "grad_norm": 2.074744462966919, + "learning_rate": 4.5389681668496156e-07, + "loss": 1.1922, + "mean_token_accuracy": 0.6602569818496704, + "num_tokens": 21403818.0, + "step": 828 + }, + { + "epoch": 0.09103887546672523, + "grad_norm": 2.587895631790161, + "learning_rate": 4.5444566410537867e-07, + "loss": 1.0818, + "mean_token_accuracy": 0.6783239841461182, + "num_tokens": 21426445.0, + "step": 829 + }, + { + "epoch": 0.09114869316933889, + "grad_norm": 2.2945828437805176, + "learning_rate": 4.5499451152579584e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.6714868545532227, + "num_tokens": 21451893.0, + "step": 830 + }, + { + "epoch": 0.09125851087195255, + "grad_norm": 2.3328254222869873, + "learning_rate": 4.5554335894621295e-07, + "loss": 1.0736, + "mean_token_accuracy": 0.6768536567687988, + "num_tokens": 21477961.0, + "step": 831 + }, + { + "epoch": 0.09136832857456623, + "grad_norm": 2.152310371398926, + "learning_rate": 4.5609220636663e-07, + "loss": 1.1376, + "mean_token_accuracy": 0.6676874160766602, + "num_tokens": 21508356.0, + "step": 832 + }, + { + "epoch": 0.09147814627717989, + "grad_norm": 2.6916701793670654, + "learning_rate": 4.566410537870472e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6842162609100342, + "num_tokens": 21528952.0, + "step": 833 + }, + { + "epoch": 0.09158796397979355, + "grad_norm": 2.537349224090576, + "learning_rate": 4.571899012074643e-07, + "loss": 1.0215, + "mean_token_accuracy": 0.6929155588150024, + "num_tokens": 21549910.0, + "step": 834 + }, + { + "epoch": 0.0916977816824072, + "grad_norm": 2.306070327758789, + "learning_rate": 4.577387486278814e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.6874949932098389, + "num_tokens": 21576602.0, + "step": 835 + }, + { + "epoch": 0.09180759938502087, + "grad_norm": 2.312488555908203, + "learning_rate": 4.5828759604829853e-07, + "loss": 1.0113, + "mean_token_accuracy": 0.7020975351333618, + "num_tokens": 21600769.0, + "step": 836 + }, + { + "epoch": 0.09191741708763453, + "grad_norm": 2.6314313411712646, + "learning_rate": 4.588364434687157e-07, + "loss": 1.0404, + "mean_token_accuracy": 0.6917909979820251, + "num_tokens": 21621350.0, + "step": 837 + }, + { + "epoch": 0.09202723479024819, + "grad_norm": 2.619922399520874, + "learning_rate": 4.593852908891328e-07, + "loss": 1.1252, + "mean_token_accuracy": 0.6677731275558472, + "num_tokens": 21642615.0, + "step": 838 + }, + { + "epoch": 0.09213705249286185, + "grad_norm": 2.502244472503662, + "learning_rate": 4.5993413830954993e-07, + "loss": 1.033, + "mean_token_accuracy": 0.699161171913147, + "num_tokens": 21666180.0, + "step": 839 + }, + { + "epoch": 0.0922468701954755, + "grad_norm": 2.1840662956237793, + "learning_rate": 4.6048298572996704e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6756247282028198, + "num_tokens": 21696633.0, + "step": 840 + }, + { + "epoch": 0.09235668789808917, + "grad_norm": 3.035963296890259, + "learning_rate": 4.6103183315038416e-07, + "loss": 0.9818, + "mean_token_accuracy": 0.7125397324562073, + "num_tokens": 21715499.0, + "step": 841 + }, + { + "epoch": 0.09246650560070284, + "grad_norm": 2.3471615314483643, + "learning_rate": 4.615806805708013e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6824620962142944, + "num_tokens": 21741874.0, + "step": 842 + }, + { + "epoch": 0.0925763233033165, + "grad_norm": 2.0660436153411865, + "learning_rate": 4.621295279912184e-07, + "loss": 1.1749, + "mean_token_accuracy": 0.6609817743301392, + "num_tokens": 21773494.0, + "step": 843 + }, + { + "epoch": 0.09268614100593016, + "grad_norm": 2.401029109954834, + "learning_rate": 4.6267837541163556e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6942924857139587, + "num_tokens": 21799517.0, + "step": 844 + }, + { + "epoch": 0.09279595870854382, + "grad_norm": 2.5903422832489014, + "learning_rate": 4.632272228320527e-07, + "loss": 1.0262, + "mean_token_accuracy": 0.6971033215522766, + "num_tokens": 21821273.0, + "step": 845 + }, + { + "epoch": 0.09290577641115748, + "grad_norm": 2.202476978302002, + "learning_rate": 4.637760702524698e-07, + "loss": 1.0363, + "mean_token_accuracy": 0.6978000402450562, + "num_tokens": 21849975.0, + "step": 846 + }, + { + "epoch": 0.09301559411377114, + "grad_norm": 2.63012957572937, + "learning_rate": 4.643249176728869e-07, + "loss": 1.0803, + "mean_token_accuracy": 0.677477240562439, + "num_tokens": 21872824.0, + "step": 847 + }, + { + "epoch": 0.0931254118163848, + "grad_norm": 2.5352389812469482, + "learning_rate": 4.6487376509330407e-07, + "loss": 1.1529, + "mean_token_accuracy": 0.6587963104248047, + "num_tokens": 21897793.0, + "step": 848 + }, + { + "epoch": 0.09323522951899846, + "grad_norm": 2.298694610595703, + "learning_rate": 4.654226125137212e-07, + "loss": 1.0089, + "mean_token_accuracy": 0.6988579630851746, + "num_tokens": 21922246.0, + "step": 849 + }, + { + "epoch": 0.09334504722161212, + "grad_norm": 2.8420090675354004, + "learning_rate": 4.659714599341383e-07, + "loss": 1.0171, + "mean_token_accuracy": 0.6885315179824829, + "num_tokens": 21940166.0, + "step": 850 + }, + { + "epoch": 0.09345486492422579, + "grad_norm": 2.227581024169922, + "learning_rate": 4.6652030735455537e-07, + "loss": 1.0761, + "mean_token_accuracy": 0.6812645196914673, + "num_tokens": 21968365.0, + "step": 851 + }, + { + "epoch": 0.09356468262683945, + "grad_norm": 2.2148852348327637, + "learning_rate": 4.6706915477497253e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6949394941329956, + "num_tokens": 21999365.0, + "step": 852 + }, + { + "epoch": 0.09367450032945311, + "grad_norm": 2.3307297229766846, + "learning_rate": 4.6761800219538965e-07, + "loss": 1.0953, + "mean_token_accuracy": 0.6732461452484131, + "num_tokens": 22025309.0, + "step": 853 + }, + { + "epoch": 0.09378431803206677, + "grad_norm": 2.5071756839752197, + "learning_rate": 4.6816684961580676e-07, + "loss": 1.0776, + "mean_token_accuracy": 0.6846917271614075, + "num_tokens": 22047848.0, + "step": 854 + }, + { + "epoch": 0.09389413573468043, + "grad_norm": 2.430471181869507, + "learning_rate": 4.6871569703622393e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6938909292221069, + "num_tokens": 22070662.0, + "step": 855 + }, + { + "epoch": 0.09400395343729409, + "grad_norm": 2.3289504051208496, + "learning_rate": 4.6926454445664105e-07, + "loss": 1.0814, + "mean_token_accuracy": 0.6793934106826782, + "num_tokens": 22095763.0, + "step": 856 + }, + { + "epoch": 0.09411377113990775, + "grad_norm": 2.2405576705932617, + "learning_rate": 4.6981339187705816e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6900722980499268, + "num_tokens": 22121997.0, + "step": 857 + }, + { + "epoch": 0.09422358884252141, + "grad_norm": 2.2799901962280273, + "learning_rate": 4.703622392974753e-07, + "loss": 1.0177, + "mean_token_accuracy": 0.696343183517456, + "num_tokens": 22148051.0, + "step": 858 + }, + { + "epoch": 0.09433340654513507, + "grad_norm": 2.347160816192627, + "learning_rate": 4.7091108671789245e-07, + "loss": 1.0502, + "mean_token_accuracy": 0.6885168552398682, + "num_tokens": 22172281.0, + "step": 859 + }, + { + "epoch": 0.09444322424774873, + "grad_norm": 2.215564489364624, + "learning_rate": 4.714599341383095e-07, + "loss": 1.0996, + "mean_token_accuracy": 0.675923764705658, + "num_tokens": 22199733.0, + "step": 860 + }, + { + "epoch": 0.0945530419503624, + "grad_norm": 2.410951614379883, + "learning_rate": 4.720087815587266e-07, + "loss": 1.0911, + "mean_token_accuracy": 0.6747097969055176, + "num_tokens": 22223089.0, + "step": 861 + }, + { + "epoch": 0.09466285965297606, + "grad_norm": 2.24066162109375, + "learning_rate": 4.7255762897914374e-07, + "loss": 1.0588, + "mean_token_accuracy": 0.6899943351745605, + "num_tokens": 22250050.0, + "step": 862 + }, + { + "epoch": 0.09477267735558972, + "grad_norm": 2.499695301055908, + "learning_rate": 4.731064763995609e-07, + "loss": 1.0931, + "mean_token_accuracy": 0.6826431751251221, + "num_tokens": 22273801.0, + "step": 863 + }, + { + "epoch": 0.09488249505820338, + "grad_norm": 2.193009614944458, + "learning_rate": 4.73655323819978e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6954149007797241, + "num_tokens": 22301976.0, + "step": 864 + }, + { + "epoch": 0.09499231276081704, + "grad_norm": 1.9756901264190674, + "learning_rate": 4.7420417124039514e-07, + "loss": 1.1218, + "mean_token_accuracy": 0.666492223739624, + "num_tokens": 22339180.0, + "step": 865 + }, + { + "epoch": 0.0951021304634307, + "grad_norm": 2.5628325939178467, + "learning_rate": 4.747530186608123e-07, + "loss": 0.9898, + "mean_token_accuracy": 0.701533854007721, + "num_tokens": 22360850.0, + "step": 866 + }, + { + "epoch": 0.09521194816604436, + "grad_norm": 2.404388427734375, + "learning_rate": 4.753018660812294e-07, + "loss": 0.9867, + "mean_token_accuracy": 0.6991231441497803, + "num_tokens": 22382951.0, + "step": 867 + }, + { + "epoch": 0.09532176586865802, + "grad_norm": 2.46844744682312, + "learning_rate": 4.7585071350164654e-07, + "loss": 0.9829, + "mean_token_accuracy": 0.7010369300842285, + "num_tokens": 22407774.0, + "step": 868 + }, + { + "epoch": 0.09543158357127168, + "grad_norm": 2.1409473419189453, + "learning_rate": 4.7639956092206365e-07, + "loss": 1.0173, + "mean_token_accuracy": 0.7039119005203247, + "num_tokens": 22436723.0, + "step": 869 + }, + { + "epoch": 0.09554140127388536, + "grad_norm": 2.5333473682403564, + "learning_rate": 4.769484083424808e-07, + "loss": 1.1105, + "mean_token_accuracy": 0.6754642724990845, + "num_tokens": 22460113.0, + "step": 870 + }, + { + "epoch": 0.09565121897649902, + "grad_norm": 2.8396432399749756, + "learning_rate": 4.774972557628979e-07, + "loss": 1.0686, + "mean_token_accuracy": 0.6779893636703491, + "num_tokens": 22477693.0, + "step": 871 + }, + { + "epoch": 0.09576103667911268, + "grad_norm": 2.2257230281829834, + "learning_rate": 4.78046103183315e-07, + "loss": 1.1518, + "mean_token_accuracy": 0.667307436466217, + "num_tokens": 22505877.0, + "step": 872 + }, + { + "epoch": 0.09587085438172634, + "grad_norm": 2.6759378910064697, + "learning_rate": 4.785949506037321e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6856222748756409, + "num_tokens": 22527662.0, + "step": 873 + }, + { + "epoch": 0.09598067208434, + "grad_norm": 2.3494720458984375, + "learning_rate": 4.791437980241493e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6796286702156067, + "num_tokens": 22552614.0, + "step": 874 + }, + { + "epoch": 0.09609048978695366, + "grad_norm": 2.6843655109405518, + "learning_rate": 4.796926454445663e-07, + "loss": 1.0345, + "mean_token_accuracy": 0.6898089647293091, + "num_tokens": 22573529.0, + "step": 875 + }, + { + "epoch": 0.09620030748956732, + "grad_norm": 2.5312976837158203, + "learning_rate": 4.802414928649835e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6905860900878906, + "num_tokens": 22595759.0, + "step": 876 + }, + { + "epoch": 0.09631012519218098, + "grad_norm": 2.2463905811309814, + "learning_rate": 4.807903402854007e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.6916900277137756, + "num_tokens": 22622606.0, + "step": 877 + }, + { + "epoch": 0.09641994289479464, + "grad_norm": 2.4833905696868896, + "learning_rate": 4.813391877058177e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.7005859017372131, + "num_tokens": 22646241.0, + "step": 878 + }, + { + "epoch": 0.0965297605974083, + "grad_norm": 2.420154333114624, + "learning_rate": 4.818880351262349e-07, + "loss": 1.0549, + "mean_token_accuracy": 0.6805309057235718, + "num_tokens": 22671241.0, + "step": 879 + }, + { + "epoch": 0.09663957830002197, + "grad_norm": 2.515368938446045, + "learning_rate": 4.82436882546652e-07, + "loss": 1.0856, + "mean_token_accuracy": 0.6804119348526001, + "num_tokens": 22694307.0, + "step": 880 + }, + { + "epoch": 0.09674939600263563, + "grad_norm": 2.55827260017395, + "learning_rate": 4.829857299670691e-07, + "loss": 1.0803, + "mean_token_accuracy": 0.6755251884460449, + "num_tokens": 22716118.0, + "step": 881 + }, + { + "epoch": 0.09685921370524929, + "grad_norm": 2.2700390815734863, + "learning_rate": 4.835345773874863e-07, + "loss": 1.0371, + "mean_token_accuracy": 0.6922596096992493, + "num_tokens": 22744556.0, + "step": 882 + }, + { + "epoch": 0.09696903140786295, + "grad_norm": 2.2238941192626953, + "learning_rate": 4.840834248079034e-07, + "loss": 1.0446, + "mean_token_accuracy": 0.6909739375114441, + "num_tokens": 22774095.0, + "step": 883 + }, + { + "epoch": 0.09707884911047661, + "grad_norm": 2.5859320163726807, + "learning_rate": 4.846322722283204e-07, + "loss": 1.0126, + "mean_token_accuracy": 0.7022597789764404, + "num_tokens": 22796458.0, + "step": 884 + }, + { + "epoch": 0.09718866681309027, + "grad_norm": 2.3746562004089355, + "learning_rate": 4.851811196487376e-07, + "loss": 1.0012, + "mean_token_accuracy": 0.699434757232666, + "num_tokens": 22822413.0, + "step": 885 + }, + { + "epoch": 0.09729848451570393, + "grad_norm": 2.004987955093384, + "learning_rate": 4.857299670691548e-07, + "loss": 1.0334, + "mean_token_accuracy": 0.6966952085494995, + "num_tokens": 22854807.0, + "step": 886 + }, + { + "epoch": 0.09740830221831759, + "grad_norm": 2.9024131298065186, + "learning_rate": 4.862788144895718e-07, + "loss": 1.0925, + "mean_token_accuracy": 0.6804897785186768, + "num_tokens": 22875370.0, + "step": 887 + }, + { + "epoch": 0.09751811992093125, + "grad_norm": 2.272573232650757, + "learning_rate": 4.86827661909989e-07, + "loss": 1.031, + "mean_token_accuracy": 0.6913384795188904, + "num_tokens": 22903395.0, + "step": 888 + }, + { + "epoch": 0.09762793762354492, + "grad_norm": 2.386504888534546, + "learning_rate": 4.873765093304062e-07, + "loss": 0.9646, + "mean_token_accuracy": 0.7165899276733398, + "num_tokens": 22929550.0, + "step": 889 + }, + { + "epoch": 0.09773775532615858, + "grad_norm": 2.4665112495422363, + "learning_rate": 4.879253567508232e-07, + "loss": 0.9755, + "mean_token_accuracy": 0.7006858587265015, + "num_tokens": 22953065.0, + "step": 890 + }, + { + "epoch": 0.09784757302877224, + "grad_norm": 2.3398802280426025, + "learning_rate": 4.884742041712404e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6881603598594666, + "num_tokens": 22979228.0, + "step": 891 + }, + { + "epoch": 0.0979573907313859, + "grad_norm": 2.2313730716705322, + "learning_rate": 4.890230515916576e-07, + "loss": 1.0924, + "mean_token_accuracy": 0.6779606342315674, + "num_tokens": 23006058.0, + "step": 892 + }, + { + "epoch": 0.09806720843399956, + "grad_norm": 2.218327045440674, + "learning_rate": 4.895718990120746e-07, + "loss": 1.2255, + "mean_token_accuracy": 0.6470960974693298, + "num_tokens": 23036912.0, + "step": 893 + }, + { + "epoch": 0.09817702613661322, + "grad_norm": 2.360239028930664, + "learning_rate": 4.901207464324917e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.696025013923645, + "num_tokens": 23061443.0, + "step": 894 + }, + { + "epoch": 0.09828684383922688, + "grad_norm": 2.6822657585144043, + "learning_rate": 4.906695938529089e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6748154163360596, + "num_tokens": 23083660.0, + "step": 895 + }, + { + "epoch": 0.09839666154184054, + "grad_norm": 2.4664011001586914, + "learning_rate": 4.91218441273326e-07, + "loss": 0.9265, + "mean_token_accuracy": 0.7230502367019653, + "num_tokens": 23106674.0, + "step": 896 + }, + { + "epoch": 0.0985064792444542, + "grad_norm": 2.3146281242370605, + "learning_rate": 4.917672886937431e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6843773722648621, + "num_tokens": 23132791.0, + "step": 897 + }, + { + "epoch": 0.09861629694706786, + "grad_norm": 2.3462367057800293, + "learning_rate": 4.923161361141603e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6733270287513733, + "num_tokens": 23157563.0, + "step": 898 + }, + { + "epoch": 0.09872611464968153, + "grad_norm": 2.365957736968994, + "learning_rate": 4.928649835345773e-07, + "loss": 1.1658, + "mean_token_accuracy": 0.684265673160553, + "num_tokens": 23183059.0, + "step": 899 + }, + { + "epoch": 0.0988359323522952, + "grad_norm": 2.36690092086792, + "learning_rate": 4.934138309549945e-07, + "loss": 1.1342, + "mean_token_accuracy": 0.6659113168716431, + "num_tokens": 23211530.0, + "step": 900 + }, + { + "epoch": 0.09894575005490885, + "grad_norm": 2.4157965183258057, + "learning_rate": 4.939626783754117e-07, + "loss": 1.0996, + "mean_token_accuracy": 0.6729130744934082, + "num_tokens": 23236453.0, + "step": 901 + }, + { + "epoch": 0.09905556775752251, + "grad_norm": 2.1055095195770264, + "learning_rate": 4.945115257958287e-07, + "loss": 1.1076, + "mean_token_accuracy": 0.6803542375564575, + "num_tokens": 23268583.0, + "step": 902 + }, + { + "epoch": 0.09916538546013617, + "grad_norm": 2.328883171081543, + "learning_rate": 4.950603732162459e-07, + "loss": 0.9552, + "mean_token_accuracy": 0.7146031856536865, + "num_tokens": 23293747.0, + "step": 903 + }, + { + "epoch": 0.09927520316274983, + "grad_norm": 2.542253017425537, + "learning_rate": 4.95609220636663e-07, + "loss": 1.0582, + "mean_token_accuracy": 0.6936948895454407, + "num_tokens": 23316804.0, + "step": 904 + }, + { + "epoch": 0.0993850208653635, + "grad_norm": 2.1639645099639893, + "learning_rate": 4.961580680570801e-07, + "loss": 1.1223, + "mean_token_accuracy": 0.674518346786499, + "num_tokens": 23347046.0, + "step": 905 + }, + { + "epoch": 0.09949483856797715, + "grad_norm": 2.3612749576568604, + "learning_rate": 4.967069154774972e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6762872934341431, + "num_tokens": 23374429.0, + "step": 906 + }, + { + "epoch": 0.09960465627059081, + "grad_norm": 2.3086934089660645, + "learning_rate": 4.972557628979143e-07, + "loss": 1.106, + "mean_token_accuracy": 0.6765380501747131, + "num_tokens": 23400389.0, + "step": 907 + }, + { + "epoch": 0.09971447397320449, + "grad_norm": 2.3200008869171143, + "learning_rate": 4.978046103183315e-07, + "loss": 1.1052, + "mean_token_accuracy": 0.6756980419158936, + "num_tokens": 23428027.0, + "step": 908 + }, + { + "epoch": 0.09982429167581815, + "grad_norm": 2.159820318222046, + "learning_rate": 4.983534577387486e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6854719519615173, + "num_tokens": 23457021.0, + "step": 909 + }, + { + "epoch": 0.0999341093784318, + "grad_norm": 2.0506136417388916, + "learning_rate": 4.989023051591657e-07, + "loss": 1.1123, + "mean_token_accuracy": 0.68117356300354, + "num_tokens": 23486930.0, + "step": 910 + }, + { + "epoch": 0.10004392708104547, + "grad_norm": 2.4858484268188477, + "learning_rate": 4.994511525795829e-07, + "loss": 1.0651, + "mean_token_accuracy": 0.6814738512039185, + "num_tokens": 23510614.0, + "step": 911 + }, + { + "epoch": 0.10015374478365913, + "grad_norm": 2.047617197036743, + "learning_rate": 5e-07, + "loss": 1.1728, + "mean_token_accuracy": 0.6622492074966431, + "num_tokens": 23543957.0, + "step": 912 + }, + { + "epoch": 0.10026356248627279, + "grad_norm": 2.588361978530884, + "learning_rate": 5.005488474204171e-07, + "loss": 1.1503, + "mean_token_accuracy": 0.6618077158927917, + "num_tokens": 23566720.0, + "step": 913 + }, + { + "epoch": 0.10037338018888645, + "grad_norm": 2.5054726600646973, + "learning_rate": 5.010976948408342e-07, + "loss": 1.0113, + "mean_token_accuracy": 0.6971554160118103, + "num_tokens": 23589720.0, + "step": 914 + }, + { + "epoch": 0.1004831978915001, + "grad_norm": 2.0690252780914307, + "learning_rate": 5.016465422612514e-07, + "loss": 1.1137, + "mean_token_accuracy": 0.6747691631317139, + "num_tokens": 23621661.0, + "step": 915 + }, + { + "epoch": 0.10059301559411377, + "grad_norm": 2.42343807220459, + "learning_rate": 5.021953896816685e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.6869454383850098, + "num_tokens": 23645751.0, + "step": 916 + }, + { + "epoch": 0.10070283329672743, + "grad_norm": 2.453974723815918, + "learning_rate": 5.027442371020856e-07, + "loss": 1.0844, + "mean_token_accuracy": 0.6837403178215027, + "num_tokens": 23670677.0, + "step": 917 + }, + { + "epoch": 0.1008126509993411, + "grad_norm": 2.3581833839416504, + "learning_rate": 5.032930845225028e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.6996740102767944, + "num_tokens": 23696957.0, + "step": 918 + }, + { + "epoch": 0.10092246870195476, + "grad_norm": 2.167725086212158, + "learning_rate": 5.038419319429198e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6878770589828491, + "num_tokens": 23725382.0, + "step": 919 + }, + { + "epoch": 0.10103228640456842, + "grad_norm": 2.1599178314208984, + "learning_rate": 5.04390779363337e-07, + "loss": 1.0273, + "mean_token_accuracy": 0.6924800872802734, + "num_tokens": 23754163.0, + "step": 920 + }, + { + "epoch": 0.10114210410718208, + "grad_norm": 2.338192939758301, + "learning_rate": 5.049396267837542e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6959739923477173, + "num_tokens": 23780669.0, + "step": 921 + }, + { + "epoch": 0.10125192180979574, + "grad_norm": 2.3341763019561768, + "learning_rate": 5.054884742041711e-07, + "loss": 1.0761, + "mean_token_accuracy": 0.6775417327880859, + "num_tokens": 23805037.0, + "step": 922 + }, + { + "epoch": 0.1013617395124094, + "grad_norm": 2.41827654838562, + "learning_rate": 5.060373216245883e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6863421201705933, + "num_tokens": 23829718.0, + "step": 923 + }, + { + "epoch": 0.10147155721502306, + "grad_norm": 2.0453271865844727, + "learning_rate": 5.065861690450055e-07, + "loss": 1.1443, + "mean_token_accuracy": 0.6643802523612976, + "num_tokens": 23860745.0, + "step": 924 + }, + { + "epoch": 0.10158137491763672, + "grad_norm": 2.3413779735565186, + "learning_rate": 5.071350164654225e-07, + "loss": 1.1464, + "mean_token_accuracy": 0.6620912551879883, + "num_tokens": 23889457.0, + "step": 925 + }, + { + "epoch": 0.10169119262025038, + "grad_norm": 2.853504180908203, + "learning_rate": 5.076838638858397e-07, + "loss": 1.0094, + "mean_token_accuracy": 0.6992480754852295, + "num_tokens": 23908688.0, + "step": 926 + }, + { + "epoch": 0.10180101032286405, + "grad_norm": 2.5344386100769043, + "learning_rate": 5.082327113062569e-07, + "loss": 1.0221, + "mean_token_accuracy": 0.6912761926651001, + "num_tokens": 23930010.0, + "step": 927 + }, + { + "epoch": 0.10191082802547771, + "grad_norm": 2.185608148574829, + "learning_rate": 5.087815587266739e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6890936493873596, + "num_tokens": 23959660.0, + "step": 928 + }, + { + "epoch": 0.10202064572809137, + "grad_norm": 2.4299190044403076, + "learning_rate": 5.093304061470911e-07, + "loss": 1.089, + "mean_token_accuracy": 0.6736373901367188, + "num_tokens": 23985205.0, + "step": 929 + }, + { + "epoch": 0.10213046343070503, + "grad_norm": 2.5130701065063477, + "learning_rate": 5.098792535675082e-07, + "loss": 1.0644, + "mean_token_accuracy": 0.6908049583435059, + "num_tokens": 24009096.0, + "step": 930 + }, + { + "epoch": 0.10224028113331869, + "grad_norm": 2.307579755783081, + "learning_rate": 5.104281009879253e-07, + "loss": 1.1225, + "mean_token_accuracy": 0.6685721278190613, + "num_tokens": 24039034.0, + "step": 931 + }, + { + "epoch": 0.10235009883593235, + "grad_norm": 2.4020135402679443, + "learning_rate": 5.109769484083425e-07, + "loss": 1.1318, + "mean_token_accuracy": 0.6671626567840576, + "num_tokens": 24065785.0, + "step": 932 + }, + { + "epoch": 0.10245991653854601, + "grad_norm": 2.340824604034424, + "learning_rate": 5.115257958287596e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.6887220144271851, + "num_tokens": 24093137.0, + "step": 933 + }, + { + "epoch": 0.10256973424115967, + "grad_norm": 2.398787498474121, + "learning_rate": 5.120746432491767e-07, + "loss": 1.0762, + "mean_token_accuracy": 0.6878340840339661, + "num_tokens": 24117309.0, + "step": 934 + }, + { + "epoch": 0.10267955194377333, + "grad_norm": 2.304546594619751, + "learning_rate": 5.126234906695939e-07, + "loss": 1.1287, + "mean_token_accuracy": 0.6742352247238159, + "num_tokens": 24143990.0, + "step": 935 + }, + { + "epoch": 0.10278936964638699, + "grad_norm": 2.444589853286743, + "learning_rate": 5.13172338090011e-07, + "loss": 1.0817, + "mean_token_accuracy": 0.6789841055870056, + "num_tokens": 24169468.0, + "step": 936 + }, + { + "epoch": 0.10289918734900066, + "grad_norm": 2.376354455947876, + "learning_rate": 5.137211855104281e-07, + "loss": 1.0884, + "mean_token_accuracy": 0.6785037517547607, + "num_tokens": 24196761.0, + "step": 937 + }, + { + "epoch": 0.10300900505161432, + "grad_norm": 2.6013216972351074, + "learning_rate": 5.142700329308453e-07, + "loss": 1.0153, + "mean_token_accuracy": 0.695171594619751, + "num_tokens": 24217959.0, + "step": 938 + }, + { + "epoch": 0.10311882275422798, + "grad_norm": 2.761552095413208, + "learning_rate": 5.148188803512624e-07, + "loss": 0.9277, + "mean_token_accuracy": 0.717647910118103, + "num_tokens": 24236885.0, + "step": 939 + }, + { + "epoch": 0.10322864045684164, + "grad_norm": 2.2886266708374023, + "learning_rate": 5.153677277716795e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6939096450805664, + "num_tokens": 24262406.0, + "step": 940 + }, + { + "epoch": 0.1033384581594553, + "grad_norm": 2.463798761367798, + "learning_rate": 5.159165751920965e-07, + "loss": 0.9935, + "mean_token_accuracy": 0.7040731906890869, + "num_tokens": 24287953.0, + "step": 941 + }, + { + "epoch": 0.10344827586206896, + "grad_norm": 2.148329257965088, + "learning_rate": 5.164654226125136e-07, + "loss": 1.1556, + "mean_token_accuracy": 0.6583994030952454, + "num_tokens": 24319169.0, + "step": 942 + }, + { + "epoch": 0.10355809356468262, + "grad_norm": 2.223951578140259, + "learning_rate": 5.170142700329308e-07, + "loss": 1.116, + "mean_token_accuracy": 0.669447660446167, + "num_tokens": 24348183.0, + "step": 943 + }, + { + "epoch": 0.10366791126729628, + "grad_norm": 2.4036827087402344, + "learning_rate": 5.175631174533479e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6766678094863892, + "num_tokens": 24372545.0, + "step": 944 + }, + { + "epoch": 0.10377772896990994, + "grad_norm": 2.1464569568634033, + "learning_rate": 5.18111964873765e-07, + "loss": 1.0235, + "mean_token_accuracy": 0.6951125860214233, + "num_tokens": 24401830.0, + "step": 945 + }, + { + "epoch": 0.10388754667252362, + "grad_norm": 2.2714991569519043, + "learning_rate": 5.186608122941822e-07, + "loss": 0.9703, + "mean_token_accuracy": 0.7028940916061401, + "num_tokens": 24428871.0, + "step": 946 + }, + { + "epoch": 0.10399736437513728, + "grad_norm": 2.3777973651885986, + "learning_rate": 5.192096597145993e-07, + "loss": 1.0746, + "mean_token_accuracy": 0.6824490427970886, + "num_tokens": 24453724.0, + "step": 947 + }, + { + "epoch": 0.10410718207775094, + "grad_norm": 2.5874722003936768, + "learning_rate": 5.197585071350164e-07, + "loss": 0.9969, + "mean_token_accuracy": 0.7006206512451172, + "num_tokens": 24475260.0, + "step": 948 + }, + { + "epoch": 0.1042169997803646, + "grad_norm": 2.387071371078491, + "learning_rate": 5.203073545554336e-07, + "loss": 1.1258, + "mean_token_accuracy": 0.6596475839614868, + "num_tokens": 24499286.0, + "step": 949 + }, + { + "epoch": 0.10432681748297826, + "grad_norm": 2.1473541259765625, + "learning_rate": 5.208562019758507e-07, + "loss": 1.097, + "mean_token_accuracy": 0.6727919578552246, + "num_tokens": 24528795.0, + "step": 950 + }, + { + "epoch": 0.10443663518559192, + "grad_norm": 2.2819671630859375, + "learning_rate": 5.214050493962678e-07, + "loss": 1.1128, + "mean_token_accuracy": 0.6744477152824402, + "num_tokens": 24555175.0, + "step": 951 + }, + { + "epoch": 0.10454645288820558, + "grad_norm": 2.7956082820892334, + "learning_rate": 5.219538968166849e-07, + "loss": 1.0713, + "mean_token_accuracy": 0.6854521036148071, + "num_tokens": 24575292.0, + "step": 952 + }, + { + "epoch": 0.10465627059081924, + "grad_norm": 2.210749864578247, + "learning_rate": 5.225027442371021e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.6930984854698181, + "num_tokens": 24601927.0, + "step": 953 + }, + { + "epoch": 0.1047660882934329, + "grad_norm": 2.444714307785034, + "learning_rate": 5.230515916575192e-07, + "loss": 0.8982, + "mean_token_accuracy": 0.7164467573165894, + "num_tokens": 24623376.0, + "step": 954 + }, + { + "epoch": 0.10487590599604656, + "grad_norm": 2.7190945148468018, + "learning_rate": 5.236004390779363e-07, + "loss": 1.0008, + "mean_token_accuracy": 0.699238657951355, + "num_tokens": 24643419.0, + "step": 955 + }, + { + "epoch": 0.10498572369866023, + "grad_norm": 2.3600192070007324, + "learning_rate": 5.241492864983535e-07, + "loss": 1.067, + "mean_token_accuracy": 0.6878836154937744, + "num_tokens": 24668512.0, + "step": 956 + }, + { + "epoch": 0.10509554140127389, + "grad_norm": 2.3574514389038086, + "learning_rate": 5.246981339187706e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6768792271614075, + "num_tokens": 24695919.0, + "step": 957 + }, + { + "epoch": 0.10520535910388755, + "grad_norm": 2.2165911197662354, + "learning_rate": 5.252469813391877e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6893429160118103, + "num_tokens": 24723202.0, + "step": 958 + }, + { + "epoch": 0.10531517680650121, + "grad_norm": 2.454404830932617, + "learning_rate": 5.257958287596049e-07, + "loss": 1.0531, + "mean_token_accuracy": 0.6880818605422974, + "num_tokens": 24746964.0, + "step": 959 + }, + { + "epoch": 0.10542499450911487, + "grad_norm": 2.203321933746338, + "learning_rate": 5.263446761800219e-07, + "loss": 1.118, + "mean_token_accuracy": 0.6700640916824341, + "num_tokens": 24773681.0, + "step": 960 + }, + { + "epoch": 0.10553481221172853, + "grad_norm": 2.1595048904418945, + "learning_rate": 5.26893523600439e-07, + "loss": 1.0024, + "mean_token_accuracy": 0.6995117664337158, + "num_tokens": 24803912.0, + "step": 961 + }, + { + "epoch": 0.10564462991434219, + "grad_norm": 2.276789426803589, + "learning_rate": 5.274423710208562e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.7041915655136108, + "num_tokens": 24830738.0, + "step": 962 + }, + { + "epoch": 0.10575444761695585, + "grad_norm": 2.4062554836273193, + "learning_rate": 5.279912184412732e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6872603297233582, + "num_tokens": 24855273.0, + "step": 963 + }, + { + "epoch": 0.10586426531956951, + "grad_norm": 2.1050808429718018, + "learning_rate": 5.285400658616904e-07, + "loss": 1.1095, + "mean_token_accuracy": 0.675512433052063, + "num_tokens": 24887234.0, + "step": 964 + }, + { + "epoch": 0.10597408302218318, + "grad_norm": 2.3733530044555664, + "learning_rate": 5.290889132821076e-07, + "loss": 1.0636, + "mean_token_accuracy": 0.6936889886856079, + "num_tokens": 24911252.0, + "step": 965 + }, + { + "epoch": 0.10608390072479684, + "grad_norm": 2.224881172180176, + "learning_rate": 5.296377607025246e-07, + "loss": 1.0745, + "mean_token_accuracy": 0.6952739357948303, + "num_tokens": 24937252.0, + "step": 966 + }, + { + "epoch": 0.1061937184274105, + "grad_norm": 2.0235612392425537, + "learning_rate": 5.301866081229418e-07, + "loss": 1.0629, + "mean_token_accuracy": 0.6792044639587402, + "num_tokens": 24967734.0, + "step": 967 + }, + { + "epoch": 0.10630353613002416, + "grad_norm": 2.3373141288757324, + "learning_rate": 5.30735455543359e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6929903626441956, + "num_tokens": 24993830.0, + "step": 968 + }, + { + "epoch": 0.10641335383263782, + "grad_norm": 2.5837161540985107, + "learning_rate": 5.31284302963776e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.693687379360199, + "num_tokens": 25014743.0, + "step": 969 + }, + { + "epoch": 0.10652317153525148, + "grad_norm": 2.3752286434173584, + "learning_rate": 5.318331503841932e-07, + "loss": 0.9605, + "mean_token_accuracy": 0.7101156711578369, + "num_tokens": 25039092.0, + "step": 970 + }, + { + "epoch": 0.10663298923786514, + "grad_norm": 2.1868374347686768, + "learning_rate": 5.323819978046103e-07, + "loss": 1.0334, + "mean_token_accuracy": 0.6909586191177368, + "num_tokens": 25068513.0, + "step": 971 + }, + { + "epoch": 0.1067428069404788, + "grad_norm": 2.606801986694336, + "learning_rate": 5.329308452250274e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.6909224987030029, + "num_tokens": 25089764.0, + "step": 972 + }, + { + "epoch": 0.10685262464309246, + "grad_norm": 2.0672898292541504, + "learning_rate": 5.334796926454446e-07, + "loss": 1.0099, + "mean_token_accuracy": 0.7013018131256104, + "num_tokens": 25119263.0, + "step": 973 + }, + { + "epoch": 0.10696244234570612, + "grad_norm": 2.154813766479492, + "learning_rate": 5.340285400658617e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.683044970035553, + "num_tokens": 25147471.0, + "step": 974 + }, + { + "epoch": 0.1070722600483198, + "grad_norm": 2.595747232437134, + "learning_rate": 5.345773874862788e-07, + "loss": 1.0143, + "mean_token_accuracy": 0.6889299750328064, + "num_tokens": 25167742.0, + "step": 975 + }, + { + "epoch": 0.10718207775093345, + "grad_norm": 2.533710479736328, + "learning_rate": 5.35126234906696e-07, + "loss": 0.9657, + "mean_token_accuracy": 0.7057723999023438, + "num_tokens": 25189341.0, + "step": 976 + }, + { + "epoch": 0.10729189545354711, + "grad_norm": 2.180846691131592, + "learning_rate": 5.35675082327113e-07, + "loss": 1.0813, + "mean_token_accuracy": 0.6797252297401428, + "num_tokens": 25217912.0, + "step": 977 + }, + { + "epoch": 0.10740171315616077, + "grad_norm": 2.453178644180298, + "learning_rate": 5.362239297475302e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6721063852310181, + "num_tokens": 25240119.0, + "step": 978 + }, + { + "epoch": 0.10751153085877443, + "grad_norm": 2.3632588386535645, + "learning_rate": 5.367727771679473e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.6962274312973022, + "num_tokens": 25266245.0, + "step": 979 + }, + { + "epoch": 0.1076213485613881, + "grad_norm": 2.3859896659851074, + "learning_rate": 5.373216245883643e-07, + "loss": 0.9857, + "mean_token_accuracy": 0.7015424370765686, + "num_tokens": 25290150.0, + "step": 980 + }, + { + "epoch": 0.10773116626400175, + "grad_norm": 2.1616125106811523, + "learning_rate": 5.378704720087815e-07, + "loss": 1.1355, + "mean_token_accuracy": 0.6747676730155945, + "num_tokens": 25320551.0, + "step": 981 + }, + { + "epoch": 0.10784098396661541, + "grad_norm": 2.0514392852783203, + "learning_rate": 5.384193194291986e-07, + "loss": 1.0596, + "mean_token_accuracy": 0.6948171257972717, + "num_tokens": 25350476.0, + "step": 982 + }, + { + "epoch": 0.10795080166922907, + "grad_norm": 2.23030686378479, + "learning_rate": 5.389681668496157e-07, + "loss": 1.1687, + "mean_token_accuracy": 0.6647801995277405, + "num_tokens": 25380167.0, + "step": 983 + }, + { + "epoch": 0.10806061937184275, + "grad_norm": 2.140598773956299, + "learning_rate": 5.395170142700329e-07, + "loss": 1.0688, + "mean_token_accuracy": 0.678547203540802, + "num_tokens": 25409083.0, + "step": 984 + }, + { + "epoch": 0.10817043707445641, + "grad_norm": 2.41489577293396, + "learning_rate": 5.4006586169045e-07, + "loss": 1.0432, + "mean_token_accuracy": 0.6920883655548096, + "num_tokens": 25434326.0, + "step": 985 + }, + { + "epoch": 0.10828025477707007, + "grad_norm": 2.5185720920562744, + "learning_rate": 5.406147091108671e-07, + "loss": 1.0715, + "mean_token_accuracy": 0.6824748516082764, + "num_tokens": 25456985.0, + "step": 986 + }, + { + "epoch": 0.10839007247968373, + "grad_norm": 2.4522128105163574, + "learning_rate": 5.411635565312843e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.6895917057991028, + "num_tokens": 25479640.0, + "step": 987 + }, + { + "epoch": 0.10849989018229739, + "grad_norm": 2.2129769325256348, + "learning_rate": 5.417124039517014e-07, + "loss": 1.1561, + "mean_token_accuracy": 0.6661331057548523, + "num_tokens": 25508935.0, + "step": 988 + }, + { + "epoch": 0.10860970788491105, + "grad_norm": 2.3411431312561035, + "learning_rate": 5.422612513721185e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.6850426197052002, + "num_tokens": 25535588.0, + "step": 989 + }, + { + "epoch": 0.1087195255875247, + "grad_norm": 2.3178789615631104, + "learning_rate": 5.428100987925357e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6922120451927185, + "num_tokens": 25561553.0, + "step": 990 + }, + { + "epoch": 0.10882934329013837, + "grad_norm": 2.4539942741394043, + "learning_rate": 5.433589462129528e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.7023168206214905, + "num_tokens": 25584950.0, + "step": 991 + }, + { + "epoch": 0.10893916099275203, + "grad_norm": 2.311992645263672, + "learning_rate": 5.439077936333699e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.696457028388977, + "num_tokens": 25610408.0, + "step": 992 + }, + { + "epoch": 0.10904897869536569, + "grad_norm": 2.571066379547119, + "learning_rate": 5.44456641053787e-07, + "loss": 1.0724, + "mean_token_accuracy": 0.6928174495697021, + "num_tokens": 25632495.0, + "step": 993 + }, + { + "epoch": 0.10915879639797936, + "grad_norm": 2.0548157691955566, + "learning_rate": 5.450054884742042e-07, + "loss": 1.1306, + "mean_token_accuracy": 0.6607271432876587, + "num_tokens": 25665317.0, + "step": 994 + }, + { + "epoch": 0.10926861410059302, + "grad_norm": 2.541700601577759, + "learning_rate": 5.455543358946213e-07, + "loss": 0.9982, + "mean_token_accuracy": 0.7030100226402283, + "num_tokens": 25687663.0, + "step": 995 + }, + { + "epoch": 0.10937843180320668, + "grad_norm": 2.4812264442443848, + "learning_rate": 5.461031833150384e-07, + "loss": 0.9731, + "mean_token_accuracy": 0.7062811851501465, + "num_tokens": 25710254.0, + "step": 996 + }, + { + "epoch": 0.10948824950582034, + "grad_norm": 2.4395933151245117, + "learning_rate": 5.466520307354556e-07, + "loss": 1.0523, + "mean_token_accuracy": 0.6874176859855652, + "num_tokens": 25734559.0, + "step": 997 + }, + { + "epoch": 0.109598067208434, + "grad_norm": 2.476914644241333, + "learning_rate": 5.472008781558726e-07, + "loss": 1.0343, + "mean_token_accuracy": 0.693940281867981, + "num_tokens": 25757096.0, + "step": 998 + }, + { + "epoch": 0.10970788491104766, + "grad_norm": 2.399264097213745, + "learning_rate": 5.477497255762897e-07, + "loss": 1.1717, + "mean_token_accuracy": 0.6696358323097229, + "num_tokens": 25784397.0, + "step": 999 + }, + { + "epoch": 0.10981770261366132, + "grad_norm": 2.1053168773651123, + "learning_rate": 5.482985729967069e-07, + "loss": 0.9909, + "mean_token_accuracy": 0.6992924213409424, + "num_tokens": 25811567.0, + "step": 1000 + }, + { + "epoch": 0.10992752031627498, + "grad_norm": 2.389148712158203, + "learning_rate": 5.48847420417124e-07, + "loss": 1.0427, + "mean_token_accuracy": 0.685346245765686, + "num_tokens": 25837777.0, + "step": 1001 + }, + { + "epoch": 0.11003733801888864, + "grad_norm": 2.1505126953125, + "learning_rate": 5.493962678375411e-07, + "loss": 1.1567, + "mean_token_accuracy": 0.6646157503128052, + "num_tokens": 25868009.0, + "step": 1002 + }, + { + "epoch": 0.11014715572150231, + "grad_norm": 2.327882766723633, + "learning_rate": 5.499451152579583e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.6815264225006104, + "num_tokens": 25895714.0, + "step": 1003 + }, + { + "epoch": 0.11025697342411597, + "grad_norm": 2.2322475910186768, + "learning_rate": 5.504939626783753e-07, + "loss": 1.0134, + "mean_token_accuracy": 0.6936473846435547, + "num_tokens": 25922742.0, + "step": 1004 + }, + { + "epoch": 0.11036679112672963, + "grad_norm": 2.365285873413086, + "learning_rate": 5.510428100987925e-07, + "loss": 1.1554, + "mean_token_accuracy": 0.6560394763946533, + "num_tokens": 25948010.0, + "step": 1005 + }, + { + "epoch": 0.11047660882934329, + "grad_norm": 2.7707321643829346, + "learning_rate": 5.515916575192097e-07, + "loss": 0.9924, + "mean_token_accuracy": 0.700779914855957, + "num_tokens": 25966147.0, + "step": 1006 + }, + { + "epoch": 0.11058642653195695, + "grad_norm": 2.379115343093872, + "learning_rate": 5.521405049396267e-07, + "loss": 1.0741, + "mean_token_accuracy": 0.6850048899650574, + "num_tokens": 25991527.0, + "step": 1007 + }, + { + "epoch": 0.11069624423457061, + "grad_norm": 2.3863611221313477, + "learning_rate": 5.526893523600439e-07, + "loss": 0.9655, + "mean_token_accuracy": 0.7067258358001709, + "num_tokens": 26014825.0, + "step": 1008 + }, + { + "epoch": 0.11080606193718427, + "grad_norm": 2.6967763900756836, + "learning_rate": 5.532381997804611e-07, + "loss": 0.9697, + "mean_token_accuracy": 0.7105035185813904, + "num_tokens": 26033734.0, + "step": 1009 + }, + { + "epoch": 0.11091587963979793, + "grad_norm": 2.326188564300537, + "learning_rate": 5.537870472008781e-07, + "loss": 1.0888, + "mean_token_accuracy": 0.676512598991394, + "num_tokens": 26058606.0, + "step": 1010 + }, + { + "epoch": 0.11102569734241159, + "grad_norm": 2.5041563510894775, + "learning_rate": 5.543358946212953e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.7056800127029419, + "num_tokens": 26081200.0, + "step": 1011 + }, + { + "epoch": 0.11113551504502525, + "grad_norm": 2.2983205318450928, + "learning_rate": 5.548847420417125e-07, + "loss": 1.0975, + "mean_token_accuracy": 0.6716651916503906, + "num_tokens": 26107864.0, + "step": 1012 + }, + { + "epoch": 0.11124533274763893, + "grad_norm": 2.0430099964141846, + "learning_rate": 5.554335894621295e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6931126117706299, + "num_tokens": 26139911.0, + "step": 1013 + }, + { + "epoch": 0.11135515045025259, + "grad_norm": 2.1664650440216064, + "learning_rate": 5.559824368825467e-07, + "loss": 1.1256, + "mean_token_accuracy": 0.6709228157997131, + "num_tokens": 26169846.0, + "step": 1014 + }, + { + "epoch": 0.11146496815286625, + "grad_norm": 2.261256456375122, + "learning_rate": 5.565312843029637e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6851999759674072, + "num_tokens": 26195926.0, + "step": 1015 + }, + { + "epoch": 0.1115747858554799, + "grad_norm": 2.5819647312164307, + "learning_rate": 5.570801317233809e-07, + "loss": 0.9572, + "mean_token_accuracy": 0.7103766202926636, + "num_tokens": 26216099.0, + "step": 1016 + }, + { + "epoch": 0.11168460355809356, + "grad_norm": 2.3266446590423584, + "learning_rate": 5.57628979143798e-07, + "loss": 1.0594, + "mean_token_accuracy": 0.6852415800094604, + "num_tokens": 26242802.0, + "step": 1017 + }, + { + "epoch": 0.11179442126070722, + "grad_norm": 2.4078407287597656, + "learning_rate": 5.58177826564215e-07, + "loss": 0.9941, + "mean_token_accuracy": 0.694800853729248, + "num_tokens": 26266574.0, + "step": 1018 + }, + { + "epoch": 0.11190423896332088, + "grad_norm": 2.382528066635132, + "learning_rate": 5.587266739846322e-07, + "loss": 0.9892, + "mean_token_accuracy": 0.7013718485832214, + "num_tokens": 26290866.0, + "step": 1019 + }, + { + "epoch": 0.11201405666593454, + "grad_norm": 2.2091569900512695, + "learning_rate": 5.592755214050494e-07, + "loss": 1.1304, + "mean_token_accuracy": 0.6655460000038147, + "num_tokens": 26319375.0, + "step": 1020 + }, + { + "epoch": 0.1121238743685482, + "grad_norm": 2.2180817127227783, + "learning_rate": 5.598243688254664e-07, + "loss": 1.1119, + "mean_token_accuracy": 0.6718626022338867, + "num_tokens": 26347924.0, + "step": 1021 + }, + { + "epoch": 0.11223369207116188, + "grad_norm": 2.569020986557007, + "learning_rate": 5.603732162458836e-07, + "loss": 1.1007, + "mean_token_accuracy": 0.6807947754859924, + "num_tokens": 26371727.0, + "step": 1022 + }, + { + "epoch": 0.11234350977377554, + "grad_norm": 2.1764333248138428, + "learning_rate": 5.609220636663008e-07, + "loss": 0.9638, + "mean_token_accuracy": 0.7098456025123596, + "num_tokens": 26398356.0, + "step": 1023 + }, + { + "epoch": 0.1124533274763892, + "grad_norm": 2.308670997619629, + "learning_rate": 5.614709110867178e-07, + "loss": 1.0799, + "mean_token_accuracy": 0.6774786710739136, + "num_tokens": 26424095.0, + "step": 1024 + }, + { + "epoch": 0.11256314517900286, + "grad_norm": 2.2973058223724365, + "learning_rate": 5.62019758507135e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.690901517868042, + "num_tokens": 26451022.0, + "step": 1025 + }, + { + "epoch": 0.11267296288161652, + "grad_norm": 2.2490291595458984, + "learning_rate": 5.625686059275521e-07, + "loss": 1.0827, + "mean_token_accuracy": 0.6783904433250427, + "num_tokens": 26478780.0, + "step": 1026 + }, + { + "epoch": 0.11278278058423018, + "grad_norm": 2.413708209991455, + "learning_rate": 5.631174533479692e-07, + "loss": 1.0944, + "mean_token_accuracy": 0.6831425428390503, + "num_tokens": 26502641.0, + "step": 1027 + }, + { + "epoch": 0.11289259828684384, + "grad_norm": 2.486215591430664, + "learning_rate": 5.636663007683864e-07, + "loss": 1.1216, + "mean_token_accuracy": 0.6673924326896667, + "num_tokens": 26525481.0, + "step": 1028 + }, + { + "epoch": 0.1130024159894575, + "grad_norm": 2.330268144607544, + "learning_rate": 5.642151481888035e-07, + "loss": 1.1768, + "mean_token_accuracy": 0.655981719493866, + "num_tokens": 26553971.0, + "step": 1029 + }, + { + "epoch": 0.11311223369207116, + "grad_norm": 2.220914840698242, + "learning_rate": 5.647639956092206e-07, + "loss": 1.0868, + "mean_token_accuracy": 0.6773991584777832, + "num_tokens": 26581263.0, + "step": 1030 + }, + { + "epoch": 0.11322205139468482, + "grad_norm": 2.4760262966156006, + "learning_rate": 5.653128430296378e-07, + "loss": 1.0314, + "mean_token_accuracy": 0.6988601088523865, + "num_tokens": 26603801.0, + "step": 1031 + }, + { + "epoch": 0.11333186909729849, + "grad_norm": 2.32004714012146, + "learning_rate": 5.658616904500549e-07, + "loss": 1.0797, + "mean_token_accuracy": 0.6775333881378174, + "num_tokens": 26630845.0, + "step": 1032 + }, + { + "epoch": 0.11344168679991215, + "grad_norm": 2.1159417629241943, + "learning_rate": 5.66410537870472e-07, + "loss": 1.1265, + "mean_token_accuracy": 0.6725302934646606, + "num_tokens": 26661708.0, + "step": 1033 + }, + { + "epoch": 0.11355150450252581, + "grad_norm": 2.1417248249053955, + "learning_rate": 5.669593852908892e-07, + "loss": 1.147, + "mean_token_accuracy": 0.6624059677124023, + "num_tokens": 26690602.0, + "step": 1034 + }, + { + "epoch": 0.11366132220513947, + "grad_norm": 2.3925373554229736, + "learning_rate": 5.675082327113063e-07, + "loss": 1.01, + "mean_token_accuracy": 0.695700466632843, + "num_tokens": 26714408.0, + "step": 1035 + }, + { + "epoch": 0.11377113990775313, + "grad_norm": 2.6416382789611816, + "learning_rate": 5.680570801317233e-07, + "loss": 1.0783, + "mean_token_accuracy": 0.6919487118721008, + "num_tokens": 26735606.0, + "step": 1036 + }, + { + "epoch": 0.11388095761036679, + "grad_norm": 2.332092523574829, + "learning_rate": 5.686059275521404e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6920899748802185, + "num_tokens": 26761088.0, + "step": 1037 + }, + { + "epoch": 0.11399077531298045, + "grad_norm": 2.5638277530670166, + "learning_rate": 5.691547749725576e-07, + "loss": 1.0356, + "mean_token_accuracy": 0.6915007829666138, + "num_tokens": 26784162.0, + "step": 1038 + }, + { + "epoch": 0.11410059301559411, + "grad_norm": 2.25168776512146, + "learning_rate": 5.697036223929747e-07, + "loss": 1.1562, + "mean_token_accuracy": 0.6607928276062012, + "num_tokens": 26812675.0, + "step": 1039 + }, + { + "epoch": 0.11421041071820777, + "grad_norm": 2.278782606124878, + "learning_rate": 5.702524698133918e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6999066472053528, + "num_tokens": 26837602.0, + "step": 1040 + }, + { + "epoch": 0.11432022842082144, + "grad_norm": 2.1279001235961914, + "learning_rate": 5.70801317233809e-07, + "loss": 1.1736, + "mean_token_accuracy": 0.6490647792816162, + "num_tokens": 26870647.0, + "step": 1041 + }, + { + "epoch": 0.1144300461234351, + "grad_norm": 2.2457380294799805, + "learning_rate": 5.713501646542261e-07, + "loss": 1.0938, + "mean_token_accuracy": 0.6707664132118225, + "num_tokens": 26898567.0, + "step": 1042 + }, + { + "epoch": 0.11453986382604876, + "grad_norm": 1.9920151233673096, + "learning_rate": 5.718990120746432e-07, + "loss": 1.086, + "mean_token_accuracy": 0.6763495802879333, + "num_tokens": 26930971.0, + "step": 1043 + }, + { + "epoch": 0.11464968152866242, + "grad_norm": 2.697927951812744, + "learning_rate": 5.724478594950604e-07, + "loss": 0.9449, + "mean_token_accuracy": 0.7108691334724426, + "num_tokens": 26949455.0, + "step": 1044 + }, + { + "epoch": 0.11475949923127608, + "grad_norm": 2.4581494331359863, + "learning_rate": 5.729967069154775e-07, + "loss": 1.1271, + "mean_token_accuracy": 0.6667665243148804, + "num_tokens": 26973711.0, + "step": 1045 + }, + { + "epoch": 0.11486931693388974, + "grad_norm": 2.3198487758636475, + "learning_rate": 5.735455543358946e-07, + "loss": 0.998, + "mean_token_accuracy": 0.7009760141372681, + "num_tokens": 26998410.0, + "step": 1046 + }, + { + "epoch": 0.1149791346365034, + "grad_norm": 2.1665711402893066, + "learning_rate": 5.740944017563118e-07, + "loss": 1.0393, + "mean_token_accuracy": 0.6876307725906372, + "num_tokens": 27027082.0, + "step": 1047 + }, + { + "epoch": 0.11508895233911706, + "grad_norm": 2.1092512607574463, + "learning_rate": 5.746432491767288e-07, + "loss": 1.133, + "mean_token_accuracy": 0.6634030342102051, + "num_tokens": 27056554.0, + "step": 1048 + }, + { + "epoch": 0.11519877004173072, + "grad_norm": 2.1919164657592773, + "learning_rate": 5.75192096597146e-07, + "loss": 1.0956, + "mean_token_accuracy": 0.672532856464386, + "num_tokens": 27086078.0, + "step": 1049 + }, + { + "epoch": 0.11530858774434438, + "grad_norm": 2.4031898975372314, + "learning_rate": 5.757409440175632e-07, + "loss": 1.0253, + "mean_token_accuracy": 0.6892629861831665, + "num_tokens": 27112359.0, + "step": 1050 + }, + { + "epoch": 0.11541840544695806, + "grad_norm": 2.3360323905944824, + "learning_rate": 5.762897914379802e-07, + "loss": 1.058, + "mean_token_accuracy": 0.6954743266105652, + "num_tokens": 27138282.0, + "step": 1051 + }, + { + "epoch": 0.11552822314957172, + "grad_norm": 2.502500057220459, + "learning_rate": 5.768386388583974e-07, + "loss": 1.0048, + "mean_token_accuracy": 0.7011315822601318, + "num_tokens": 27163510.0, + "step": 1052 + }, + { + "epoch": 0.11563804085218538, + "grad_norm": 2.428762197494507, + "learning_rate": 5.773874862788145e-07, + "loss": 1.1722, + "mean_token_accuracy": 0.6598687767982483, + "num_tokens": 27189510.0, + "step": 1053 + }, + { + "epoch": 0.11574785855479904, + "grad_norm": 2.3028981685638428, + "learning_rate": 5.779363336992316e-07, + "loss": 1.1099, + "mean_token_accuracy": 0.6644243001937866, + "num_tokens": 27219007.0, + "step": 1054 + }, + { + "epoch": 0.1158576762574127, + "grad_norm": 2.1013190746307373, + "learning_rate": 5.784851811196487e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6693684458732605, + "num_tokens": 27251763.0, + "step": 1055 + }, + { + "epoch": 0.11596749396002635, + "grad_norm": 1.9702305793762207, + "learning_rate": 5.790340285400658e-07, + "loss": 1.0727, + "mean_token_accuracy": 0.6871154308319092, + "num_tokens": 27285673.0, + "step": 1056 + }, + { + "epoch": 0.11607731166264001, + "grad_norm": 2.5482866764068604, + "learning_rate": 5.795828759604829e-07, + "loss": 1.1895, + "mean_token_accuracy": 0.6551116704940796, + "num_tokens": 27308013.0, + "step": 1057 + }, + { + "epoch": 0.11618712936525367, + "grad_norm": 2.3413689136505127, + "learning_rate": 5.801317233809001e-07, + "loss": 1.1065, + "mean_token_accuracy": 0.6867877244949341, + "num_tokens": 27333653.0, + "step": 1058 + }, + { + "epoch": 0.11629694706786733, + "grad_norm": 2.408393144607544, + "learning_rate": 5.806805708013171e-07, + "loss": 1.0904, + "mean_token_accuracy": 0.6789568662643433, + "num_tokens": 27359747.0, + "step": 1059 + }, + { + "epoch": 0.11640676477048101, + "grad_norm": 2.3890299797058105, + "learning_rate": 5.812294182217343e-07, + "loss": 1.0507, + "mean_token_accuracy": 0.6893719434738159, + "num_tokens": 27385857.0, + "step": 1060 + }, + { + "epoch": 0.11651658247309467, + "grad_norm": 2.3074393272399902, + "learning_rate": 5.817782656421515e-07, + "loss": 1.062, + "mean_token_accuracy": 0.6875464916229248, + "num_tokens": 27410886.0, + "step": 1061 + }, + { + "epoch": 0.11662640017570833, + "grad_norm": 2.584667921066284, + "learning_rate": 5.823271130625685e-07, + "loss": 1.0466, + "mean_token_accuracy": 0.6903299689292908, + "num_tokens": 27433415.0, + "step": 1062 + }, + { + "epoch": 0.11673621787832199, + "grad_norm": 2.334160566329956, + "learning_rate": 5.828759604829857e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6941704750061035, + "num_tokens": 27459369.0, + "step": 1063 + }, + { + "epoch": 0.11684603558093565, + "grad_norm": 2.395864725112915, + "learning_rate": 5.834248079034029e-07, + "loss": 1.0644, + "mean_token_accuracy": 0.6861156225204468, + "num_tokens": 27484005.0, + "step": 1064 + }, + { + "epoch": 0.11695585328354931, + "grad_norm": 2.1608383655548096, + "learning_rate": 5.839736553238199e-07, + "loss": 1.1515, + "mean_token_accuracy": 0.6757978200912476, + "num_tokens": 27514945.0, + "step": 1065 + }, + { + "epoch": 0.11706567098616297, + "grad_norm": 2.110546588897705, + "learning_rate": 5.845225027442371e-07, + "loss": 1.1524, + "mean_token_accuracy": 0.6591464877128601, + "num_tokens": 27545050.0, + "step": 1066 + }, + { + "epoch": 0.11717548868877663, + "grad_norm": 2.455357551574707, + "learning_rate": 5.850713501646543e-07, + "loss": 1.0067, + "mean_token_accuracy": 0.6973764896392822, + "num_tokens": 27569625.0, + "step": 1067 + }, + { + "epoch": 0.11728530639139029, + "grad_norm": 2.1243340969085693, + "learning_rate": 5.856201975850713e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.6796094179153442, + "num_tokens": 27599131.0, + "step": 1068 + }, + { + "epoch": 0.11739512409400395, + "grad_norm": 2.0603599548339844, + "learning_rate": 5.861690450054885e-07, + "loss": 1.1595, + "mean_token_accuracy": 0.6577808856964111, + "num_tokens": 27631478.0, + "step": 1069 + }, + { + "epoch": 0.11750494179661762, + "grad_norm": 2.2686243057250977, + "learning_rate": 5.867178924259056e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6947687864303589, + "num_tokens": 27657677.0, + "step": 1070 + }, + { + "epoch": 0.11761475949923128, + "grad_norm": 2.6247572898864746, + "learning_rate": 5.872667398463227e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.695607602596283, + "num_tokens": 27678125.0, + "step": 1071 + }, + { + "epoch": 0.11772457720184494, + "grad_norm": 2.176896095275879, + "learning_rate": 5.878155872667399e-07, + "loss": 1.0926, + "mean_token_accuracy": 0.6772715449333191, + "num_tokens": 27706635.0, + "step": 1072 + }, + { + "epoch": 0.1178343949044586, + "grad_norm": 2.678057909011841, + "learning_rate": 5.88364434687157e-07, + "loss": 1.0453, + "mean_token_accuracy": 0.6887712478637695, + "num_tokens": 27727525.0, + "step": 1073 + }, + { + "epoch": 0.11794421260707226, + "grad_norm": 2.5706539154052734, + "learning_rate": 5.88913282107574e-07, + "loss": 1.1503, + "mean_token_accuracy": 0.6771278381347656, + "num_tokens": 27749969.0, + "step": 1074 + }, + { + "epoch": 0.11805403030968592, + "grad_norm": 2.2860183715820312, + "learning_rate": 5.894621295279912e-07, + "loss": 1.0374, + "mean_token_accuracy": 0.6907364726066589, + "num_tokens": 27777797.0, + "step": 1075 + }, + { + "epoch": 0.11816384801229958, + "grad_norm": 2.137749433517456, + "learning_rate": 5.900109769484083e-07, + "loss": 1.0633, + "mean_token_accuracy": 0.6809074878692627, + "num_tokens": 27805716.0, + "step": 1076 + }, + { + "epoch": 0.11827366571491324, + "grad_norm": 2.364269971847534, + "learning_rate": 5.905598243688254e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6790717840194702, + "num_tokens": 27831634.0, + "step": 1077 + }, + { + "epoch": 0.1183834834175269, + "grad_norm": 2.6127991676330566, + "learning_rate": 5.911086717892426e-07, + "loss": 1.1536, + "mean_token_accuracy": 0.6675060987472534, + "num_tokens": 27855048.0, + "step": 1078 + }, + { + "epoch": 0.11849330112014057, + "grad_norm": 2.452211618423462, + "learning_rate": 5.916575192096597e-07, + "loss": 1.1493, + "mean_token_accuracy": 0.6612519025802612, + "num_tokens": 27879921.0, + "step": 1079 + }, + { + "epoch": 0.11860311882275423, + "grad_norm": 2.0684263706207275, + "learning_rate": 5.922063666300768e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6765913963317871, + "num_tokens": 27912401.0, + "step": 1080 + }, + { + "epoch": 0.1187129365253679, + "grad_norm": 2.2896933555603027, + "learning_rate": 5.927552140504939e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.6984656453132629, + "num_tokens": 27937537.0, + "step": 1081 + }, + { + "epoch": 0.11882275422798155, + "grad_norm": 2.5896029472351074, + "learning_rate": 5.93304061470911e-07, + "loss": 1.0668, + "mean_token_accuracy": 0.680735170841217, + "num_tokens": 27960588.0, + "step": 1082 + }, + { + "epoch": 0.11893257193059521, + "grad_norm": 2.3826379776000977, + "learning_rate": 5.938529088913282e-07, + "loss": 1.0811, + "mean_token_accuracy": 0.6849210858345032, + "num_tokens": 27985675.0, + "step": 1083 + }, + { + "epoch": 0.11904238963320887, + "grad_norm": 2.2673678398132324, + "learning_rate": 5.944017563117453e-07, + "loss": 1.0953, + "mean_token_accuracy": 0.6777523159980774, + "num_tokens": 28013002.0, + "step": 1084 + }, + { + "epoch": 0.11915220733582253, + "grad_norm": 2.520960807800293, + "learning_rate": 5.949506037321624e-07, + "loss": 1.1436, + "mean_token_accuracy": 0.6721333265304565, + "num_tokens": 28037078.0, + "step": 1085 + }, + { + "epoch": 0.11926202503843619, + "grad_norm": 2.40736722946167, + "learning_rate": 5.954994511525796e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6822535395622253, + "num_tokens": 28062092.0, + "step": 1086 + }, + { + "epoch": 0.11937184274104985, + "grad_norm": 2.1927592754364014, + "learning_rate": 5.960482985729967e-07, + "loss": 1.1201, + "mean_token_accuracy": 0.6667667031288147, + "num_tokens": 28090139.0, + "step": 1087 + }, + { + "epoch": 0.11948166044366351, + "grad_norm": 2.2952561378479004, + "learning_rate": 5.965971459934138e-07, + "loss": 1.0879, + "mean_token_accuracy": 0.6744300127029419, + "num_tokens": 28116697.0, + "step": 1088 + }, + { + "epoch": 0.11959147814627719, + "grad_norm": 2.6065289974212646, + "learning_rate": 5.971459934138309e-07, + "loss": 0.9893, + "mean_token_accuracy": 0.7010440230369568, + "num_tokens": 28138325.0, + "step": 1089 + }, + { + "epoch": 0.11970129584889085, + "grad_norm": 2.2025277614593506, + "learning_rate": 5.976948408342481e-07, + "loss": 1.0834, + "mean_token_accuracy": 0.6779444813728333, + "num_tokens": 28167479.0, + "step": 1090 + }, + { + "epoch": 0.1198111135515045, + "grad_norm": 2.5210659503936768, + "learning_rate": 5.982436882546652e-07, + "loss": 0.9921, + "mean_token_accuracy": 0.6998966336250305, + "num_tokens": 28190354.0, + "step": 1091 + }, + { + "epoch": 0.11992093125411817, + "grad_norm": 2.2658751010894775, + "learning_rate": 5.987925356750822e-07, + "loss": 1.1256, + "mean_token_accuracy": 0.6682490706443787, + "num_tokens": 28219514.0, + "step": 1092 + }, + { + "epoch": 0.12003074895673183, + "grad_norm": 2.305911064147949, + "learning_rate": 5.993413830954994e-07, + "loss": 1.0228, + "mean_token_accuracy": 0.6976816058158875, + "num_tokens": 28245877.0, + "step": 1093 + }, + { + "epoch": 0.12014056665934549, + "grad_norm": 2.3149328231811523, + "learning_rate": 5.998902305159165e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.6916624903678894, + "num_tokens": 28269400.0, + "step": 1094 + }, + { + "epoch": 0.12025038436195915, + "grad_norm": 2.2133076190948486, + "learning_rate": 6.004390779363336e-07, + "loss": 1.0877, + "mean_token_accuracy": 0.6805165410041809, + "num_tokens": 28299347.0, + "step": 1095 + }, + { + "epoch": 0.1203602020645728, + "grad_norm": 2.534313917160034, + "learning_rate": 6.009879253567508e-07, + "loss": 1.0447, + "mean_token_accuracy": 0.6968567967414856, + "num_tokens": 28323100.0, + "step": 1096 + }, + { + "epoch": 0.12047001976718646, + "grad_norm": 2.3982033729553223, + "learning_rate": 6.015367727771679e-07, + "loss": 1.0705, + "mean_token_accuracy": 0.6943545341491699, + "num_tokens": 28347970.0, + "step": 1097 + }, + { + "epoch": 0.12057983746980014, + "grad_norm": 2.3428759574890137, + "learning_rate": 6.02085620197585e-07, + "loss": 1.0427, + "mean_token_accuracy": 0.6935408711433411, + "num_tokens": 28374735.0, + "step": 1098 + }, + { + "epoch": 0.1206896551724138, + "grad_norm": 2.1995983123779297, + "learning_rate": 6.026344676180022e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6779260635375977, + "num_tokens": 28402989.0, + "step": 1099 + }, + { + "epoch": 0.12079947287502746, + "grad_norm": 2.283501625061035, + "learning_rate": 6.031833150384192e-07, + "loss": 1.0663, + "mean_token_accuracy": 0.6894546747207642, + "num_tokens": 28429225.0, + "step": 1100 + }, + { + "epoch": 0.12090929057764112, + "grad_norm": 2.3859684467315674, + "learning_rate": 6.037321624588364e-07, + "loss": 1.1362, + "mean_token_accuracy": 0.6700706481933594, + "num_tokens": 28454989.0, + "step": 1101 + }, + { + "epoch": 0.12101910828025478, + "grad_norm": 2.3037478923797607, + "learning_rate": 6.042810098792536e-07, + "loss": 1.0779, + "mean_token_accuracy": 0.6828193068504333, + "num_tokens": 28481627.0, + "step": 1102 + }, + { + "epoch": 0.12112892598286844, + "grad_norm": 2.698392391204834, + "learning_rate": 6.048298572996706e-07, + "loss": 0.9978, + "mean_token_accuracy": 0.7004851698875427, + "num_tokens": 28501702.0, + "step": 1103 + }, + { + "epoch": 0.1212387436854821, + "grad_norm": 2.41110897064209, + "learning_rate": 6.053787047200878e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.6911273002624512, + "num_tokens": 28525170.0, + "step": 1104 + }, + { + "epoch": 0.12134856138809576, + "grad_norm": 2.5193352699279785, + "learning_rate": 6.05927552140505e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.7056028842926025, + "num_tokens": 28546203.0, + "step": 1105 + }, + { + "epoch": 0.12145837909070942, + "grad_norm": 2.3023948669433594, + "learning_rate": 6.06476399560922e-07, + "loss": 1.046, + "mean_token_accuracy": 0.6880849003791809, + "num_tokens": 28572280.0, + "step": 1106 + }, + { + "epoch": 0.12156819679332308, + "grad_norm": 2.5612564086914062, + "learning_rate": 6.070252469813392e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6809903383255005, + "num_tokens": 28595107.0, + "step": 1107 + }, + { + "epoch": 0.12167801449593675, + "grad_norm": 2.3941617012023926, + "learning_rate": 6.075740944017564e-07, + "loss": 1.1, + "mean_token_accuracy": 0.683085024356842, + "num_tokens": 28622284.0, + "step": 1108 + }, + { + "epoch": 0.12178783219855041, + "grad_norm": 2.3073930740356445, + "learning_rate": 6.081229418221734e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6858001947402954, + "num_tokens": 28647760.0, + "step": 1109 + }, + { + "epoch": 0.12189764990116407, + "grad_norm": 2.4685912132263184, + "learning_rate": 6.086717892425906e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6851031184196472, + "num_tokens": 28674115.0, + "step": 1110 + }, + { + "epoch": 0.12200746760377773, + "grad_norm": 2.215299367904663, + "learning_rate": 6.092206366630076e-07, + "loss": 1.002, + "mean_token_accuracy": 0.6983633041381836, + "num_tokens": 28699317.0, + "step": 1111 + }, + { + "epoch": 0.12211728530639139, + "grad_norm": 2.441901683807373, + "learning_rate": 6.097694840834247e-07, + "loss": 1.0118, + "mean_token_accuracy": 0.6928917169570923, + "num_tokens": 28722523.0, + "step": 1112 + }, + { + "epoch": 0.12222710300900505, + "grad_norm": 2.20092511177063, + "learning_rate": 6.103183315038419e-07, + "loss": 1.037, + "mean_token_accuracy": 0.702886700630188, + "num_tokens": 28748640.0, + "step": 1113 + }, + { + "epoch": 0.12233692071161871, + "grad_norm": 2.351240634918213, + "learning_rate": 6.10867178924259e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.676638126373291, + "num_tokens": 28774868.0, + "step": 1114 + }, + { + "epoch": 0.12244673841423237, + "grad_norm": 2.5815701484680176, + "learning_rate": 6.114160263446761e-07, + "loss": 0.9882, + "mean_token_accuracy": 0.7050041556358337, + "num_tokens": 28795707.0, + "step": 1115 + }, + { + "epoch": 0.12255655611684603, + "grad_norm": 2.2287230491638184, + "learning_rate": 6.119648737650933e-07, + "loss": 1.0264, + "mean_token_accuracy": 0.6962502002716064, + "num_tokens": 28822307.0, + "step": 1116 + }, + { + "epoch": 0.1226663738194597, + "grad_norm": 2.450183153152466, + "learning_rate": 6.125137211855103e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.6931032538414001, + "num_tokens": 28845463.0, + "step": 1117 + }, + { + "epoch": 0.12277619152207336, + "grad_norm": 2.6760497093200684, + "learning_rate": 6.130625686059275e-07, + "loss": 1.1198, + "mean_token_accuracy": 0.668645441532135, + "num_tokens": 28866996.0, + "step": 1118 + }, + { + "epoch": 0.12288600922468702, + "grad_norm": 2.242969274520874, + "learning_rate": 6.136114160263447e-07, + "loss": 0.9903, + "mean_token_accuracy": 0.7021206617355347, + "num_tokens": 28891739.0, + "step": 1119 + }, + { + "epoch": 0.12299582692730068, + "grad_norm": 2.2325801849365234, + "learning_rate": 6.141602634467617e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6941454410552979, + "num_tokens": 28918155.0, + "step": 1120 + }, + { + "epoch": 0.12310564462991434, + "grad_norm": 2.1416993141174316, + "learning_rate": 6.147091108671789e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6789547204971313, + "num_tokens": 28945978.0, + "step": 1121 + }, + { + "epoch": 0.123215462332528, + "grad_norm": 2.4764280319213867, + "learning_rate": 6.15257958287596e-07, + "loss": 1.0329, + "mean_token_accuracy": 0.6896899342536926, + "num_tokens": 28969561.0, + "step": 1122 + }, + { + "epoch": 0.12332528003514166, + "grad_norm": 2.3898801803588867, + "learning_rate": 6.158068057080131e-07, + "loss": 1.0617, + "mean_token_accuracy": 0.6759624481201172, + "num_tokens": 28995780.0, + "step": 1123 + }, + { + "epoch": 0.12343509773775532, + "grad_norm": 2.381911039352417, + "learning_rate": 6.163556531284303e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6846404075622559, + "num_tokens": 29020518.0, + "step": 1124 + }, + { + "epoch": 0.12354491544036898, + "grad_norm": 2.2284412384033203, + "learning_rate": 6.169045005488474e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6858395934104919, + "num_tokens": 29047809.0, + "step": 1125 + }, + { + "epoch": 0.12365473314298264, + "grad_norm": 2.3830409049987793, + "learning_rate": 6.174533479692645e-07, + "loss": 0.9951, + "mean_token_accuracy": 0.7051640748977661, + "num_tokens": 29070861.0, + "step": 1126 + }, + { + "epoch": 0.12376455084559632, + "grad_norm": 2.277287483215332, + "learning_rate": 6.180021953896817e-07, + "loss": 1.0691, + "mean_token_accuracy": 0.6818389892578125, + "num_tokens": 29096873.0, + "step": 1127 + }, + { + "epoch": 0.12387436854820998, + "grad_norm": 2.1370158195495605, + "learning_rate": 6.185510428100988e-07, + "loss": 1.1359, + "mean_token_accuracy": 0.6602637767791748, + "num_tokens": 29126710.0, + "step": 1128 + }, + { + "epoch": 0.12398418625082364, + "grad_norm": 2.1369001865386963, + "learning_rate": 6.190998902305159e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.6924254894256592, + "num_tokens": 29154682.0, + "step": 1129 + }, + { + "epoch": 0.1240940039534373, + "grad_norm": 2.339015007019043, + "learning_rate": 6.196487376509331e-07, + "loss": 1.021, + "mean_token_accuracy": 0.6941938400268555, + "num_tokens": 29177695.0, + "step": 1130 + }, + { + "epoch": 0.12420382165605096, + "grad_norm": 2.3048150539398193, + "learning_rate": 6.201975850713501e-07, + "loss": 1.1175, + "mean_token_accuracy": 0.6656689643859863, + "num_tokens": 29203041.0, + "step": 1131 + }, + { + "epoch": 0.12431363935866462, + "grad_norm": 2.531092882156372, + "learning_rate": 6.207464324917672e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6841883063316345, + "num_tokens": 29228325.0, + "step": 1132 + }, + { + "epoch": 0.12442345706127828, + "grad_norm": 2.3770804405212402, + "learning_rate": 6.212952799121843e-07, + "loss": 1.0534, + "mean_token_accuracy": 0.6856409311294556, + "num_tokens": 29252274.0, + "step": 1133 + }, + { + "epoch": 0.12453327476389194, + "grad_norm": 2.4369382858276367, + "learning_rate": 6.218441273326015e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6897122263908386, + "num_tokens": 29276631.0, + "step": 1134 + }, + { + "epoch": 0.1246430924665056, + "grad_norm": 2.383995771408081, + "learning_rate": 6.223929747530186e-07, + "loss": 1.0716, + "mean_token_accuracy": 0.6798090934753418, + "num_tokens": 29302086.0, + "step": 1135 + }, + { + "epoch": 0.12475291016911927, + "grad_norm": 2.4416308403015137, + "learning_rate": 6.229418221734357e-07, + "loss": 0.9868, + "mean_token_accuracy": 0.6996421813964844, + "num_tokens": 29323877.0, + "step": 1136 + }, + { + "epoch": 0.12486272787173293, + "grad_norm": 2.4521613121032715, + "learning_rate": 6.234906695938529e-07, + "loss": 1.1128, + "mean_token_accuracy": 0.6714131832122803, + "num_tokens": 29348237.0, + "step": 1137 + }, + { + "epoch": 0.12497254557434659, + "grad_norm": 2.500068187713623, + "learning_rate": 6.2403951701427e-07, + "loss": 1.0436, + "mean_token_accuracy": 0.6941812634468079, + "num_tokens": 29371380.0, + "step": 1138 + }, + { + "epoch": 0.12508236327696023, + "grad_norm": 2.2965140342712402, + "learning_rate": 6.245883644346871e-07, + "loss": 1.0016, + "mean_token_accuracy": 0.6978554725646973, + "num_tokens": 29397485.0, + "step": 1139 + }, + { + "epoch": 0.1251921809795739, + "grad_norm": 2.1807281970977783, + "learning_rate": 6.251372118551043e-07, + "loss": 1.0295, + "mean_token_accuracy": 0.6948522329330444, + "num_tokens": 29424259.0, + "step": 1140 + }, + { + "epoch": 0.12530199868218758, + "grad_norm": 2.168221950531006, + "learning_rate": 6.256860592755214e-07, + "loss": 1.119, + "mean_token_accuracy": 0.667301595211029, + "num_tokens": 29452414.0, + "step": 1141 + }, + { + "epoch": 0.12541181638480123, + "grad_norm": 2.620177745819092, + "learning_rate": 6.262349066959385e-07, + "loss": 1.1076, + "mean_token_accuracy": 0.6796584129333496, + "num_tokens": 29475305.0, + "step": 1142 + }, + { + "epoch": 0.1255216340874149, + "grad_norm": 2.400400161743164, + "learning_rate": 6.267837541163557e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6824926137924194, + "num_tokens": 29500626.0, + "step": 1143 + }, + { + "epoch": 0.12563145179002855, + "grad_norm": 2.4798882007598877, + "learning_rate": 6.273326015367727e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6895133256912231, + "num_tokens": 29522970.0, + "step": 1144 + }, + { + "epoch": 0.12574126949264222, + "grad_norm": 2.515589952468872, + "learning_rate": 6.278814489571899e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6906593441963196, + "num_tokens": 29545052.0, + "step": 1145 + }, + { + "epoch": 0.12585108719525587, + "grad_norm": 2.330944776535034, + "learning_rate": 6.284302963776071e-07, + "loss": 1.1208, + "mean_token_accuracy": 0.6730299592018127, + "num_tokens": 29570471.0, + "step": 1146 + }, + { + "epoch": 0.12596090489786954, + "grad_norm": 2.3126425743103027, + "learning_rate": 6.289791437980241e-07, + "loss": 0.9155, + "mean_token_accuracy": 0.7259045839309692, + "num_tokens": 29597201.0, + "step": 1147 + }, + { + "epoch": 0.1260707226004832, + "grad_norm": 2.3430161476135254, + "learning_rate": 6.295279912184413e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6792085766792297, + "num_tokens": 29621188.0, + "step": 1148 + }, + { + "epoch": 0.12618054030309686, + "grad_norm": 2.8498611450195312, + "learning_rate": 6.300768386388585e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.6902414560317993, + "num_tokens": 29641182.0, + "step": 1149 + }, + { + "epoch": 0.1262903580057105, + "grad_norm": 1.8806636333465576, + "learning_rate": 6.306256860592754e-07, + "loss": 1.003, + "mean_token_accuracy": 0.6977424621582031, + "num_tokens": 29676842.0, + "step": 1150 + }, + { + "epoch": 0.12640017570832418, + "grad_norm": 2.194040060043335, + "learning_rate": 6.311745334796926e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6712744235992432, + "num_tokens": 29706524.0, + "step": 1151 + }, + { + "epoch": 0.12650999341093785, + "grad_norm": 2.383185625076294, + "learning_rate": 6.317233809001098e-07, + "loss": 1.0438, + "mean_token_accuracy": 0.6997005939483643, + "num_tokens": 29731835.0, + "step": 1152 + }, + { + "epoch": 0.1266198111135515, + "grad_norm": 2.215147018432617, + "learning_rate": 6.322722283205268e-07, + "loss": 1.1154, + "mean_token_accuracy": 0.6706527471542358, + "num_tokens": 29759624.0, + "step": 1153 + }, + { + "epoch": 0.12672962881616517, + "grad_norm": 2.4883413314819336, + "learning_rate": 6.32821075740944e-07, + "loss": 0.9978, + "mean_token_accuracy": 0.6977500915527344, + "num_tokens": 29784258.0, + "step": 1154 + }, + { + "epoch": 0.12683944651877882, + "grad_norm": 2.1408605575561523, + "learning_rate": 6.33369923161361e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6890366673469543, + "num_tokens": 29813072.0, + "step": 1155 + }, + { + "epoch": 0.1269492642213925, + "grad_norm": 2.1293280124664307, + "learning_rate": 6.339187705817782e-07, + "loss": 1.0485, + "mean_token_accuracy": 0.691077709197998, + "num_tokens": 29841848.0, + "step": 1156 + }, + { + "epoch": 0.12705908192400614, + "grad_norm": 2.2744455337524414, + "learning_rate": 6.344676180021954e-07, + "loss": 0.9419, + "mean_token_accuracy": 0.715090274810791, + "num_tokens": 29866779.0, + "step": 1157 + }, + { + "epoch": 0.1271688996266198, + "grad_norm": 2.382485866546631, + "learning_rate": 6.350164654226124e-07, + "loss": 1.0959, + "mean_token_accuracy": 0.6752576231956482, + "num_tokens": 29891420.0, + "step": 1158 + }, + { + "epoch": 0.12727871732923346, + "grad_norm": 2.3463714122772217, + "learning_rate": 6.355653128430296e-07, + "loss": 1.033, + "mean_token_accuracy": 0.6903937458992004, + "num_tokens": 29915290.0, + "step": 1159 + }, + { + "epoch": 0.12738853503184713, + "grad_norm": 2.260510206222534, + "learning_rate": 6.361141602634468e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.6921989321708679, + "num_tokens": 29941352.0, + "step": 1160 + }, + { + "epoch": 0.1274983527344608, + "grad_norm": 2.133344888687134, + "learning_rate": 6.366630076838638e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7020260095596313, + "num_tokens": 29969379.0, + "step": 1161 + }, + { + "epoch": 0.12760817043707445, + "grad_norm": 2.2887308597564697, + "learning_rate": 6.37211855104281e-07, + "loss": 1.1003, + "mean_token_accuracy": 0.6679004430770874, + "num_tokens": 29997129.0, + "step": 1162 + }, + { + "epoch": 0.12771798813968813, + "grad_norm": 2.2983181476593018, + "learning_rate": 6.377607025246982e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.6766012907028198, + "num_tokens": 30023175.0, + "step": 1163 + }, + { + "epoch": 0.12782780584230177, + "grad_norm": 2.201188564300537, + "learning_rate": 6.383095499451152e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6889074444770813, + "num_tokens": 30052859.0, + "step": 1164 + }, + { + "epoch": 0.12793762354491545, + "grad_norm": 2.2210593223571777, + "learning_rate": 6.388583973655324e-07, + "loss": 1.0567, + "mean_token_accuracy": 0.6889038681983948, + "num_tokens": 30082284.0, + "step": 1165 + }, + { + "epoch": 0.1280474412475291, + "grad_norm": 2.312206745147705, + "learning_rate": 6.394072447859495e-07, + "loss": 1.0442, + "mean_token_accuracy": 0.686639130115509, + "num_tokens": 30107279.0, + "step": 1166 + }, + { + "epoch": 0.12815725895014277, + "grad_norm": 2.174410581588745, + "learning_rate": 6.399560922063666e-07, + "loss": 0.9615, + "mean_token_accuracy": 0.7108774185180664, + "num_tokens": 30134521.0, + "step": 1167 + }, + { + "epoch": 0.1282670766527564, + "grad_norm": 2.3407180309295654, + "learning_rate": 6.405049396267838e-07, + "loss": 1.0042, + "mean_token_accuracy": 0.7029858827590942, + "num_tokens": 30160609.0, + "step": 1168 + }, + { + "epoch": 0.12837689435537009, + "grad_norm": 2.4217731952667236, + "learning_rate": 6.410537870472008e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.691474199295044, + "num_tokens": 30183959.0, + "step": 1169 + }, + { + "epoch": 0.12848671205798376, + "grad_norm": 2.7318079471588135, + "learning_rate": 6.416026344676179e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6820782423019409, + "num_tokens": 30204192.0, + "step": 1170 + }, + { + "epoch": 0.1285965297605974, + "grad_norm": 2.140798568725586, + "learning_rate": 6.421514818880351e-07, + "loss": 1.0039, + "mean_token_accuracy": 0.7028688192367554, + "num_tokens": 30232658.0, + "step": 1171 + }, + { + "epoch": 0.12870634746321108, + "grad_norm": 2.4022409915924072, + "learning_rate": 6.427003293084522e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.691609263420105, + "num_tokens": 30255365.0, + "step": 1172 + }, + { + "epoch": 0.12881616516582473, + "grad_norm": 2.3015003204345703, + "learning_rate": 6.432491767288693e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6848340630531311, + "num_tokens": 30281952.0, + "step": 1173 + }, + { + "epoch": 0.1289259828684384, + "grad_norm": 2.0095314979553223, + "learning_rate": 6.437980241492865e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6776495575904846, + "num_tokens": 30313618.0, + "step": 1174 + }, + { + "epoch": 0.12903580057105205, + "grad_norm": 2.499014139175415, + "learning_rate": 6.443468715697036e-07, + "loss": 1.0501, + "mean_token_accuracy": 0.6972795724868774, + "num_tokens": 30337050.0, + "step": 1175 + }, + { + "epoch": 0.12914561827366572, + "grad_norm": 2.2999496459960938, + "learning_rate": 6.448957189901207e-07, + "loss": 1.0523, + "mean_token_accuracy": 0.6998319625854492, + "num_tokens": 30361944.0, + "step": 1176 + }, + { + "epoch": 0.12925543597627936, + "grad_norm": 2.430856704711914, + "learning_rate": 6.454445664105378e-07, + "loss": 1.0792, + "mean_token_accuracy": 0.6858905553817749, + "num_tokens": 30387153.0, + "step": 1177 + }, + { + "epoch": 0.12936525367889304, + "grad_norm": 2.4938833713531494, + "learning_rate": 6.45993413830955e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.6989525556564331, + "num_tokens": 30409379.0, + "step": 1178 + }, + { + "epoch": 0.1294750713815067, + "grad_norm": 2.708317518234253, + "learning_rate": 6.465422612513721e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6877904534339905, + "num_tokens": 30430006.0, + "step": 1179 + }, + { + "epoch": 0.12958488908412036, + "grad_norm": 2.501291513442993, + "learning_rate": 6.470911086717892e-07, + "loss": 1.0794, + "mean_token_accuracy": 0.6817722320556641, + "num_tokens": 30452872.0, + "step": 1180 + }, + { + "epoch": 0.12969470678673403, + "grad_norm": 2.107937812805176, + "learning_rate": 6.476399560922064e-07, + "loss": 1.1534, + "mean_token_accuracy": 0.6618674993515015, + "num_tokens": 30484946.0, + "step": 1181 + }, + { + "epoch": 0.12980452448934768, + "grad_norm": 2.6968369483947754, + "learning_rate": 6.481888035126235e-07, + "loss": 1.0125, + "mean_token_accuracy": 0.7031172513961792, + "num_tokens": 30505796.0, + "step": 1182 + }, + { + "epoch": 0.12991434219196135, + "grad_norm": 2.458834648132324, + "learning_rate": 6.487376509330406e-07, + "loss": 0.9909, + "mean_token_accuracy": 0.7075953483581543, + "num_tokens": 30528725.0, + "step": 1183 + }, + { + "epoch": 0.130024159894575, + "grad_norm": 2.171175956726074, + "learning_rate": 6.492864983534578e-07, + "loss": 1.1624, + "mean_token_accuracy": 0.6682311296463013, + "num_tokens": 30559139.0, + "step": 1184 + }, + { + "epoch": 0.13013397759718867, + "grad_norm": 2.235226631164551, + "learning_rate": 6.498353457738749e-07, + "loss": 1.0054, + "mean_token_accuracy": 0.6976103782653809, + "num_tokens": 30584350.0, + "step": 1185 + }, + { + "epoch": 0.13024379529980232, + "grad_norm": 2.122676134109497, + "learning_rate": 6.50384193194292e-07, + "loss": 1.0473, + "mean_token_accuracy": 0.6944661140441895, + "num_tokens": 30613289.0, + "step": 1186 + }, + { + "epoch": 0.130353613002416, + "grad_norm": 2.4987213611602783, + "learning_rate": 6.509330406147092e-07, + "loss": 1.0477, + "mean_token_accuracy": 0.6841434240341187, + "num_tokens": 30637415.0, + "step": 1187 + }, + { + "epoch": 0.13046343070502964, + "grad_norm": 3.203216314315796, + "learning_rate": 6.514818880351261e-07, + "loss": 0.9336, + "mean_token_accuracy": 0.716621458530426, + "num_tokens": 30652950.0, + "step": 1188 + }, + { + "epoch": 0.1305732484076433, + "grad_norm": 2.6410739421844482, + "learning_rate": 6.520307354555433e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6767250299453735, + "num_tokens": 30674347.0, + "step": 1189 + }, + { + "epoch": 0.13068306611025698, + "grad_norm": 2.436249256134033, + "learning_rate": 6.525795828759604e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.6965001821517944, + "num_tokens": 30696884.0, + "step": 1190 + }, + { + "epoch": 0.13079288381287063, + "grad_norm": 2.452373504638672, + "learning_rate": 6.531284302963775e-07, + "loss": 1.0702, + "mean_token_accuracy": 0.6858769059181213, + "num_tokens": 30719915.0, + "step": 1191 + }, + { + "epoch": 0.1309027015154843, + "grad_norm": 2.468759059906006, + "learning_rate": 6.536772777167947e-07, + "loss": 0.9891, + "mean_token_accuracy": 0.7041344046592712, + "num_tokens": 30742732.0, + "step": 1192 + }, + { + "epoch": 0.13101251921809795, + "grad_norm": 2.3489763736724854, + "learning_rate": 6.542261251372118e-07, + "loss": 1.0578, + "mean_token_accuracy": 0.6933557987213135, + "num_tokens": 30767684.0, + "step": 1193 + }, + { + "epoch": 0.13112233692071162, + "grad_norm": 2.3935444355010986, + "learning_rate": 6.547749725576289e-07, + "loss": 1.0401, + "mean_token_accuracy": 0.6966735124588013, + "num_tokens": 30791975.0, + "step": 1194 + }, + { + "epoch": 0.13123215462332527, + "grad_norm": 2.3122034072875977, + "learning_rate": 6.553238199780461e-07, + "loss": 1.1112, + "mean_token_accuracy": 0.6793274879455566, + "num_tokens": 30819228.0, + "step": 1195 + }, + { + "epoch": 0.13134197232593894, + "grad_norm": 2.429879665374756, + "learning_rate": 6.558726673984632e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.696843683719635, + "num_tokens": 30843583.0, + "step": 1196 + }, + { + "epoch": 0.1314517900285526, + "grad_norm": 2.366037130355835, + "learning_rate": 6.564215148188803e-07, + "loss": 1.0489, + "mean_token_accuracy": 0.688827395439148, + "num_tokens": 30867803.0, + "step": 1197 + }, + { + "epoch": 0.13156160773116626, + "grad_norm": 2.4720118045806885, + "learning_rate": 6.569703622392975e-07, + "loss": 1.0339, + "mean_token_accuracy": 0.6869875192642212, + "num_tokens": 30891050.0, + "step": 1198 + }, + { + "epoch": 0.13167142543377994, + "grad_norm": 2.308208465576172, + "learning_rate": 6.575192096597145e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.7097751498222351, + "num_tokens": 30914576.0, + "step": 1199 + }, + { + "epoch": 0.13178124313639358, + "grad_norm": 2.439586877822876, + "learning_rate": 6.580680570801317e-07, + "loss": 1.0065, + "mean_token_accuracy": 0.6942379474639893, + "num_tokens": 30937629.0, + "step": 1200 + }, + { + "epoch": 0.13189106083900726, + "grad_norm": 2.1337947845458984, + "learning_rate": 6.586169045005489e-07, + "loss": 0.9819, + "mean_token_accuracy": 0.7025573253631592, + "num_tokens": 30964557.0, + "step": 1201 + }, + { + "epoch": 0.1320008785416209, + "grad_norm": 2.265439033508301, + "learning_rate": 6.591657519209659e-07, + "loss": 1.0122, + "mean_token_accuracy": 0.7059030532836914, + "num_tokens": 30989672.0, + "step": 1202 + }, + { + "epoch": 0.13211069624423458, + "grad_norm": 2.104048490524292, + "learning_rate": 6.597145993413831e-07, + "loss": 1.1276, + "mean_token_accuracy": 0.6651740074157715, + "num_tokens": 31019716.0, + "step": 1203 + }, + { + "epoch": 0.13222051394684822, + "grad_norm": 2.253207206726074, + "learning_rate": 6.602634467618003e-07, + "loss": 1.0292, + "mean_token_accuracy": 0.6929373741149902, + "num_tokens": 31047988.0, + "step": 1204 + }, + { + "epoch": 0.1323303316494619, + "grad_norm": 1.9199823141098022, + "learning_rate": 6.608122941822173e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.6771625280380249, + "num_tokens": 31085619.0, + "step": 1205 + }, + { + "epoch": 0.13244014935207554, + "grad_norm": 2.2944726943969727, + "learning_rate": 6.613611416026345e-07, + "loss": 1.0326, + "mean_token_accuracy": 0.6851361989974976, + "num_tokens": 31111597.0, + "step": 1206 + }, + { + "epoch": 0.13254996705468922, + "grad_norm": 2.2231528759002686, + "learning_rate": 6.619099890230515e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6829867362976074, + "num_tokens": 31139881.0, + "step": 1207 + }, + { + "epoch": 0.1326597847573029, + "grad_norm": 2.2201085090637207, + "learning_rate": 6.624588364434686e-07, + "loss": 1.1073, + "mean_token_accuracy": 0.6775826215744019, + "num_tokens": 31168364.0, + "step": 1208 + }, + { + "epoch": 0.13276960245991654, + "grad_norm": 2.4737231731414795, + "learning_rate": 6.630076838638858e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.6838774681091309, + "num_tokens": 31191975.0, + "step": 1209 + }, + { + "epoch": 0.1328794201625302, + "grad_norm": 2.130108118057251, + "learning_rate": 6.635565312843029e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.6909551024436951, + "num_tokens": 31220226.0, + "step": 1210 + }, + { + "epoch": 0.13298923786514386, + "grad_norm": 2.04752779006958, + "learning_rate": 6.6410537870472e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6815102696418762, + "num_tokens": 31252627.0, + "step": 1211 + }, + { + "epoch": 0.13309905556775753, + "grad_norm": 2.068401336669922, + "learning_rate": 6.646542261251372e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6785667538642883, + "num_tokens": 31284636.0, + "step": 1212 + }, + { + "epoch": 0.13320887327037118, + "grad_norm": 2.2947003841400146, + "learning_rate": 6.652030735455543e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.6979983448982239, + "num_tokens": 31310821.0, + "step": 1213 + }, + { + "epoch": 0.13331869097298485, + "grad_norm": 2.263803720474243, + "learning_rate": 6.657519209659714e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6866909265518188, + "num_tokens": 31337065.0, + "step": 1214 + }, + { + "epoch": 0.1334285086755985, + "grad_norm": 2.5900003910064697, + "learning_rate": 6.663007683863886e-07, + "loss": 1.0311, + "mean_token_accuracy": 0.6976968050003052, + "num_tokens": 31358761.0, + "step": 1215 + }, + { + "epoch": 0.13353832637821217, + "grad_norm": 2.556107997894287, + "learning_rate": 6.668496158068057e-07, + "loss": 1.0701, + "mean_token_accuracy": 0.6820998191833496, + "num_tokens": 31381894.0, + "step": 1216 + }, + { + "epoch": 0.13364814408082584, + "grad_norm": 2.335381269454956, + "learning_rate": 6.673984632272228e-07, + "loss": 0.9753, + "mean_token_accuracy": 0.6998112201690674, + "num_tokens": 31405768.0, + "step": 1217 + }, + { + "epoch": 0.1337579617834395, + "grad_norm": 2.420207977294922, + "learning_rate": 6.679473106476399e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7013595104217529, + "num_tokens": 31428816.0, + "step": 1218 + }, + { + "epoch": 0.13386777948605316, + "grad_norm": 2.165796995162964, + "learning_rate": 6.684961580680571e-07, + "loss": 1.0656, + "mean_token_accuracy": 0.6772662401199341, + "num_tokens": 31457346.0, + "step": 1219 + }, + { + "epoch": 0.1339775971886668, + "grad_norm": 2.2077252864837646, + "learning_rate": 6.690450054884742e-07, + "loss": 0.962, + "mean_token_accuracy": 0.7110503315925598, + "num_tokens": 31485649.0, + "step": 1220 + }, + { + "epoch": 0.13408741489128048, + "grad_norm": 2.259345531463623, + "learning_rate": 6.695938529088913e-07, + "loss": 1.0301, + "mean_token_accuracy": 0.6897085309028625, + "num_tokens": 31511905.0, + "step": 1221 + }, + { + "epoch": 0.13419723259389413, + "grad_norm": 2.489964008331299, + "learning_rate": 6.701427003293085e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6938719153404236, + "num_tokens": 31535648.0, + "step": 1222 + }, + { + "epoch": 0.1343070502965078, + "grad_norm": 2.1358847618103027, + "learning_rate": 6.706915477497256e-07, + "loss": 0.9852, + "mean_token_accuracy": 0.7016880512237549, + "num_tokens": 31565741.0, + "step": 1223 + }, + { + "epoch": 0.13441686799912145, + "grad_norm": 2.487318515777588, + "learning_rate": 6.712403951701427e-07, + "loss": 1.1402, + "mean_token_accuracy": 0.6660042405128479, + "num_tokens": 31589617.0, + "step": 1224 + }, + { + "epoch": 0.13452668570173512, + "grad_norm": 2.191786289215088, + "learning_rate": 6.717892425905599e-07, + "loss": 1.1356, + "mean_token_accuracy": 0.6612992286682129, + "num_tokens": 31618759.0, + "step": 1225 + }, + { + "epoch": 0.13463650340434877, + "grad_norm": 2.6822643280029297, + "learning_rate": 6.723380900109769e-07, + "loss": 0.9236, + "mean_token_accuracy": 0.7246822118759155, + "num_tokens": 31637603.0, + "step": 1226 + }, + { + "epoch": 0.13474632110696244, + "grad_norm": 2.6717190742492676, + "learning_rate": 6.72886937431394e-07, + "loss": 1.0519, + "mean_token_accuracy": 0.6911882162094116, + "num_tokens": 31660391.0, + "step": 1227 + }, + { + "epoch": 0.13485613880957611, + "grad_norm": 2.056469202041626, + "learning_rate": 6.734357848518111e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6886296272277832, + "num_tokens": 31690392.0, + "step": 1228 + }, + { + "epoch": 0.13496595651218976, + "grad_norm": 2.207240581512451, + "learning_rate": 6.739846322722282e-07, + "loss": 1.0368, + "mean_token_accuracy": 0.6843494176864624, + "num_tokens": 31720242.0, + "step": 1229 + }, + { + "epoch": 0.13507577421480343, + "grad_norm": 2.277027130126953, + "learning_rate": 6.745334796926454e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.6816381216049194, + "num_tokens": 31746011.0, + "step": 1230 + }, + { + "epoch": 0.13518559191741708, + "grad_norm": 2.0630011558532715, + "learning_rate": 6.750823271130625e-07, + "loss": 1.1196, + "mean_token_accuracy": 0.6655032634735107, + "num_tokens": 31777825.0, + "step": 1231 + }, + { + "epoch": 0.13529540962003075, + "grad_norm": 2.032392740249634, + "learning_rate": 6.756311745334796e-07, + "loss": 1.0442, + "mean_token_accuracy": 0.6878349781036377, + "num_tokens": 31808659.0, + "step": 1232 + }, + { + "epoch": 0.1354052273226444, + "grad_norm": 2.410310745239258, + "learning_rate": 6.761800219538968e-07, + "loss": 1.1149, + "mean_token_accuracy": 0.6782659292221069, + "num_tokens": 31832936.0, + "step": 1233 + }, + { + "epoch": 0.13551504502525807, + "grad_norm": 2.1796000003814697, + "learning_rate": 6.767288693743139e-07, + "loss": 1.0721, + "mean_token_accuracy": 0.6882843971252441, + "num_tokens": 31860537.0, + "step": 1234 + }, + { + "epoch": 0.13562486272787172, + "grad_norm": 2.3189499378204346, + "learning_rate": 6.77277716794731e-07, + "loss": 0.9674, + "mean_token_accuracy": 0.7126190066337585, + "num_tokens": 31885947.0, + "step": 1235 + }, + { + "epoch": 0.1357346804304854, + "grad_norm": 2.2598257064819336, + "learning_rate": 6.778265642151482e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6802015900611877, + "num_tokens": 31912741.0, + "step": 1236 + }, + { + "epoch": 0.13584449813309907, + "grad_norm": 2.174281120300293, + "learning_rate": 6.783754116355653e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.68333899974823, + "num_tokens": 31941759.0, + "step": 1237 + }, + { + "epoch": 0.1359543158357127, + "grad_norm": 2.268903970718384, + "learning_rate": 6.789242590559824e-07, + "loss": 1.1212, + "mean_token_accuracy": 0.6698650121688843, + "num_tokens": 31969130.0, + "step": 1238 + }, + { + "epoch": 0.1360641335383264, + "grad_norm": 2.4337093830108643, + "learning_rate": 6.794731064763996e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.6882074475288391, + "num_tokens": 31992700.0, + "step": 1239 + }, + { + "epoch": 0.13617395124094003, + "grad_norm": 2.351947784423828, + "learning_rate": 6.800219538968166e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.6959898471832275, + "num_tokens": 32016422.0, + "step": 1240 + }, + { + "epoch": 0.1362837689435537, + "grad_norm": 2.2075886726379395, + "learning_rate": 6.805708013172338e-07, + "loss": 0.9789, + "mean_token_accuracy": 0.7030123472213745, + "num_tokens": 32044022.0, + "step": 1241 + }, + { + "epoch": 0.13639358664616735, + "grad_norm": 2.57595157623291, + "learning_rate": 6.81119648737651e-07, + "loss": 1.0347, + "mean_token_accuracy": 0.6910072565078735, + "num_tokens": 32065051.0, + "step": 1242 + }, + { + "epoch": 0.13650340434878103, + "grad_norm": 2.2866411209106445, + "learning_rate": 6.81668496158068e-07, + "loss": 1.0517, + "mean_token_accuracy": 0.6833513975143433, + "num_tokens": 32092318.0, + "step": 1243 + }, + { + "epoch": 0.13661322205139467, + "grad_norm": 2.233417510986328, + "learning_rate": 6.822173435784852e-07, + "loss": 1.0382, + "mean_token_accuracy": 0.6895139217376709, + "num_tokens": 32118847.0, + "step": 1244 + }, + { + "epoch": 0.13672303975400835, + "grad_norm": 2.3548455238342285, + "learning_rate": 6.827661909989023e-07, + "loss": 1.0199, + "mean_token_accuracy": 0.6941131353378296, + "num_tokens": 32142427.0, + "step": 1245 + }, + { + "epoch": 0.13683285745662202, + "grad_norm": 2.3125264644622803, + "learning_rate": 6.833150384193193e-07, + "loss": 1.1292, + "mean_token_accuracy": 0.6802918910980225, + "num_tokens": 32171537.0, + "step": 1246 + }, + { + "epoch": 0.13694267515923567, + "grad_norm": 2.254636764526367, + "learning_rate": 6.838638858397365e-07, + "loss": 1.1044, + "mean_token_accuracy": 0.6760549545288086, + "num_tokens": 32200681.0, + "step": 1247 + }, + { + "epoch": 0.13705249286184934, + "grad_norm": 2.7911009788513184, + "learning_rate": 6.844127332601537e-07, + "loss": 0.9488, + "mean_token_accuracy": 0.710587739944458, + "num_tokens": 32219511.0, + "step": 1248 + }, + { + "epoch": 0.13716231056446299, + "grad_norm": 2.151371955871582, + "learning_rate": 6.849615806805707e-07, + "loss": 1.1502, + "mean_token_accuracy": 0.6700464487075806, + "num_tokens": 32252648.0, + "step": 1249 + }, + { + "epoch": 0.13727212826707666, + "grad_norm": 2.331972360610962, + "learning_rate": 6.855104281009879e-07, + "loss": 1.0002, + "mean_token_accuracy": 0.6977202296257019, + "num_tokens": 32280367.0, + "step": 1250 + }, + { + "epoch": 0.1373819459696903, + "grad_norm": 2.4369518756866455, + "learning_rate": 6.86059275521405e-07, + "loss": 1.0809, + "mean_token_accuracy": 0.6700700521469116, + "num_tokens": 32304565.0, + "step": 1251 + }, + { + "epoch": 0.13749176367230398, + "grad_norm": 2.3659002780914307, + "learning_rate": 6.866081229418221e-07, + "loss": 1.0705, + "mean_token_accuracy": 0.6824170351028442, + "num_tokens": 32329535.0, + "step": 1252 + }, + { + "epoch": 0.13760158137491763, + "grad_norm": 2.310889959335327, + "learning_rate": 6.871569703622393e-07, + "loss": 0.9415, + "mean_token_accuracy": 0.710905909538269, + "num_tokens": 32353133.0, + "step": 1253 + }, + { + "epoch": 0.1377113990775313, + "grad_norm": 2.075207233428955, + "learning_rate": 6.877058177826564e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6833997368812561, + "num_tokens": 32384697.0, + "step": 1254 + }, + { + "epoch": 0.13782121678014497, + "grad_norm": 2.324702262878418, + "learning_rate": 6.882546652030735e-07, + "loss": 0.9465, + "mean_token_accuracy": 0.7168238162994385, + "num_tokens": 32410867.0, + "step": 1255 + }, + { + "epoch": 0.13793103448275862, + "grad_norm": 2.222012758255005, + "learning_rate": 6.888035126234907e-07, + "loss": 1.0907, + "mean_token_accuracy": 0.6844081878662109, + "num_tokens": 32437715.0, + "step": 1256 + }, + { + "epoch": 0.1380408521853723, + "grad_norm": 1.93733549118042, + "learning_rate": 6.893523600439078e-07, + "loss": 1.047, + "mean_token_accuracy": 0.6884164810180664, + "num_tokens": 32470251.0, + "step": 1257 + }, + { + "epoch": 0.13815066988798594, + "grad_norm": 2.3892059326171875, + "learning_rate": 6.899012074643249e-07, + "loss": 1.0653, + "mean_token_accuracy": 0.6850738525390625, + "num_tokens": 32497181.0, + "step": 1258 + }, + { + "epoch": 0.1382604875905996, + "grad_norm": 2.388350248336792, + "learning_rate": 6.904500548847421e-07, + "loss": 1.0576, + "mean_token_accuracy": 0.6868467926979065, + "num_tokens": 32522125.0, + "step": 1259 + }, + { + "epoch": 0.13837030529321326, + "grad_norm": 2.3962793350219727, + "learning_rate": 6.909989023051592e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6660006642341614, + "num_tokens": 32546903.0, + "step": 1260 + }, + { + "epoch": 0.13848012299582693, + "grad_norm": 2.587024450302124, + "learning_rate": 6.915477497255763e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6788479089736938, + "num_tokens": 32569434.0, + "step": 1261 + }, + { + "epoch": 0.13858994069844058, + "grad_norm": 2.501140594482422, + "learning_rate": 6.920965971459934e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6736894845962524, + "num_tokens": 32593566.0, + "step": 1262 + }, + { + "epoch": 0.13869975840105425, + "grad_norm": 2.227433443069458, + "learning_rate": 6.926454445664105e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.6886022090911865, + "num_tokens": 32621452.0, + "step": 1263 + }, + { + "epoch": 0.1388095761036679, + "grad_norm": 2.4085729122161865, + "learning_rate": 6.931942919868276e-07, + "loss": 0.9878, + "mean_token_accuracy": 0.6988086104393005, + "num_tokens": 32644830.0, + "step": 1264 + }, + { + "epoch": 0.13891939380628157, + "grad_norm": 2.27949857711792, + "learning_rate": 6.937431394072447e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.6884022951126099, + "num_tokens": 32672924.0, + "step": 1265 + }, + { + "epoch": 0.13902921150889525, + "grad_norm": 2.3591647148132324, + "learning_rate": 6.942919868276618e-07, + "loss": 1.0084, + "mean_token_accuracy": 0.6984051465988159, + "num_tokens": 32698399.0, + "step": 1266 + }, + { + "epoch": 0.1391390292115089, + "grad_norm": 2.5082476139068604, + "learning_rate": 6.94840834248079e-07, + "loss": 1.0773, + "mean_token_accuracy": 0.6802060604095459, + "num_tokens": 32721535.0, + "step": 1267 + }, + { + "epoch": 0.13924884691412256, + "grad_norm": 2.1106083393096924, + "learning_rate": 6.953896816684961e-07, + "loss": 1.0643, + "mean_token_accuracy": 0.6875066757202148, + "num_tokens": 32751862.0, + "step": 1268 + }, + { + "epoch": 0.1393586646167362, + "grad_norm": 2.372363805770874, + "learning_rate": 6.959385290889132e-07, + "loss": 0.9295, + "mean_token_accuracy": 0.7188851833343506, + "num_tokens": 32775855.0, + "step": 1269 + }, + { + "epoch": 0.13946848231934988, + "grad_norm": 2.2342309951782227, + "learning_rate": 6.964873765093304e-07, + "loss": 1.1007, + "mean_token_accuracy": 0.6747185587882996, + "num_tokens": 32805194.0, + "step": 1270 + }, + { + "epoch": 0.13957830002196353, + "grad_norm": 2.310378074645996, + "learning_rate": 6.970362239297475e-07, + "loss": 0.9642, + "mean_token_accuracy": 0.7153271436691284, + "num_tokens": 32828906.0, + "step": 1271 + }, + { + "epoch": 0.1396881177245772, + "grad_norm": 2.306941270828247, + "learning_rate": 6.975850713501646e-07, + "loss": 1.093, + "mean_token_accuracy": 0.6819630861282349, + "num_tokens": 32856362.0, + "step": 1272 + }, + { + "epoch": 0.13979793542719085, + "grad_norm": 2.172541379928589, + "learning_rate": 6.981339187705817e-07, + "loss": 1.0418, + "mean_token_accuracy": 0.696723461151123, + "num_tokens": 32883972.0, + "step": 1273 + }, + { + "epoch": 0.13990775312980452, + "grad_norm": 1.9504097700119019, + "learning_rate": 6.986827661909989e-07, + "loss": 1.1053, + "mean_token_accuracy": 0.6678376793861389, + "num_tokens": 32920645.0, + "step": 1274 + }, + { + "epoch": 0.1400175708324182, + "grad_norm": 2.500145435333252, + "learning_rate": 6.99231613611416e-07, + "loss": 1.0289, + "mean_token_accuracy": 0.6968287229537964, + "num_tokens": 32942832.0, + "step": 1275 + }, + { + "epoch": 0.14012738853503184, + "grad_norm": 2.541275978088379, + "learning_rate": 6.997804610318331e-07, + "loss": 1.0599, + "mean_token_accuracy": 0.699080765247345, + "num_tokens": 32965523.0, + "step": 1276 + }, + { + "epoch": 0.14023720623764552, + "grad_norm": 2.2448461055755615, + "learning_rate": 7.003293084522503e-07, + "loss": 1.1381, + "mean_token_accuracy": 0.6663135290145874, + "num_tokens": 32993525.0, + "step": 1277 + }, + { + "epoch": 0.14034702394025916, + "grad_norm": 2.665855646133423, + "learning_rate": 7.008781558726674e-07, + "loss": 1.0498, + "mean_token_accuracy": 0.6830029487609863, + "num_tokens": 33012946.0, + "step": 1278 + }, + { + "epoch": 0.14045684164287284, + "grad_norm": 2.2594645023345947, + "learning_rate": 7.014270032930845e-07, + "loss": 1.0727, + "mean_token_accuracy": 0.681010365486145, + "num_tokens": 33040062.0, + "step": 1279 + }, + { + "epoch": 0.14056665934548648, + "grad_norm": 2.4598731994628906, + "learning_rate": 7.019758507135017e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6946839094161987, + "num_tokens": 33062058.0, + "step": 1280 + }, + { + "epoch": 0.14067647704810016, + "grad_norm": 2.2706971168518066, + "learning_rate": 7.025246981339188e-07, + "loss": 0.9945, + "mean_token_accuracy": 0.7043309211730957, + "num_tokens": 33086982.0, + "step": 1281 + }, + { + "epoch": 0.1407862947507138, + "grad_norm": 2.1959543228149414, + "learning_rate": 7.030735455543358e-07, + "loss": 0.9731, + "mean_token_accuracy": 0.7098414897918701, + "num_tokens": 33112153.0, + "step": 1282 + }, + { + "epoch": 0.14089611245332748, + "grad_norm": 2.433028221130371, + "learning_rate": 7.03622392974753e-07, + "loss": 1.0179, + "mean_token_accuracy": 0.6983472108840942, + "num_tokens": 33135050.0, + "step": 1283 + }, + { + "epoch": 0.14100593015594115, + "grad_norm": 2.3958330154418945, + "learning_rate": 7.0417124039517e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6995292901992798, + "num_tokens": 33158376.0, + "step": 1284 + }, + { + "epoch": 0.1411157478585548, + "grad_norm": 2.464420795440674, + "learning_rate": 7.047200878155872e-07, + "loss": 1.1015, + "mean_token_accuracy": 0.6846817135810852, + "num_tokens": 33180814.0, + "step": 1285 + }, + { + "epoch": 0.14122556556116847, + "grad_norm": 2.5547990798950195, + "learning_rate": 7.052689352360044e-07, + "loss": 0.9907, + "mean_token_accuracy": 0.6984452605247498, + "num_tokens": 33202903.0, + "step": 1286 + }, + { + "epoch": 0.14133538326378212, + "grad_norm": 2.264357089996338, + "learning_rate": 7.058177826564214e-07, + "loss": 1.0764, + "mean_token_accuracy": 0.690090000629425, + "num_tokens": 33229270.0, + "step": 1287 + }, + { + "epoch": 0.1414452009663958, + "grad_norm": 2.3580589294433594, + "learning_rate": 7.063666300768386e-07, + "loss": 1.1431, + "mean_token_accuracy": 0.6619696021080017, + "num_tokens": 33258139.0, + "step": 1288 + }, + { + "epoch": 0.14155501866900944, + "grad_norm": 2.1387271881103516, + "learning_rate": 7.069154774972558e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6776858568191528, + "num_tokens": 33288162.0, + "step": 1289 + }, + { + "epoch": 0.1416648363716231, + "grad_norm": 2.3246190547943115, + "learning_rate": 7.074643249176728e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.6965397596359253, + "num_tokens": 33312588.0, + "step": 1290 + }, + { + "epoch": 0.14177465407423676, + "grad_norm": 2.4183268547058105, + "learning_rate": 7.0801317233809e-07, + "loss": 0.9469, + "mean_token_accuracy": 0.7192294001579285, + "num_tokens": 33335415.0, + "step": 1291 + }, + { + "epoch": 0.14188447177685043, + "grad_norm": 2.4759726524353027, + "learning_rate": 7.085620197585072e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.6760246157646179, + "num_tokens": 33361103.0, + "step": 1292 + }, + { + "epoch": 0.1419942894794641, + "grad_norm": 1.9145234823226929, + "learning_rate": 7.091108671789242e-07, + "loss": 1.1054, + "mean_token_accuracy": 0.6741507053375244, + "num_tokens": 33395228.0, + "step": 1293 + }, + { + "epoch": 0.14210410718207775, + "grad_norm": 2.2252631187438965, + "learning_rate": 7.096597145993414e-07, + "loss": 1.0128, + "mean_token_accuracy": 0.6926039457321167, + "num_tokens": 33421456.0, + "step": 1294 + }, + { + "epoch": 0.14221392488469142, + "grad_norm": 2.480959177017212, + "learning_rate": 7.102085620197584e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6721041798591614, + "num_tokens": 33444415.0, + "step": 1295 + }, + { + "epoch": 0.14232374258730507, + "grad_norm": 2.2051334381103516, + "learning_rate": 7.107574094401756e-07, + "loss": 1.0153, + "mean_token_accuracy": 0.7048266530036926, + "num_tokens": 33469155.0, + "step": 1296 + }, + { + "epoch": 0.14243356028991874, + "grad_norm": 2.3750851154327393, + "learning_rate": 7.113062568605928e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.6879945397377014, + "num_tokens": 33492389.0, + "step": 1297 + }, + { + "epoch": 0.1425433779925324, + "grad_norm": 2.1226072311401367, + "learning_rate": 7.118551042810098e-07, + "loss": 1.0561, + "mean_token_accuracy": 0.6845663785934448, + "num_tokens": 33519232.0, + "step": 1298 + }, + { + "epoch": 0.14265319569514606, + "grad_norm": 2.3626084327697754, + "learning_rate": 7.12403951701427e-07, + "loss": 1.1412, + "mean_token_accuracy": 0.6628658771514893, + "num_tokens": 33544933.0, + "step": 1299 + }, + { + "epoch": 0.1427630133977597, + "grad_norm": 2.2089955806732178, + "learning_rate": 7.129527991218442e-07, + "loss": 1.0412, + "mean_token_accuracy": 0.685169517993927, + "num_tokens": 33570322.0, + "step": 1300 + }, + { + "epoch": 0.14287283110037338, + "grad_norm": 2.3130993843078613, + "learning_rate": 7.135016465422611e-07, + "loss": 1.0452, + "mean_token_accuracy": 0.7000476717948914, + "num_tokens": 33595366.0, + "step": 1301 + }, + { + "epoch": 0.14298264880298703, + "grad_norm": 2.341707229614258, + "learning_rate": 7.140504939626783e-07, + "loss": 0.9099, + "mean_token_accuracy": 0.7227811813354492, + "num_tokens": 33619103.0, + "step": 1302 + }, + { + "epoch": 0.1430924665056007, + "grad_norm": 2.1320862770080566, + "learning_rate": 7.145993413830955e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6975702047348022, + "num_tokens": 33647881.0, + "step": 1303 + }, + { + "epoch": 0.14320228420821438, + "grad_norm": 2.103644847869873, + "learning_rate": 7.151481888035125e-07, + "loss": 0.9971, + "mean_token_accuracy": 0.7024332880973816, + "num_tokens": 33676475.0, + "step": 1304 + }, + { + "epoch": 0.14331210191082802, + "grad_norm": 2.265986919403076, + "learning_rate": 7.156970362239297e-07, + "loss": 0.9848, + "mean_token_accuracy": 0.705594539642334, + "num_tokens": 33701077.0, + "step": 1305 + }, + { + "epoch": 0.1434219196134417, + "grad_norm": 1.9774998426437378, + "learning_rate": 7.162458836443468e-07, + "loss": 1.0764, + "mean_token_accuracy": 0.685038685798645, + "num_tokens": 33734555.0, + "step": 1306 + }, + { + "epoch": 0.14353173731605534, + "grad_norm": 2.2740893363952637, + "learning_rate": 7.167947310647639e-07, + "loss": 0.9747, + "mean_token_accuracy": 0.7279050946235657, + "num_tokens": 33758524.0, + "step": 1307 + }, + { + "epoch": 0.14364155501866901, + "grad_norm": 1.9783962965011597, + "learning_rate": 7.173435784851811e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6788167953491211, + "num_tokens": 33789758.0, + "step": 1308 + }, + { + "epoch": 0.14375137272128266, + "grad_norm": 2.2841856479644775, + "learning_rate": 7.178924259055982e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6843613386154175, + "num_tokens": 33815324.0, + "step": 1309 + }, + { + "epoch": 0.14386119042389633, + "grad_norm": 2.2222933769226074, + "learning_rate": 7.184412733260153e-07, + "loss": 0.9224, + "mean_token_accuracy": 0.7211000323295593, + "num_tokens": 33843524.0, + "step": 1310 + }, + { + "epoch": 0.14397100812650998, + "grad_norm": 2.318150281906128, + "learning_rate": 7.189901207464325e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.6881918907165527, + "num_tokens": 33869456.0, + "step": 1311 + }, + { + "epoch": 0.14408082582912365, + "grad_norm": 2.3406691551208496, + "learning_rate": 7.195389681668496e-07, + "loss": 1.0688, + "mean_token_accuracy": 0.6811075806617737, + "num_tokens": 33894255.0, + "step": 1312 + }, + { + "epoch": 0.14419064353173733, + "grad_norm": 2.517151117324829, + "learning_rate": 7.200878155872667e-07, + "loss": 0.9457, + "mean_token_accuracy": 0.7079068422317505, + "num_tokens": 33913444.0, + "step": 1313 + }, + { + "epoch": 0.14430046123435097, + "grad_norm": 2.384812355041504, + "learning_rate": 7.206366630076838e-07, + "loss": 1.0532, + "mean_token_accuracy": 0.6856403350830078, + "num_tokens": 33937048.0, + "step": 1314 + }, + { + "epoch": 0.14441027893696465, + "grad_norm": 2.3286008834838867, + "learning_rate": 7.21185510428101e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6967281103134155, + "num_tokens": 33964165.0, + "step": 1315 + }, + { + "epoch": 0.1445200966395783, + "grad_norm": 1.9759920835494995, + "learning_rate": 7.217343578485181e-07, + "loss": 1.0188, + "mean_token_accuracy": 0.6984091997146606, + "num_tokens": 33997024.0, + "step": 1316 + }, + { + "epoch": 0.14462991434219197, + "grad_norm": 2.1808109283447266, + "learning_rate": 7.222832052689352e-07, + "loss": 1.0017, + "mean_token_accuracy": 0.7062618136405945, + "num_tokens": 34023820.0, + "step": 1317 + }, + { + "epoch": 0.1447397320448056, + "grad_norm": 2.3351070880889893, + "learning_rate": 7.228320526893524e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6964146494865417, + "num_tokens": 34048789.0, + "step": 1318 + }, + { + "epoch": 0.1448495497474193, + "grad_norm": 2.082139015197754, + "learning_rate": 7.233809001097695e-07, + "loss": 1.077, + "mean_token_accuracy": 0.6787333488464355, + "num_tokens": 34078586.0, + "step": 1319 + }, + { + "epoch": 0.14495936745003293, + "grad_norm": 2.2396910190582275, + "learning_rate": 7.239297475301865e-07, + "loss": 1.1106, + "mean_token_accuracy": 0.6723518967628479, + "num_tokens": 34105381.0, + "step": 1320 + }, + { + "epoch": 0.1450691851526466, + "grad_norm": 1.8713182210922241, + "learning_rate": 7.244785949506037e-07, + "loss": 1.1327, + "mean_token_accuracy": 0.6667565107345581, + "num_tokens": 34143094.0, + "step": 1321 + }, + { + "epoch": 0.14517900285526028, + "grad_norm": 2.5130505561828613, + "learning_rate": 7.250274423710208e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.6756371259689331, + "num_tokens": 34166332.0, + "step": 1322 + }, + { + "epoch": 0.14528882055787393, + "grad_norm": 2.232736587524414, + "learning_rate": 7.255762897914379e-07, + "loss": 1.0654, + "mean_token_accuracy": 0.684740424156189, + "num_tokens": 34193124.0, + "step": 1323 + }, + { + "epoch": 0.1453986382604876, + "grad_norm": 1.9943208694458008, + "learning_rate": 7.261251372118551e-07, + "loss": 1.1164, + "mean_token_accuracy": 0.6794610023498535, + "num_tokens": 34226801.0, + "step": 1324 + }, + { + "epoch": 0.14550845596310125, + "grad_norm": 2.2547576427459717, + "learning_rate": 7.266739846322721e-07, + "loss": 1.1094, + "mean_token_accuracy": 0.6666133403778076, + "num_tokens": 34254543.0, + "step": 1325 + }, + { + "epoch": 0.14561827366571492, + "grad_norm": 2.372673273086548, + "learning_rate": 7.272228320526893e-07, + "loss": 0.8958, + "mean_token_accuracy": 0.726413369178772, + "num_tokens": 34278171.0, + "step": 1326 + }, + { + "epoch": 0.14572809136832857, + "grad_norm": 2.53460431098938, + "learning_rate": 7.277716794731065e-07, + "loss": 0.9776, + "mean_token_accuracy": 0.7026708722114563, + "num_tokens": 34299509.0, + "step": 1327 + }, + { + "epoch": 0.14583790907094224, + "grad_norm": 2.191556215286255, + "learning_rate": 7.283205268935235e-07, + "loss": 1.0125, + "mean_token_accuracy": 0.6969195008277893, + "num_tokens": 34326265.0, + "step": 1328 + }, + { + "epoch": 0.14594772677355589, + "grad_norm": 2.4968340396881104, + "learning_rate": 7.288693743139407e-07, + "loss": 0.9969, + "mean_token_accuracy": 0.6963183879852295, + "num_tokens": 34348330.0, + "step": 1329 + }, + { + "epoch": 0.14605754447616956, + "grad_norm": 2.3837687969207764, + "learning_rate": 7.294182217343579e-07, + "loss": 0.9969, + "mean_token_accuracy": 0.6962734460830688, + "num_tokens": 34371870.0, + "step": 1330 + }, + { + "epoch": 0.14616736217878323, + "grad_norm": 2.209792375564575, + "learning_rate": 7.299670691547749e-07, + "loss": 1.0886, + "mean_token_accuracy": 0.6827858686447144, + "num_tokens": 34397588.0, + "step": 1331 + }, + { + "epoch": 0.14627717988139688, + "grad_norm": 2.5140540599823, + "learning_rate": 7.305159165751921e-07, + "loss": 1.016, + "mean_token_accuracy": 0.6970952749252319, + "num_tokens": 34420268.0, + "step": 1332 + }, + { + "epoch": 0.14638699758401055, + "grad_norm": 2.356855630874634, + "learning_rate": 7.310647639956093e-07, + "loss": 1.0954, + "mean_token_accuracy": 0.6747195720672607, + "num_tokens": 34447551.0, + "step": 1333 + }, + { + "epoch": 0.1464968152866242, + "grad_norm": 2.4058473110198975, + "learning_rate": 7.316136114160263e-07, + "loss": 1.0619, + "mean_token_accuracy": 0.6849764585494995, + "num_tokens": 34472082.0, + "step": 1334 + }, + { + "epoch": 0.14660663298923787, + "grad_norm": 2.1378767490386963, + "learning_rate": 7.321624588364435e-07, + "loss": 0.9913, + "mean_token_accuracy": 0.7043039202690125, + "num_tokens": 34499892.0, + "step": 1335 + }, + { + "epoch": 0.14671645069185152, + "grad_norm": 2.2102129459381104, + "learning_rate": 7.327113062568605e-07, + "loss": 1.1337, + "mean_token_accuracy": 0.6681075096130371, + "num_tokens": 34528205.0, + "step": 1336 + }, + { + "epoch": 0.1468262683944652, + "grad_norm": 2.6815531253814697, + "learning_rate": 7.332601536772777e-07, + "loss": 1.1023, + "mean_token_accuracy": 0.6803075075149536, + "num_tokens": 34550578.0, + "step": 1337 + }, + { + "epoch": 0.14693608609707884, + "grad_norm": 2.2838187217712402, + "learning_rate": 7.338090010976949e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.691199541091919, + "num_tokens": 34576587.0, + "step": 1338 + }, + { + "epoch": 0.1470459037996925, + "grad_norm": 2.3646371364593506, + "learning_rate": 7.343578485181118e-07, + "loss": 1.0875, + "mean_token_accuracy": 0.6777167320251465, + "num_tokens": 34601076.0, + "step": 1339 + }, + { + "epoch": 0.14715572150230616, + "grad_norm": 2.5331740379333496, + "learning_rate": 7.34906695938529e-07, + "loss": 1.0765, + "mean_token_accuracy": 0.6753513813018799, + "num_tokens": 34621592.0, + "step": 1340 + }, + { + "epoch": 0.14726553920491983, + "grad_norm": 2.2956929206848145, + "learning_rate": 7.354555433589462e-07, + "loss": 0.9303, + "mean_token_accuracy": 0.7126259207725525, + "num_tokens": 34643526.0, + "step": 1341 + }, + { + "epoch": 0.1473753569075335, + "grad_norm": 2.2242794036865234, + "learning_rate": 7.360043907793632e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6838982105255127, + "num_tokens": 34669540.0, + "step": 1342 + }, + { + "epoch": 0.14748517461014715, + "grad_norm": 2.2233316898345947, + "learning_rate": 7.365532381997804e-07, + "loss": 1.0989, + "mean_token_accuracy": 0.6686229705810547, + "num_tokens": 34697762.0, + "step": 1343 + }, + { + "epoch": 0.14759499231276083, + "grad_norm": 2.440218210220337, + "learning_rate": 7.371020856201976e-07, + "loss": 0.9293, + "mean_token_accuracy": 0.7107264399528503, + "num_tokens": 34719352.0, + "step": 1344 + }, + { + "epoch": 0.14770481001537447, + "grad_norm": 2.3952994346618652, + "learning_rate": 7.376509330406146e-07, + "loss": 1.0181, + "mean_token_accuracy": 0.7001185417175293, + "num_tokens": 34746136.0, + "step": 1345 + }, + { + "epoch": 0.14781462771798815, + "grad_norm": 2.35957407951355, + "learning_rate": 7.381997804610318e-07, + "loss": 1.0799, + "mean_token_accuracy": 0.688084602355957, + "num_tokens": 34770519.0, + "step": 1346 + }, + { + "epoch": 0.1479244454206018, + "grad_norm": 2.3548645973205566, + "learning_rate": 7.387486278814489e-07, + "loss": 1.0442, + "mean_token_accuracy": 0.6899288296699524, + "num_tokens": 34794829.0, + "step": 1347 + }, + { + "epoch": 0.14803426312321546, + "grad_norm": 2.382566213607788, + "learning_rate": 7.39297475301866e-07, + "loss": 1.1035, + "mean_token_accuracy": 0.6720000505447388, + "num_tokens": 34819683.0, + "step": 1348 + }, + { + "epoch": 0.1481440808258291, + "grad_norm": 2.1950902938842773, + "learning_rate": 7.398463227222832e-07, + "loss": 1.0615, + "mean_token_accuracy": 0.6842712163925171, + "num_tokens": 34847837.0, + "step": 1349 + }, + { + "epoch": 0.14825389852844278, + "grad_norm": 2.5300328731536865, + "learning_rate": 7.403951701427003e-07, + "loss": 1.015, + "mean_token_accuracy": 0.6929871439933777, + "num_tokens": 34870050.0, + "step": 1350 + }, + { + "epoch": 0.14836371623105646, + "grad_norm": 2.187634229660034, + "learning_rate": 7.409440175631174e-07, + "loss": 1.099, + "mean_token_accuracy": 0.6784944534301758, + "num_tokens": 34898835.0, + "step": 1351 + }, + { + "epoch": 0.1484735339336701, + "grad_norm": 2.524291753768921, + "learning_rate": 7.414928649835346e-07, + "loss": 0.9955, + "mean_token_accuracy": 0.6964452266693115, + "num_tokens": 34919416.0, + "step": 1352 + }, + { + "epoch": 0.14858335163628378, + "grad_norm": 2.166205644607544, + "learning_rate": 7.420417124039517e-07, + "loss": 1.0066, + "mean_token_accuracy": 0.6937492489814758, + "num_tokens": 34947091.0, + "step": 1353 + }, + { + "epoch": 0.14869316933889742, + "grad_norm": 2.2300732135772705, + "learning_rate": 7.425905598243688e-07, + "loss": 0.9792, + "mean_token_accuracy": 0.7039631605148315, + "num_tokens": 34971956.0, + "step": 1354 + }, + { + "epoch": 0.1488029870415111, + "grad_norm": 2.3922882080078125, + "learning_rate": 7.43139407244786e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.678548276424408, + "num_tokens": 34996624.0, + "step": 1355 + }, + { + "epoch": 0.14891280474412474, + "grad_norm": 2.3343324661254883, + "learning_rate": 7.436882546652031e-07, + "loss": 1.001, + "mean_token_accuracy": 0.6984245777130127, + "num_tokens": 35021421.0, + "step": 1356 + }, + { + "epoch": 0.14902262244673842, + "grad_norm": 2.038109540939331, + "learning_rate": 7.442371020856202e-07, + "loss": 1.0248, + "mean_token_accuracy": 0.6937536001205444, + "num_tokens": 35053251.0, + "step": 1357 + }, + { + "epoch": 0.14913244014935206, + "grad_norm": 2.462204694747925, + "learning_rate": 7.447859495060372e-07, + "loss": 1.0588, + "mean_token_accuracy": 0.6892133355140686, + "num_tokens": 35076065.0, + "step": 1358 + }, + { + "epoch": 0.14924225785196574, + "grad_norm": 2.1705524921417236, + "learning_rate": 7.453347969264544e-07, + "loss": 0.9965, + "mean_token_accuracy": 0.6944243907928467, + "num_tokens": 35102598.0, + "step": 1359 + }, + { + "epoch": 0.1493520755545794, + "grad_norm": 2.369403600692749, + "learning_rate": 7.458836443468715e-07, + "loss": 1.0351, + "mean_token_accuracy": 0.6932587623596191, + "num_tokens": 35127150.0, + "step": 1360 + }, + { + "epoch": 0.14946189325719306, + "grad_norm": 2.2072699069976807, + "learning_rate": 7.464324917672886e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.6854552626609802, + "num_tokens": 35156357.0, + "step": 1361 + }, + { + "epoch": 0.14957171095980673, + "grad_norm": 2.369600296020508, + "learning_rate": 7.469813391877058e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.6784845590591431, + "num_tokens": 35182510.0, + "step": 1362 + }, + { + "epoch": 0.14968152866242038, + "grad_norm": 2.0737268924713135, + "learning_rate": 7.475301866081229e-07, + "loss": 1.0924, + "mean_token_accuracy": 0.6761667728424072, + "num_tokens": 35212284.0, + "step": 1363 + }, + { + "epoch": 0.14979134636503405, + "grad_norm": 2.1481382846832275, + "learning_rate": 7.4807903402854e-07, + "loss": 0.9889, + "mean_token_accuracy": 0.7028484344482422, + "num_tokens": 35241984.0, + "step": 1364 + }, + { + "epoch": 0.1499011640676477, + "grad_norm": 2.0728096961975098, + "learning_rate": 7.486278814489572e-07, + "loss": 1.0504, + "mean_token_accuracy": 0.6837591528892517, + "num_tokens": 35272523.0, + "step": 1365 + }, + { + "epoch": 0.15001098177026137, + "grad_norm": 2.1806647777557373, + "learning_rate": 7.491767288693743e-07, + "loss": 1.118, + "mean_token_accuracy": 0.6794604063034058, + "num_tokens": 35301675.0, + "step": 1366 + }, + { + "epoch": 0.15012079947287502, + "grad_norm": 2.4592316150665283, + "learning_rate": 7.497255762897914e-07, + "loss": 0.9759, + "mean_token_accuracy": 0.7051310539245605, + "num_tokens": 35324699.0, + "step": 1367 + }, + { + "epoch": 0.1502306171754887, + "grad_norm": 2.2018065452575684, + "learning_rate": 7.502744237102086e-07, + "loss": 1.0175, + "mean_token_accuracy": 0.6990983486175537, + "num_tokens": 35352504.0, + "step": 1368 + }, + { + "epoch": 0.15034043487810236, + "grad_norm": 2.406151533126831, + "learning_rate": 7.508232711306256e-07, + "loss": 1.0506, + "mean_token_accuracy": 0.6850656270980835, + "num_tokens": 35375920.0, + "step": 1369 + }, + { + "epoch": 0.150450252580716, + "grad_norm": 2.3846817016601562, + "learning_rate": 7.513721185510428e-07, + "loss": 1.0713, + "mean_token_accuracy": 0.6799846887588501, + "num_tokens": 35403417.0, + "step": 1370 + }, + { + "epoch": 0.15056007028332968, + "grad_norm": 2.348493814468384, + "learning_rate": 7.5192096597146e-07, + "loss": 1.0064, + "mean_token_accuracy": 0.7020888328552246, + "num_tokens": 35428034.0, + "step": 1371 + }, + { + "epoch": 0.15066988798594333, + "grad_norm": 2.1899027824401855, + "learning_rate": 7.52469813391877e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6956349611282349, + "num_tokens": 35455125.0, + "step": 1372 + }, + { + "epoch": 0.150779705688557, + "grad_norm": 2.1217737197875977, + "learning_rate": 7.530186608122942e-07, + "loss": 1.1677, + "mean_token_accuracy": 0.6620120406150818, + "num_tokens": 35486984.0, + "step": 1373 + }, + { + "epoch": 0.15088952339117065, + "grad_norm": 2.6656205654144287, + "learning_rate": 7.535675082327113e-07, + "loss": 1.0373, + "mean_token_accuracy": 0.6890026330947876, + "num_tokens": 35508094.0, + "step": 1374 + }, + { + "epoch": 0.15099934109378432, + "grad_norm": 2.429743766784668, + "learning_rate": 7.541163556531284e-07, + "loss": 0.9985, + "mean_token_accuracy": 0.7016690373420715, + "num_tokens": 35530825.0, + "step": 1375 + }, + { + "epoch": 0.15110915879639797, + "grad_norm": 2.386054277420044, + "learning_rate": 7.546652030735456e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.6847687363624573, + "num_tokens": 35555579.0, + "step": 1376 + }, + { + "epoch": 0.15121897649901164, + "grad_norm": 2.358800172805786, + "learning_rate": 7.552140504939627e-07, + "loss": 0.9739, + "mean_token_accuracy": 0.7027828693389893, + "num_tokens": 35579173.0, + "step": 1377 + }, + { + "epoch": 0.1513287942016253, + "grad_norm": 2.053912401199341, + "learning_rate": 7.557628979143797e-07, + "loss": 1.0947, + "mean_token_accuracy": 0.6835090517997742, + "num_tokens": 35608608.0, + "step": 1378 + }, + { + "epoch": 0.15143861190423896, + "grad_norm": 2.4191126823425293, + "learning_rate": 7.563117453347969e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6833609342575073, + "num_tokens": 35631896.0, + "step": 1379 + }, + { + "epoch": 0.15154842960685264, + "grad_norm": 2.1086630821228027, + "learning_rate": 7.568605927552139e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6864771842956543, + "num_tokens": 35659520.0, + "step": 1380 + }, + { + "epoch": 0.15165824730946628, + "grad_norm": 2.420973539352417, + "learning_rate": 7.574094401756311e-07, + "loss": 0.9724, + "mean_token_accuracy": 0.7062900066375732, + "num_tokens": 35683276.0, + "step": 1381 + }, + { + "epoch": 0.15176806501207996, + "grad_norm": 2.353480100631714, + "learning_rate": 7.579582875960483e-07, + "loss": 1.1276, + "mean_token_accuracy": 0.6609391570091248, + "num_tokens": 35707625.0, + "step": 1382 + }, + { + "epoch": 0.1518778827146936, + "grad_norm": 2.2905161380767822, + "learning_rate": 7.585071350164653e-07, + "loss": 1.0806, + "mean_token_accuracy": 0.6768396496772766, + "num_tokens": 35735417.0, + "step": 1383 + }, + { + "epoch": 0.15198770041730728, + "grad_norm": 2.2017388343811035, + "learning_rate": 7.590559824368825e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7099335193634033, + "num_tokens": 35759947.0, + "step": 1384 + }, + { + "epoch": 0.15209751811992092, + "grad_norm": 2.133969306945801, + "learning_rate": 7.596048298572997e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6682755947113037, + "num_tokens": 35789554.0, + "step": 1385 + }, + { + "epoch": 0.1522073358225346, + "grad_norm": 2.4642598628997803, + "learning_rate": 7.601536772777167e-07, + "loss": 1.022, + "mean_token_accuracy": 0.6931875944137573, + "num_tokens": 35814038.0, + "step": 1386 + }, + { + "epoch": 0.15231715352514824, + "grad_norm": 2.007112979888916, + "learning_rate": 7.607025246981339e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6783881187438965, + "num_tokens": 35845257.0, + "step": 1387 + }, + { + "epoch": 0.15242697122776191, + "grad_norm": 2.136127233505249, + "learning_rate": 7.612513721185511e-07, + "loss": 1.0201, + "mean_token_accuracy": 0.692333459854126, + "num_tokens": 35872589.0, + "step": 1388 + }, + { + "epoch": 0.1525367889303756, + "grad_norm": 2.3932268619537354, + "learning_rate": 7.618002195389681e-07, + "loss": 1.1025, + "mean_token_accuracy": 0.6735714077949524, + "num_tokens": 35899016.0, + "step": 1389 + }, + { + "epoch": 0.15264660663298923, + "grad_norm": 2.2572171688079834, + "learning_rate": 7.623490669593853e-07, + "loss": 1.0533, + "mean_token_accuracy": 0.6871049404144287, + "num_tokens": 35926864.0, + "step": 1390 + }, + { + "epoch": 0.1527564243356029, + "grad_norm": 2.4204695224761963, + "learning_rate": 7.628979143798024e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.6925952434539795, + "num_tokens": 35950373.0, + "step": 1391 + }, + { + "epoch": 0.15286624203821655, + "grad_norm": 2.2808752059936523, + "learning_rate": 7.634467618002195e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.6881481409072876, + "num_tokens": 35977252.0, + "step": 1392 + }, + { + "epoch": 0.15297605974083023, + "grad_norm": 2.4310946464538574, + "learning_rate": 7.639956092206367e-07, + "loss": 1.0664, + "mean_token_accuracy": 0.6828088164329529, + "num_tokens": 36000722.0, + "step": 1393 + }, + { + "epoch": 0.15308587744344387, + "grad_norm": 2.273108959197998, + "learning_rate": 7.645444566410538e-07, + "loss": 1.1125, + "mean_token_accuracy": 0.6693401336669922, + "num_tokens": 36028075.0, + "step": 1394 + }, + { + "epoch": 0.15319569514605755, + "grad_norm": 2.2174477577209473, + "learning_rate": 7.650933040614709e-07, + "loss": 0.9375, + "mean_token_accuracy": 0.7275620102882385, + "num_tokens": 36054110.0, + "step": 1395 + }, + { + "epoch": 0.1533055128486712, + "grad_norm": 2.18259859085083, + "learning_rate": 7.656421514818881e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.688517153263092, + "num_tokens": 36082729.0, + "step": 1396 + }, + { + "epoch": 0.15341533055128487, + "grad_norm": 2.31372332572937, + "learning_rate": 7.66190998902305e-07, + "loss": 1.0241, + "mean_token_accuracy": 0.700005054473877, + "num_tokens": 36108841.0, + "step": 1397 + }, + { + "epoch": 0.15352514825389854, + "grad_norm": 1.936745524406433, + "learning_rate": 7.667398463227222e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.6841425895690918, + "num_tokens": 36142306.0, + "step": 1398 + }, + { + "epoch": 0.1536349659565122, + "grad_norm": 2.1762139797210693, + "learning_rate": 7.672886937431394e-07, + "loss": 1.1277, + "mean_token_accuracy": 0.6658376455307007, + "num_tokens": 36171301.0, + "step": 1399 + }, + { + "epoch": 0.15374478365912586, + "grad_norm": 2.52713942527771, + "learning_rate": 7.678375411635564e-07, + "loss": 1.0422, + "mean_token_accuracy": 0.6888754367828369, + "num_tokens": 36193331.0, + "step": 1400 + }, + { + "epoch": 0.1538546013617395, + "grad_norm": 2.43285870552063, + "learning_rate": 7.683863885839736e-07, + "loss": 1.006, + "mean_token_accuracy": 0.6994996666908264, + "num_tokens": 36215365.0, + "step": 1401 + }, + { + "epoch": 0.15396441906435318, + "grad_norm": 2.1144216060638428, + "learning_rate": 7.689352360043907e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.6848171949386597, + "num_tokens": 36244206.0, + "step": 1402 + }, + { + "epoch": 0.15407423676696683, + "grad_norm": 2.3209400177001953, + "learning_rate": 7.694840834248078e-07, + "loss": 1.0978, + "mean_token_accuracy": 0.6862040758132935, + "num_tokens": 36269726.0, + "step": 1403 + }, + { + "epoch": 0.1541840544695805, + "grad_norm": 2.40747332572937, + "learning_rate": 7.70032930845225e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6698656678199768, + "num_tokens": 36292749.0, + "step": 1404 + }, + { + "epoch": 0.15429387217219415, + "grad_norm": 2.168182373046875, + "learning_rate": 7.705817782656421e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7073327302932739, + "num_tokens": 36318879.0, + "step": 1405 + }, + { + "epoch": 0.15440368987480782, + "grad_norm": 2.431025266647339, + "learning_rate": 7.711306256860592e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6903561353683472, + "num_tokens": 36341366.0, + "step": 1406 + }, + { + "epoch": 0.1545135075774215, + "grad_norm": 2.7083394527435303, + "learning_rate": 7.716794731064764e-07, + "loss": 1.0404, + "mean_token_accuracy": 0.6857359409332275, + "num_tokens": 36362546.0, + "step": 1407 + }, + { + "epoch": 0.15462332528003514, + "grad_norm": 2.5637731552124023, + "learning_rate": 7.722283205268935e-07, + "loss": 1.02, + "mean_token_accuracy": 0.6994880437850952, + "num_tokens": 36384944.0, + "step": 1408 + }, + { + "epoch": 0.1547331429826488, + "grad_norm": 2.408661365509033, + "learning_rate": 7.727771679473106e-07, + "loss": 1.0387, + "mean_token_accuracy": 0.6884039640426636, + "num_tokens": 36409264.0, + "step": 1409 + }, + { + "epoch": 0.15484296068526246, + "grad_norm": 2.2424063682556152, + "learning_rate": 7.733260153677278e-07, + "loss": 1.1133, + "mean_token_accuracy": 0.678756058216095, + "num_tokens": 36436366.0, + "step": 1410 + }, + { + "epoch": 0.15495277838787613, + "grad_norm": 2.437101125717163, + "learning_rate": 7.738748627881449e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6886452436447144, + "num_tokens": 36460916.0, + "step": 1411 + }, + { + "epoch": 0.15506259609048978, + "grad_norm": 2.312267303466797, + "learning_rate": 7.74423710208562e-07, + "loss": 1.0253, + "mean_token_accuracy": 0.7037920951843262, + "num_tokens": 36487005.0, + "step": 1412 + }, + { + "epoch": 0.15517241379310345, + "grad_norm": 2.1808860301971436, + "learning_rate": 7.749725576289791e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6792519092559814, + "num_tokens": 36513325.0, + "step": 1413 + }, + { + "epoch": 0.1552822314957171, + "grad_norm": 2.431476593017578, + "learning_rate": 7.755214050493963e-07, + "loss": 1.048, + "mean_token_accuracy": 0.6876606941223145, + "num_tokens": 36535194.0, + "step": 1414 + }, + { + "epoch": 0.15539204919833077, + "grad_norm": 2.08186674118042, + "learning_rate": 7.760702524698134e-07, + "loss": 0.9448, + "mean_token_accuracy": 0.7123320698738098, + "num_tokens": 36561950.0, + "step": 1415 + }, + { + "epoch": 0.15550186690094442, + "grad_norm": 2.2612249851226807, + "learning_rate": 7.766190998902304e-07, + "loss": 1.0032, + "mean_token_accuracy": 0.7048555612564087, + "num_tokens": 36587245.0, + "step": 1416 + }, + { + "epoch": 0.1556116846035581, + "grad_norm": 2.2319977283477783, + "learning_rate": 7.771679473106476e-07, + "loss": 0.9749, + "mean_token_accuracy": 0.7083872556686401, + "num_tokens": 36611902.0, + "step": 1417 + }, + { + "epoch": 0.15572150230617177, + "grad_norm": 2.2532973289489746, + "learning_rate": 7.777167947310647e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.687932550907135, + "num_tokens": 36638782.0, + "step": 1418 + }, + { + "epoch": 0.1558313200087854, + "grad_norm": 2.196115255355835, + "learning_rate": 7.782656421514818e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.699591338634491, + "num_tokens": 36664716.0, + "step": 1419 + }, + { + "epoch": 0.15594113771139909, + "grad_norm": 2.0334432125091553, + "learning_rate": 7.78814489571899e-07, + "loss": 1.1093, + "mean_token_accuracy": 0.6727045774459839, + "num_tokens": 36699423.0, + "step": 1420 + }, + { + "epoch": 0.15605095541401273, + "grad_norm": 2.1976218223571777, + "learning_rate": 7.793633369923161e-07, + "loss": 1.075, + "mean_token_accuracy": 0.678813636302948, + "num_tokens": 36728926.0, + "step": 1421 + }, + { + "epoch": 0.1561607731166264, + "grad_norm": 2.5442278385162354, + "learning_rate": 7.799121844127332e-07, + "loss": 1.1309, + "mean_token_accuracy": 0.6649329662322998, + "num_tokens": 36751947.0, + "step": 1422 + }, + { + "epoch": 0.15627059081924005, + "grad_norm": 2.186750888824463, + "learning_rate": 7.804610318331504e-07, + "loss": 1.1361, + "mean_token_accuracy": 0.6619511842727661, + "num_tokens": 36781844.0, + "step": 1423 + }, + { + "epoch": 0.15638040852185373, + "grad_norm": 2.3753278255462646, + "learning_rate": 7.810098792535674e-07, + "loss": 1.0223, + "mean_token_accuracy": 0.6940719485282898, + "num_tokens": 36805887.0, + "step": 1424 + }, + { + "epoch": 0.15649022622446737, + "grad_norm": 2.3852407932281494, + "learning_rate": 7.815587266739846e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6970892548561096, + "num_tokens": 36830005.0, + "step": 1425 + }, + { + "epoch": 0.15660004392708105, + "grad_norm": 2.275900363922119, + "learning_rate": 7.821075740944018e-07, + "loss": 1.0696, + "mean_token_accuracy": 0.6894403100013733, + "num_tokens": 36857539.0, + "step": 1426 + }, + { + "epoch": 0.15670986162969472, + "grad_norm": 2.038562536239624, + "learning_rate": 7.826564215148188e-07, + "loss": 1.0905, + "mean_token_accuracy": 0.6803295612335205, + "num_tokens": 36890169.0, + "step": 1427 + }, + { + "epoch": 0.15681967933230836, + "grad_norm": 2.3716142177581787, + "learning_rate": 7.83205268935236e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6814667582511902, + "num_tokens": 36913683.0, + "step": 1428 + }, + { + "epoch": 0.15692949703492204, + "grad_norm": 2.167816638946533, + "learning_rate": 7.837541163556532e-07, + "loss": 1.0063, + "mean_token_accuracy": 0.6981418132781982, + "num_tokens": 36941170.0, + "step": 1429 + }, + { + "epoch": 0.15703931473753568, + "grad_norm": 2.1580734252929688, + "learning_rate": 7.843029637760702e-07, + "loss": 0.973, + "mean_token_accuracy": 0.7181742787361145, + "num_tokens": 36967443.0, + "step": 1430 + }, + { + "epoch": 0.15714913244014936, + "grad_norm": 2.402902126312256, + "learning_rate": 7.848518111964874e-07, + "loss": 0.9112, + "mean_token_accuracy": 0.7183106541633606, + "num_tokens": 36988350.0, + "step": 1431 + }, + { + "epoch": 0.157258950142763, + "grad_norm": 2.4805073738098145, + "learning_rate": 7.854006586169045e-07, + "loss": 1.0524, + "mean_token_accuracy": 0.6844382882118225, + "num_tokens": 37010718.0, + "step": 1432 + }, + { + "epoch": 0.15736876784537668, + "grad_norm": 2.4635567665100098, + "learning_rate": 7.859495060373216e-07, + "loss": 1.1116, + "mean_token_accuracy": 0.6736903190612793, + "num_tokens": 37035245.0, + "step": 1433 + }, + { + "epoch": 0.15747858554799032, + "grad_norm": 2.220958948135376, + "learning_rate": 7.864983534577388e-07, + "loss": 0.9563, + "mean_token_accuracy": 0.7113574743270874, + "num_tokens": 37061637.0, + "step": 1434 + }, + { + "epoch": 0.157588403250604, + "grad_norm": 2.526224136352539, + "learning_rate": 7.870472008781557e-07, + "loss": 1.0322, + "mean_token_accuracy": 0.6962623596191406, + "num_tokens": 37085688.0, + "step": 1435 + }, + { + "epoch": 0.15769822095321767, + "grad_norm": 2.299208164215088, + "learning_rate": 7.875960482985729e-07, + "loss": 0.9423, + "mean_token_accuracy": 0.7210569381713867, + "num_tokens": 37109412.0, + "step": 1436 + }, + { + "epoch": 0.15780803865583132, + "grad_norm": 2.219951868057251, + "learning_rate": 7.881448957189901e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6714613437652588, + "num_tokens": 37137832.0, + "step": 1437 + }, + { + "epoch": 0.157917856358445, + "grad_norm": 2.1788198947906494, + "learning_rate": 7.886937431394071e-07, + "loss": 1.0813, + "mean_token_accuracy": 0.674034833908081, + "num_tokens": 37166765.0, + "step": 1438 + }, + { + "epoch": 0.15802767406105864, + "grad_norm": 2.378258466720581, + "learning_rate": 7.892425905598243e-07, + "loss": 0.9907, + "mean_token_accuracy": 0.6986385583877563, + "num_tokens": 37190511.0, + "step": 1439 + }, + { + "epoch": 0.1581374917636723, + "grad_norm": 2.3341064453125, + "learning_rate": 7.897914379802415e-07, + "loss": 0.9456, + "mean_token_accuracy": 0.7155054807662964, + "num_tokens": 37215189.0, + "step": 1440 + }, + { + "epoch": 0.15824730946628596, + "grad_norm": 2.317101001739502, + "learning_rate": 7.903402854006585e-07, + "loss": 0.983, + "mean_token_accuracy": 0.7023878693580627, + "num_tokens": 37240251.0, + "step": 1441 + }, + { + "epoch": 0.15835712716889963, + "grad_norm": 2.557347297668457, + "learning_rate": 7.908891328210757e-07, + "loss": 1.06, + "mean_token_accuracy": 0.6880472302436829, + "num_tokens": 37264027.0, + "step": 1442 + }, + { + "epoch": 0.15846694487151328, + "grad_norm": 2.4403469562530518, + "learning_rate": 7.914379802414928e-07, + "loss": 0.9123, + "mean_token_accuracy": 0.7212167978286743, + "num_tokens": 37286230.0, + "step": 1443 + }, + { + "epoch": 0.15857676257412695, + "grad_norm": 2.0311732292175293, + "learning_rate": 7.919868276619099e-07, + "loss": 0.9394, + "mean_token_accuracy": 0.7110665440559387, + "num_tokens": 37316208.0, + "step": 1444 + }, + { + "epoch": 0.15868658027674062, + "grad_norm": 2.281578540802002, + "learning_rate": 7.925356750823271e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6969175338745117, + "num_tokens": 37340571.0, + "step": 1445 + }, + { + "epoch": 0.15879639797935427, + "grad_norm": 1.9251388311386108, + "learning_rate": 7.930845225027442e-07, + "loss": 0.981, + "mean_token_accuracy": 0.6999503374099731, + "num_tokens": 37371623.0, + "step": 1446 + }, + { + "epoch": 0.15890621568196794, + "grad_norm": 2.2013959884643555, + "learning_rate": 7.936333699231613e-07, + "loss": 1.0926, + "mean_token_accuracy": 0.6759529113769531, + "num_tokens": 37399398.0, + "step": 1447 + }, + { + "epoch": 0.1590160333845816, + "grad_norm": 2.5261878967285156, + "learning_rate": 7.941822173435785e-07, + "loss": 0.9978, + "mean_token_accuracy": 0.7004930973052979, + "num_tokens": 37421357.0, + "step": 1448 + }, + { + "epoch": 0.15912585108719526, + "grad_norm": 2.3398094177246094, + "learning_rate": 7.947310647639956e-07, + "loss": 1.0463, + "mean_token_accuracy": 0.6892502307891846, + "num_tokens": 37445588.0, + "step": 1449 + }, + { + "epoch": 0.1592356687898089, + "grad_norm": 2.260737419128418, + "learning_rate": 7.952799121844127e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7077124118804932, + "num_tokens": 37470169.0, + "step": 1450 + }, + { + "epoch": 0.15934548649242258, + "grad_norm": 2.349696636199951, + "learning_rate": 7.958287596048299e-07, + "loss": 1.0511, + "mean_token_accuracy": 0.6818383932113647, + "num_tokens": 37495057.0, + "step": 1451 + }, + { + "epoch": 0.15945530419503623, + "grad_norm": 2.1020753383636475, + "learning_rate": 7.96377607025247e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.7106344103813171, + "num_tokens": 37523296.0, + "step": 1452 + }, + { + "epoch": 0.1595651218976499, + "grad_norm": 2.2874903678894043, + "learning_rate": 7.969264544456641e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.6900336742401123, + "num_tokens": 37546847.0, + "step": 1453 + }, + { + "epoch": 0.15967493960026355, + "grad_norm": 2.4039723873138428, + "learning_rate": 7.974753018660811e-07, + "loss": 1.1459, + "mean_token_accuracy": 0.6670612096786499, + "num_tokens": 37573460.0, + "step": 1454 + }, + { + "epoch": 0.15978475730287722, + "grad_norm": 2.1106925010681152, + "learning_rate": 7.980241492864983e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.6938914656639099, + "num_tokens": 37601521.0, + "step": 1455 + }, + { + "epoch": 0.1598945750054909, + "grad_norm": 2.257035493850708, + "learning_rate": 7.985729967069154e-07, + "loss": 1.0311, + "mean_token_accuracy": 0.6873379349708557, + "num_tokens": 37629447.0, + "step": 1456 + }, + { + "epoch": 0.16000439270810454, + "grad_norm": 2.4342002868652344, + "learning_rate": 7.991218441273325e-07, + "loss": 0.9997, + "mean_token_accuracy": 0.6926524639129639, + "num_tokens": 37653780.0, + "step": 1457 + }, + { + "epoch": 0.16011421041071822, + "grad_norm": 2.4285085201263428, + "learning_rate": 7.996706915477497e-07, + "loss": 1.0215, + "mean_token_accuracy": 0.6969257593154907, + "num_tokens": 37676351.0, + "step": 1458 + }, + { + "epoch": 0.16022402811333186, + "grad_norm": 2.054798126220703, + "learning_rate": 8.002195389681668e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6951755285263062, + "num_tokens": 37707396.0, + "step": 1459 + }, + { + "epoch": 0.16033384581594554, + "grad_norm": 2.335123062133789, + "learning_rate": 8.007683863885839e-07, + "loss": 1.0693, + "mean_token_accuracy": 0.6896311640739441, + "num_tokens": 37733489.0, + "step": 1460 + }, + { + "epoch": 0.16044366351855918, + "grad_norm": 2.222598075866699, + "learning_rate": 8.013172338090011e-07, + "loss": 0.9802, + "mean_token_accuracy": 0.7096680998802185, + "num_tokens": 37760125.0, + "step": 1461 + }, + { + "epoch": 0.16055348122117286, + "grad_norm": 2.5458500385284424, + "learning_rate": 8.018660812294182e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6830626726150513, + "num_tokens": 37782324.0, + "step": 1462 + }, + { + "epoch": 0.1606632989237865, + "grad_norm": 2.221006393432617, + "learning_rate": 8.024149286498353e-07, + "loss": 0.9898, + "mean_token_accuracy": 0.7002509236335754, + "num_tokens": 37811011.0, + "step": 1463 + }, + { + "epoch": 0.16077311662640018, + "grad_norm": 2.198913812637329, + "learning_rate": 8.029637760702525e-07, + "loss": 1.029, + "mean_token_accuracy": 0.6919549703598022, + "num_tokens": 37838778.0, + "step": 1464 + }, + { + "epoch": 0.16088293432901385, + "grad_norm": 2.173797130584717, + "learning_rate": 8.035126234906695e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6866163015365601, + "num_tokens": 37866697.0, + "step": 1465 + }, + { + "epoch": 0.1609927520316275, + "grad_norm": 2.0759775638580322, + "learning_rate": 8.040614709110867e-07, + "loss": 1.0014, + "mean_token_accuracy": 0.7061452269554138, + "num_tokens": 37896466.0, + "step": 1466 + }, + { + "epoch": 0.16110256973424117, + "grad_norm": 2.283784866333008, + "learning_rate": 8.046103183315039e-07, + "loss": 0.9662, + "mean_token_accuracy": 0.7064900398254395, + "num_tokens": 37922368.0, + "step": 1467 + }, + { + "epoch": 0.16121238743685481, + "grad_norm": 2.3462610244750977, + "learning_rate": 8.051591657519209e-07, + "loss": 0.992, + "mean_token_accuracy": 0.7017149925231934, + "num_tokens": 37946656.0, + "step": 1468 + }, + { + "epoch": 0.1613222051394685, + "grad_norm": 2.232353448867798, + "learning_rate": 8.057080131723381e-07, + "loss": 1.1434, + "mean_token_accuracy": 0.6582103967666626, + "num_tokens": 37976118.0, + "step": 1469 + }, + { + "epoch": 0.16143202284208213, + "grad_norm": 2.4635159969329834, + "learning_rate": 8.062568605927553e-07, + "loss": 1.008, + "mean_token_accuracy": 0.6971036791801453, + "num_tokens": 37998602.0, + "step": 1470 + }, + { + "epoch": 0.1615418405446958, + "grad_norm": 2.4776153564453125, + "learning_rate": 8.068057080131723e-07, + "loss": 0.981, + "mean_token_accuracy": 0.7029331922531128, + "num_tokens": 38019516.0, + "step": 1471 + }, + { + "epoch": 0.16165165824730945, + "grad_norm": 2.4787096977233887, + "learning_rate": 8.073545554335894e-07, + "loss": 0.991, + "mean_token_accuracy": 0.7024984359741211, + "num_tokens": 38042515.0, + "step": 1472 + }, + { + "epoch": 0.16176147594992313, + "grad_norm": 2.176011562347412, + "learning_rate": 8.079034028540066e-07, + "loss": 1.1212, + "mean_token_accuracy": 0.6671150922775269, + "num_tokens": 38073753.0, + "step": 1473 + }, + { + "epoch": 0.1618712936525368, + "grad_norm": 2.2642757892608643, + "learning_rate": 8.084522502744236e-07, + "loss": 0.9942, + "mean_token_accuracy": 0.7013922929763794, + "num_tokens": 38101885.0, + "step": 1474 + }, + { + "epoch": 0.16198111135515045, + "grad_norm": 2.2147531509399414, + "learning_rate": 8.090010976948408e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6789680123329163, + "num_tokens": 38133231.0, + "step": 1475 + }, + { + "epoch": 0.16209092905776412, + "grad_norm": 2.4269349575042725, + "learning_rate": 8.095499451152578e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.674081563949585, + "num_tokens": 38157263.0, + "step": 1476 + }, + { + "epoch": 0.16220074676037777, + "grad_norm": 2.7061781883239746, + "learning_rate": 8.10098792535675e-07, + "loss": 0.9657, + "mean_token_accuracy": 0.709053099155426, + "num_tokens": 38176746.0, + "step": 1477 + }, + { + "epoch": 0.16231056446299144, + "grad_norm": 2.479259490966797, + "learning_rate": 8.106476399560922e-07, + "loss": 1.0268, + "mean_token_accuracy": 0.693744421005249, + "num_tokens": 38199320.0, + "step": 1478 + }, + { + "epoch": 0.1624203821656051, + "grad_norm": 2.2683730125427246, + "learning_rate": 8.111964873765092e-07, + "loss": 1.0118, + "mean_token_accuracy": 0.6951871514320374, + "num_tokens": 38224541.0, + "step": 1479 + }, + { + "epoch": 0.16253019986821876, + "grad_norm": 2.5875797271728516, + "learning_rate": 8.117453347969264e-07, + "loss": 0.9421, + "mean_token_accuracy": 0.7127686738967896, + "num_tokens": 38244885.0, + "step": 1480 + }, + { + "epoch": 0.1626400175708324, + "grad_norm": 2.369541645050049, + "learning_rate": 8.122941822173436e-07, + "loss": 1.0084, + "mean_token_accuracy": 0.6951762437820435, + "num_tokens": 38269135.0, + "step": 1481 + }, + { + "epoch": 0.16274983527344608, + "grad_norm": 2.0807273387908936, + "learning_rate": 8.128430296377606e-07, + "loss": 1.1099, + "mean_token_accuracy": 0.6686407327651978, + "num_tokens": 38301482.0, + "step": 1482 + }, + { + "epoch": 0.16285965297605975, + "grad_norm": 2.240413188934326, + "learning_rate": 8.133918770581778e-07, + "loss": 0.9603, + "mean_token_accuracy": 0.7090160846710205, + "num_tokens": 38327823.0, + "step": 1483 + }, + { + "epoch": 0.1629694706786734, + "grad_norm": 2.299295663833618, + "learning_rate": 8.13940724478595e-07, + "loss": 0.986, + "mean_token_accuracy": 0.6999321579933167, + "num_tokens": 38352760.0, + "step": 1484 + }, + { + "epoch": 0.16307928838128707, + "grad_norm": 2.413539409637451, + "learning_rate": 8.14489571899012e-07, + "loss": 1.0456, + "mean_token_accuracy": 0.6918749809265137, + "num_tokens": 38376213.0, + "step": 1485 + }, + { + "epoch": 0.16318910608390072, + "grad_norm": 2.3340792655944824, + "learning_rate": 8.150384193194292e-07, + "loss": 1.0837, + "mean_token_accuracy": 0.6855148077011108, + "num_tokens": 38403009.0, + "step": 1486 + }, + { + "epoch": 0.1632989237865144, + "grad_norm": 2.2101902961730957, + "learning_rate": 8.155872667398463e-07, + "loss": 1.1351, + "mean_token_accuracy": 0.6825194358825684, + "num_tokens": 38430141.0, + "step": 1487 + }, + { + "epoch": 0.16340874148912804, + "grad_norm": 2.9620962142944336, + "learning_rate": 8.161361141602634e-07, + "loss": 0.8702, + "mean_token_accuracy": 0.732130229473114, + "num_tokens": 38447057.0, + "step": 1488 + }, + { + "epoch": 0.1635185591917417, + "grad_norm": 2.3883347511291504, + "learning_rate": 8.166849615806806e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.6800752282142639, + "num_tokens": 38472415.0, + "step": 1489 + }, + { + "epoch": 0.16362837689435536, + "grad_norm": 2.2502477169036865, + "learning_rate": 8.172338090010977e-07, + "loss": 1.0964, + "mean_token_accuracy": 0.6788094639778137, + "num_tokens": 38503347.0, + "step": 1490 + }, + { + "epoch": 0.16373819459696903, + "grad_norm": 2.3202478885650635, + "learning_rate": 8.177826564215147e-07, + "loss": 1.086, + "mean_token_accuracy": 0.6796702146530151, + "num_tokens": 38528931.0, + "step": 1491 + }, + { + "epoch": 0.16384801229958268, + "grad_norm": 2.2192375659942627, + "learning_rate": 8.183315038419319e-07, + "loss": 1.0103, + "mean_token_accuracy": 0.7001367211341858, + "num_tokens": 38555247.0, + "step": 1492 + }, + { + "epoch": 0.16395783000219635, + "grad_norm": 2.5630974769592285, + "learning_rate": 8.18880351262349e-07, + "loss": 0.9289, + "mean_token_accuracy": 0.7122515439987183, + "num_tokens": 38574841.0, + "step": 1493 + }, + { + "epoch": 0.16406764770481003, + "grad_norm": 2.337538242340088, + "learning_rate": 8.194291986827661e-07, + "loss": 1.0412, + "mean_token_accuracy": 0.6938104629516602, + "num_tokens": 38600722.0, + "step": 1494 + }, + { + "epoch": 0.16417746540742367, + "grad_norm": 2.6462528705596924, + "learning_rate": 8.199780461031833e-07, + "loss": 0.9527, + "mean_token_accuracy": 0.7116312980651855, + "num_tokens": 38623033.0, + "step": 1495 + }, + { + "epoch": 0.16428728311003735, + "grad_norm": 2.1291582584381104, + "learning_rate": 8.205268935236004e-07, + "loss": 1.0075, + "mean_token_accuracy": 0.6959545612335205, + "num_tokens": 38651557.0, + "step": 1496 + }, + { + "epoch": 0.164397100812651, + "grad_norm": 2.5873007774353027, + "learning_rate": 8.210757409440175e-07, + "loss": 1.0689, + "mean_token_accuracy": 0.6796714067459106, + "num_tokens": 38674535.0, + "step": 1497 + }, + { + "epoch": 0.16450691851526467, + "grad_norm": 2.4706435203552246, + "learning_rate": 8.216245883644346e-07, + "loss": 1.0575, + "mean_token_accuracy": 0.6874227523803711, + "num_tokens": 38697800.0, + "step": 1498 + }, + { + "epoch": 0.1646167362178783, + "grad_norm": 2.3226914405822754, + "learning_rate": 8.221734357848518e-07, + "loss": 1.0782, + "mean_token_accuracy": 0.6869035959243774, + "num_tokens": 38722654.0, + "step": 1499 + }, + { + "epoch": 0.16472655392049199, + "grad_norm": 2.3142361640930176, + "learning_rate": 8.227222832052689e-07, + "loss": 0.9579, + "mean_token_accuracy": 0.7177810668945312, + "num_tokens": 38745804.0, + "step": 1500 + }, + { + "epoch": 0.16483637162310563, + "grad_norm": 2.0940346717834473, + "learning_rate": 8.23271130625686e-07, + "loss": 1.044, + "mean_token_accuracy": 0.690831184387207, + "num_tokens": 38774679.0, + "step": 1501 + }, + { + "epoch": 0.1649461893257193, + "grad_norm": 2.2435691356658936, + "learning_rate": 8.238199780461032e-07, + "loss": 0.9664, + "mean_token_accuracy": 0.7076376080513, + "num_tokens": 38799644.0, + "step": 1502 + }, + { + "epoch": 0.16505600702833298, + "grad_norm": 2.356931447982788, + "learning_rate": 8.243688254665203e-07, + "loss": 0.8399, + "mean_token_accuracy": 0.7369000911712646, + "num_tokens": 38820010.0, + "step": 1503 + }, + { + "epoch": 0.16516582473094663, + "grad_norm": 2.4481582641601562, + "learning_rate": 8.249176728869374e-07, + "loss": 0.9631, + "mean_token_accuracy": 0.7201657295227051, + "num_tokens": 38841622.0, + "step": 1504 + }, + { + "epoch": 0.1652756424335603, + "grad_norm": 2.46069598197937, + "learning_rate": 8.254665203073546e-07, + "loss": 0.9379, + "mean_token_accuracy": 0.7169673442840576, + "num_tokens": 38863956.0, + "step": 1505 + }, + { + "epoch": 0.16538546013617395, + "grad_norm": 2.4248416423797607, + "learning_rate": 8.260153677277717e-07, + "loss": 0.962, + "mean_token_accuracy": 0.7075158953666687, + "num_tokens": 38885651.0, + "step": 1506 + }, + { + "epoch": 0.16549527783878762, + "grad_norm": 2.5665159225463867, + "learning_rate": 8.265642151481888e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.6990289688110352, + "num_tokens": 38906585.0, + "step": 1507 + }, + { + "epoch": 0.16560509554140126, + "grad_norm": 2.41451096534729, + "learning_rate": 8.27113062568606e-07, + "loss": 0.9518, + "mean_token_accuracy": 0.7114508152008057, + "num_tokens": 38929071.0, + "step": 1508 + }, + { + "epoch": 0.16571491324401494, + "grad_norm": 2.667614698410034, + "learning_rate": 8.27661909989023e-07, + "loss": 1.043, + "mean_token_accuracy": 0.6992718577384949, + "num_tokens": 38951443.0, + "step": 1509 + }, + { + "epoch": 0.16582473094662858, + "grad_norm": 2.4145307540893555, + "learning_rate": 8.282107574094401e-07, + "loss": 1.0313, + "mean_token_accuracy": 0.6872532367706299, + "num_tokens": 38974403.0, + "step": 1510 + }, + { + "epoch": 0.16593454864924226, + "grad_norm": 2.270953416824341, + "learning_rate": 8.287596048298572e-07, + "loss": 0.9882, + "mean_token_accuracy": 0.6989889144897461, + "num_tokens": 38999971.0, + "step": 1511 + }, + { + "epoch": 0.16604436635185593, + "grad_norm": 2.2352051734924316, + "learning_rate": 8.293084522502743e-07, + "loss": 1.0481, + "mean_token_accuracy": 0.6903340816497803, + "num_tokens": 39024265.0, + "step": 1512 + }, + { + "epoch": 0.16615418405446958, + "grad_norm": 2.1541807651519775, + "learning_rate": 8.298572996706915e-07, + "loss": 1.0238, + "mean_token_accuracy": 0.6956077218055725, + "num_tokens": 39054211.0, + "step": 1513 + }, + { + "epoch": 0.16626400175708325, + "grad_norm": 2.2288777828216553, + "learning_rate": 8.304061470911086e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.6705366969108582, + "num_tokens": 39080354.0, + "step": 1514 + }, + { + "epoch": 0.1663738194596969, + "grad_norm": 2.4370999336242676, + "learning_rate": 8.309549945115257e-07, + "loss": 1.009, + "mean_token_accuracy": 0.7010023593902588, + "num_tokens": 39103294.0, + "step": 1515 + }, + { + "epoch": 0.16648363716231057, + "grad_norm": 2.072911500930786, + "learning_rate": 8.315038419319429e-07, + "loss": 0.9281, + "mean_token_accuracy": 0.7194023132324219, + "num_tokens": 39130800.0, + "step": 1516 + }, + { + "epoch": 0.16659345486492422, + "grad_norm": 2.0501551628112793, + "learning_rate": 8.3205268935236e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6956576704978943, + "num_tokens": 39161405.0, + "step": 1517 + }, + { + "epoch": 0.1667032725675379, + "grad_norm": 2.4329774379730225, + "learning_rate": 8.326015367727771e-07, + "loss": 1.1577, + "mean_token_accuracy": 0.660752534866333, + "num_tokens": 39185922.0, + "step": 1518 + }, + { + "epoch": 0.16681309027015154, + "grad_norm": 2.2885940074920654, + "learning_rate": 8.331503841931943e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.683039665222168, + "num_tokens": 39210747.0, + "step": 1519 + }, + { + "epoch": 0.1669229079727652, + "grad_norm": 2.2388694286346436, + "learning_rate": 8.336992316136113e-07, + "loss": 1.049, + "mean_token_accuracy": 0.6870353817939758, + "num_tokens": 39237507.0, + "step": 1520 + }, + { + "epoch": 0.16703272567537888, + "grad_norm": 2.5651142597198486, + "learning_rate": 8.342480790340285e-07, + "loss": 0.9575, + "mean_token_accuracy": 0.7135916948318481, + "num_tokens": 39258252.0, + "step": 1521 + }, + { + "epoch": 0.16714254337799253, + "grad_norm": 2.257844924926758, + "learning_rate": 8.347969264544457e-07, + "loss": 1.0582, + "mean_token_accuracy": 0.6843895316123962, + "num_tokens": 39284687.0, + "step": 1522 + }, + { + "epoch": 0.1672523610806062, + "grad_norm": 2.2897353172302246, + "learning_rate": 8.353457738748627e-07, + "loss": 1.0341, + "mean_token_accuracy": 0.6924951076507568, + "num_tokens": 39309268.0, + "step": 1523 + }, + { + "epoch": 0.16736217878321985, + "grad_norm": 2.224360942840576, + "learning_rate": 8.358946212952799e-07, + "loss": 0.9961, + "mean_token_accuracy": 0.7107425928115845, + "num_tokens": 39335738.0, + "step": 1524 + }, + { + "epoch": 0.16747199648583352, + "grad_norm": 2.067200183868408, + "learning_rate": 8.364434687156971e-07, + "loss": 1.052, + "mean_token_accuracy": 0.6848876476287842, + "num_tokens": 39367223.0, + "step": 1525 + }, + { + "epoch": 0.16758181418844717, + "grad_norm": 2.263737916946411, + "learning_rate": 8.369923161361141e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.6970233917236328, + "num_tokens": 39392101.0, + "step": 1526 + }, + { + "epoch": 0.16769163189106084, + "grad_norm": 2.2227137088775635, + "learning_rate": 8.375411635565313e-07, + "loss": 1.0105, + "mean_token_accuracy": 0.6967648267745972, + "num_tokens": 39418744.0, + "step": 1527 + }, + { + "epoch": 0.1678014495936745, + "grad_norm": 2.324679136276245, + "learning_rate": 8.380900109769485e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.7053209543228149, + "num_tokens": 39444101.0, + "step": 1528 + }, + { + "epoch": 0.16791126729628816, + "grad_norm": 2.1864891052246094, + "learning_rate": 8.386388583973654e-07, + "loss": 1.0663, + "mean_token_accuracy": 0.6978856325149536, + "num_tokens": 39471803.0, + "step": 1529 + }, + { + "epoch": 0.1680210849989018, + "grad_norm": 2.1805708408355713, + "learning_rate": 8.391877058177826e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6798522472381592, + "num_tokens": 39501695.0, + "step": 1530 + }, + { + "epoch": 0.16813090270151548, + "grad_norm": 2.1072938442230225, + "learning_rate": 8.397365532381997e-07, + "loss": 0.9154, + "mean_token_accuracy": 0.730258584022522, + "num_tokens": 39528578.0, + "step": 1531 + }, + { + "epoch": 0.16824072040412916, + "grad_norm": 2.4416136741638184, + "learning_rate": 8.402854006586168e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.69134521484375, + "num_tokens": 39550805.0, + "step": 1532 + }, + { + "epoch": 0.1683505381067428, + "grad_norm": 2.1552698612213135, + "learning_rate": 8.40834248079034e-07, + "loss": 1.0738, + "mean_token_accuracy": 0.6793211698532104, + "num_tokens": 39580560.0, + "step": 1533 + }, + { + "epoch": 0.16846035580935648, + "grad_norm": 2.415684461593628, + "learning_rate": 8.413830954994511e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6857391595840454, + "num_tokens": 39605822.0, + "step": 1534 + }, + { + "epoch": 0.16857017351197012, + "grad_norm": 2.6007161140441895, + "learning_rate": 8.419319429198682e-07, + "loss": 1.044, + "mean_token_accuracy": 0.6867809295654297, + "num_tokens": 39628173.0, + "step": 1535 + }, + { + "epoch": 0.1686799912145838, + "grad_norm": 2.245884656906128, + "learning_rate": 8.424807903402854e-07, + "loss": 0.9727, + "mean_token_accuracy": 0.7075297236442566, + "num_tokens": 39654226.0, + "step": 1536 + }, + { + "epoch": 0.16878980891719744, + "grad_norm": 2.3203561305999756, + "learning_rate": 8.430296377607025e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.6943848133087158, + "num_tokens": 39679787.0, + "step": 1537 + }, + { + "epoch": 0.16889962661981112, + "grad_norm": 2.5229599475860596, + "learning_rate": 8.435784851811196e-07, + "loss": 1.0651, + "mean_token_accuracy": 0.6789913773536682, + "num_tokens": 39702757.0, + "step": 1538 + }, + { + "epoch": 0.16900944432242476, + "grad_norm": 2.3743045330047607, + "learning_rate": 8.441273326015367e-07, + "loss": 1.0533, + "mean_token_accuracy": 0.69473797082901, + "num_tokens": 39728374.0, + "step": 1539 + }, + { + "epoch": 0.16911926202503844, + "grad_norm": 2.0727152824401855, + "learning_rate": 8.446761800219539e-07, + "loss": 1.1216, + "mean_token_accuracy": 0.6638309955596924, + "num_tokens": 39760553.0, + "step": 1540 + }, + { + "epoch": 0.1692290797276521, + "grad_norm": 2.271026372909546, + "learning_rate": 8.45225027442371e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6761590838432312, + "num_tokens": 39786324.0, + "step": 1541 + }, + { + "epoch": 0.16933889743026576, + "grad_norm": 2.317873954772949, + "learning_rate": 8.457738748627881e-07, + "loss": 1.0364, + "mean_token_accuracy": 0.6919193863868713, + "num_tokens": 39810163.0, + "step": 1542 + }, + { + "epoch": 0.16944871513287943, + "grad_norm": 2.2890632152557373, + "learning_rate": 8.463227222832053e-07, + "loss": 0.9695, + "mean_token_accuracy": 0.7074804306030273, + "num_tokens": 39835077.0, + "step": 1543 + }, + { + "epoch": 0.16955853283549308, + "grad_norm": 2.2661828994750977, + "learning_rate": 8.468715697036224e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.6887036561965942, + "num_tokens": 39859618.0, + "step": 1544 + }, + { + "epoch": 0.16966835053810675, + "grad_norm": 2.490210771560669, + "learning_rate": 8.474204171240395e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.698192834854126, + "num_tokens": 39882616.0, + "step": 1545 + }, + { + "epoch": 0.1697781682407204, + "grad_norm": 2.211362361907959, + "learning_rate": 8.479692645444567e-07, + "loss": 0.9628, + "mean_token_accuracy": 0.7065780162811279, + "num_tokens": 39906612.0, + "step": 1546 + }, + { + "epoch": 0.16988798594333407, + "grad_norm": 2.6255156993865967, + "learning_rate": 8.485181119648738e-07, + "loss": 1.0671, + "mean_token_accuracy": 0.6871936321258545, + "num_tokens": 39928953.0, + "step": 1547 + }, + { + "epoch": 0.16999780364594771, + "grad_norm": 2.1993203163146973, + "learning_rate": 8.490669593852908e-07, + "loss": 0.9934, + "mean_token_accuracy": 0.7014377117156982, + "num_tokens": 39955284.0, + "step": 1548 + }, + { + "epoch": 0.1701076213485614, + "grad_norm": 2.2319495677948, + "learning_rate": 8.496158068057079e-07, + "loss": 1.0193, + "mean_token_accuracy": 0.7009878754615784, + "num_tokens": 39981798.0, + "step": 1549 + }, + { + "epoch": 0.17021743905117506, + "grad_norm": 2.328913450241089, + "learning_rate": 8.50164654226125e-07, + "loss": 0.9843, + "mean_token_accuracy": 0.7022035717964172, + "num_tokens": 40006481.0, + "step": 1550 + }, + { + "epoch": 0.1703272567537887, + "grad_norm": 2.5451419353485107, + "learning_rate": 8.507135016465422e-07, + "loss": 1.0246, + "mean_token_accuracy": 0.6923145055770874, + "num_tokens": 40027747.0, + "step": 1551 + }, + { + "epoch": 0.17043707445640238, + "grad_norm": 2.3919475078582764, + "learning_rate": 8.512623490669593e-07, + "loss": 1.025, + "mean_token_accuracy": 0.6992437839508057, + "num_tokens": 40050897.0, + "step": 1552 + }, + { + "epoch": 0.17054689215901603, + "grad_norm": 2.2340118885040283, + "learning_rate": 8.518111964873764e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.682040274143219, + "num_tokens": 40077445.0, + "step": 1553 + }, + { + "epoch": 0.1706567098616297, + "grad_norm": 2.055525064468384, + "learning_rate": 8.523600439077936e-07, + "loss": 1.0018, + "mean_token_accuracy": 0.6935188174247742, + "num_tokens": 40105307.0, + "step": 1554 + }, + { + "epoch": 0.17076652756424335, + "grad_norm": 2.0122482776641846, + "learning_rate": 8.529088913282107e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6726412773132324, + "num_tokens": 40138432.0, + "step": 1555 + }, + { + "epoch": 0.17087634526685702, + "grad_norm": 2.1094231605529785, + "learning_rate": 8.534577387486278e-07, + "loss": 1.0843, + "mean_token_accuracy": 0.6702200174331665, + "num_tokens": 40168324.0, + "step": 1556 + }, + { + "epoch": 0.17098616296947067, + "grad_norm": 2.1541433334350586, + "learning_rate": 8.54006586169045e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.6986199617385864, + "num_tokens": 40196367.0, + "step": 1557 + }, + { + "epoch": 0.17109598067208434, + "grad_norm": 2.1717469692230225, + "learning_rate": 8.545554335894621e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6697202920913696, + "num_tokens": 40225683.0, + "step": 1558 + }, + { + "epoch": 0.17120579837469801, + "grad_norm": 2.182995080947876, + "learning_rate": 8.551042810098792e-07, + "loss": 1.07, + "mean_token_accuracy": 0.6819639801979065, + "num_tokens": 40254672.0, + "step": 1559 + }, + { + "epoch": 0.17131561607731166, + "grad_norm": 2.3142356872558594, + "learning_rate": 8.556531284302964e-07, + "loss": 1.1329, + "mean_token_accuracy": 0.669649064540863, + "num_tokens": 40282907.0, + "step": 1560 + }, + { + "epoch": 0.17142543377992533, + "grad_norm": 2.3619983196258545, + "learning_rate": 8.562019758507134e-07, + "loss": 1.02, + "mean_token_accuracy": 0.7007168531417847, + "num_tokens": 40306398.0, + "step": 1561 + }, + { + "epoch": 0.17153525148253898, + "grad_norm": 2.1837222576141357, + "learning_rate": 8.567508232711306e-07, + "loss": 1.0634, + "mean_token_accuracy": 0.6838825941085815, + "num_tokens": 40335863.0, + "step": 1562 + }, + { + "epoch": 0.17164506918515265, + "grad_norm": 2.3383617401123047, + "learning_rate": 8.572996706915478e-07, + "loss": 0.9623, + "mean_token_accuracy": 0.7105553150177002, + "num_tokens": 40359464.0, + "step": 1563 + }, + { + "epoch": 0.1717548868877663, + "grad_norm": 2.6405069828033447, + "learning_rate": 8.578485181119648e-07, + "loss": 0.9331, + "mean_token_accuracy": 0.7145131230354309, + "num_tokens": 40378046.0, + "step": 1564 + }, + { + "epoch": 0.17186470459037997, + "grad_norm": 2.141557455062866, + "learning_rate": 8.58397365532382e-07, + "loss": 1.0026, + "mean_token_accuracy": 0.6979125738143921, + "num_tokens": 40408110.0, + "step": 1565 + }, + { + "epoch": 0.17197452229299362, + "grad_norm": 2.253441572189331, + "learning_rate": 8.589462129527992e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6746092438697815, + "num_tokens": 40436356.0, + "step": 1566 + }, + { + "epoch": 0.1720843399956073, + "grad_norm": 2.0824174880981445, + "learning_rate": 8.594950603732161e-07, + "loss": 1.035, + "mean_token_accuracy": 0.691656768321991, + "num_tokens": 40468115.0, + "step": 1567 + }, + { + "epoch": 0.17219415769822094, + "grad_norm": 2.5126278400421143, + "learning_rate": 8.600439077936333e-07, + "loss": 0.9566, + "mean_token_accuracy": 0.7093814611434937, + "num_tokens": 40490402.0, + "step": 1568 + }, + { + "epoch": 0.1723039754008346, + "grad_norm": 2.266089916229248, + "learning_rate": 8.605927552140505e-07, + "loss": 1.0, + "mean_token_accuracy": 0.7015167474746704, + "num_tokens": 40515999.0, + "step": 1569 + }, + { + "epoch": 0.1724137931034483, + "grad_norm": 2.5370967388153076, + "learning_rate": 8.611416026344675e-07, + "loss": 0.9927, + "mean_token_accuracy": 0.7021549940109253, + "num_tokens": 40536585.0, + "step": 1570 + }, + { + "epoch": 0.17252361080606193, + "grad_norm": 2.248504400253296, + "learning_rate": 8.616904500548847e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6860519051551819, + "num_tokens": 40564423.0, + "step": 1571 + }, + { + "epoch": 0.1726334285086756, + "grad_norm": 2.179110527038574, + "learning_rate": 8.622392974753018e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.7041779160499573, + "num_tokens": 40590552.0, + "step": 1572 + }, + { + "epoch": 0.17274324621128925, + "grad_norm": 2.169837236404419, + "learning_rate": 8.627881448957189e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6729355454444885, + "num_tokens": 40622008.0, + "step": 1573 + }, + { + "epoch": 0.17285306391390293, + "grad_norm": 2.433488368988037, + "learning_rate": 8.633369923161361e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.6946872472763062, + "num_tokens": 40643804.0, + "step": 1574 + }, + { + "epoch": 0.17296288161651657, + "grad_norm": 2.032456636428833, + "learning_rate": 8.638858397365532e-07, + "loss": 0.9887, + "mean_token_accuracy": 0.7028626203536987, + "num_tokens": 40673398.0, + "step": 1575 + }, + { + "epoch": 0.17307269931913025, + "grad_norm": 2.4987480640411377, + "learning_rate": 8.644346871569703e-07, + "loss": 1.0322, + "mean_token_accuracy": 0.6906416416168213, + "num_tokens": 40696186.0, + "step": 1576 + }, + { + "epoch": 0.1731825170217439, + "grad_norm": 2.2992031574249268, + "learning_rate": 8.649835345773875e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6944969892501831, + "num_tokens": 40720249.0, + "step": 1577 + }, + { + "epoch": 0.17329233472435757, + "grad_norm": 2.3509445190429688, + "learning_rate": 8.655323819978046e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.7039437294006348, + "num_tokens": 40747812.0, + "step": 1578 + }, + { + "epoch": 0.17340215242697124, + "grad_norm": 2.3512415885925293, + "learning_rate": 8.660812294182217e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.6866644620895386, + "num_tokens": 40775158.0, + "step": 1579 + }, + { + "epoch": 0.17351197012958489, + "grad_norm": 2.3652639389038086, + "learning_rate": 8.666300768386389e-07, + "loss": 0.9355, + "mean_token_accuracy": 0.7163830995559692, + "num_tokens": 40798053.0, + "step": 1580 + }, + { + "epoch": 0.17362178783219856, + "grad_norm": 2.3272054195404053, + "learning_rate": 8.67178924259056e-07, + "loss": 1.0587, + "mean_token_accuracy": 0.6845387816429138, + "num_tokens": 40822327.0, + "step": 1581 + }, + { + "epoch": 0.1737316055348122, + "grad_norm": 2.371769428253174, + "learning_rate": 8.677277716794731e-07, + "loss": 1.0246, + "mean_token_accuracy": 0.6923296451568604, + "num_tokens": 40846640.0, + "step": 1582 + }, + { + "epoch": 0.17384142323742588, + "grad_norm": 2.1517481803894043, + "learning_rate": 8.682766190998902e-07, + "loss": 1.1177, + "mean_token_accuracy": 0.6736770868301392, + "num_tokens": 40878902.0, + "step": 1583 + }, + { + "epoch": 0.17395124094003953, + "grad_norm": 2.0208983421325684, + "learning_rate": 8.688254665203073e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6870681643486023, + "num_tokens": 40912303.0, + "step": 1584 + }, + { + "epoch": 0.1740610586426532, + "grad_norm": 2.2691211700439453, + "learning_rate": 8.693743139407245e-07, + "loss": 1.1367, + "mean_token_accuracy": 0.6652393937110901, + "num_tokens": 40940105.0, + "step": 1585 + }, + { + "epoch": 0.17417087634526685, + "grad_norm": 2.36885929107666, + "learning_rate": 8.699231613611415e-07, + "loss": 1.0131, + "mean_token_accuracy": 0.6974443197250366, + "num_tokens": 40963607.0, + "step": 1586 + }, + { + "epoch": 0.17428069404788052, + "grad_norm": 2.244568347930908, + "learning_rate": 8.704720087815586e-07, + "loss": 1.0736, + "mean_token_accuracy": 0.6901675462722778, + "num_tokens": 40989807.0, + "step": 1587 + }, + { + "epoch": 0.1743905117504942, + "grad_norm": 2.436396598815918, + "learning_rate": 8.710208562019758e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.7005980610847473, + "num_tokens": 41014005.0, + "step": 1588 + }, + { + "epoch": 0.17450032945310784, + "grad_norm": 2.1555328369140625, + "learning_rate": 8.715697036223929e-07, + "loss": 1.028, + "mean_token_accuracy": 0.6931924223899841, + "num_tokens": 41040963.0, + "step": 1589 + }, + { + "epoch": 0.1746101471557215, + "grad_norm": 2.2757811546325684, + "learning_rate": 8.7211855104281e-07, + "loss": 1.0179, + "mean_token_accuracy": 0.7010465860366821, + "num_tokens": 41065736.0, + "step": 1590 + }, + { + "epoch": 0.17471996485833516, + "grad_norm": 2.3419222831726074, + "learning_rate": 8.726673984632272e-07, + "loss": 0.9918, + "mean_token_accuracy": 0.7020325064659119, + "num_tokens": 41090932.0, + "step": 1591 + }, + { + "epoch": 0.17482978256094883, + "grad_norm": 2.2006335258483887, + "learning_rate": 8.732162458836443e-07, + "loss": 1.1026, + "mean_token_accuracy": 0.6770550608634949, + "num_tokens": 41118836.0, + "step": 1592 + }, + { + "epoch": 0.17493960026356248, + "grad_norm": 2.2277779579162598, + "learning_rate": 8.737650933040614e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.6948041915893555, + "num_tokens": 41147643.0, + "step": 1593 + }, + { + "epoch": 0.17504941796617615, + "grad_norm": 2.3942782878875732, + "learning_rate": 8.743139407244785e-07, + "loss": 0.9004, + "mean_token_accuracy": 0.722408652305603, + "num_tokens": 41168962.0, + "step": 1594 + }, + { + "epoch": 0.1751592356687898, + "grad_norm": 2.271442174911499, + "learning_rate": 8.748627881448957e-07, + "loss": 1.0411, + "mean_token_accuracy": 0.6895707845687866, + "num_tokens": 41194319.0, + "step": 1595 + }, + { + "epoch": 0.17526905337140347, + "grad_norm": 2.559937000274658, + "learning_rate": 8.754116355653128e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.684408962726593, + "num_tokens": 41215991.0, + "step": 1596 + }, + { + "epoch": 0.17537887107401715, + "grad_norm": 2.443406581878662, + "learning_rate": 8.759604829857299e-07, + "loss": 0.9548, + "mean_token_accuracy": 0.7089594602584839, + "num_tokens": 41239584.0, + "step": 1597 + }, + { + "epoch": 0.1754886887766308, + "grad_norm": 2.3661000728607178, + "learning_rate": 8.765093304061471e-07, + "loss": 1.0024, + "mean_token_accuracy": 0.6993142366409302, + "num_tokens": 41264972.0, + "step": 1598 + }, + { + "epoch": 0.17559850647924446, + "grad_norm": 2.068891763687134, + "learning_rate": 8.770581778265642e-07, + "loss": 1.0499, + "mean_token_accuracy": 0.687492847442627, + "num_tokens": 41293968.0, + "step": 1599 + }, + { + "epoch": 0.1757083241818581, + "grad_norm": 1.9560227394104004, + "learning_rate": 8.776070252469813e-07, + "loss": 0.9872, + "mean_token_accuracy": 0.7000553607940674, + "num_tokens": 41323291.0, + "step": 1600 + }, + { + "epoch": 0.17581814188447178, + "grad_norm": 2.2884521484375, + "learning_rate": 8.781558726673985e-07, + "loss": 1.0233, + "mean_token_accuracy": 0.6958910226821899, + "num_tokens": 41347617.0, + "step": 1601 + }, + { + "epoch": 0.17592795958708543, + "grad_norm": 2.668321132659912, + "learning_rate": 8.787047200878156e-07, + "loss": 0.9659, + "mean_token_accuracy": 0.7085562944412231, + "num_tokens": 41366541.0, + "step": 1602 + }, + { + "epoch": 0.1760377772896991, + "grad_norm": 2.1467487812042236, + "learning_rate": 8.792535675082327e-07, + "loss": 1.1424, + "mean_token_accuracy": 0.6654028296470642, + "num_tokens": 41394344.0, + "step": 1603 + }, + { + "epoch": 0.17614759499231275, + "grad_norm": 2.2593576908111572, + "learning_rate": 8.798024149286499e-07, + "loss": 1.0315, + "mean_token_accuracy": 0.6945195198059082, + "num_tokens": 41420146.0, + "step": 1604 + }, + { + "epoch": 0.17625741269492642, + "grad_norm": 2.3840079307556152, + "learning_rate": 8.803512623490668e-07, + "loss": 0.931, + "mean_token_accuracy": 0.7184087038040161, + "num_tokens": 41443084.0, + "step": 1605 + }, + { + "epoch": 0.17636723039754007, + "grad_norm": 2.0766761302948, + "learning_rate": 8.80900109769484e-07, + "loss": 1.0464, + "mean_token_accuracy": 0.6919640302658081, + "num_tokens": 41473547.0, + "step": 1606 + }, + { + "epoch": 0.17647704810015374, + "grad_norm": 2.5596108436584473, + "learning_rate": 8.814489571899012e-07, + "loss": 1.0201, + "mean_token_accuracy": 0.7002377510070801, + "num_tokens": 41495299.0, + "step": 1607 + }, + { + "epoch": 0.17658686580276742, + "grad_norm": 2.571641683578491, + "learning_rate": 8.819978046103182e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6974678039550781, + "num_tokens": 41516879.0, + "step": 1608 + }, + { + "epoch": 0.17669668350538106, + "grad_norm": 2.2787303924560547, + "learning_rate": 8.825466520307354e-07, + "loss": 1.0917, + "mean_token_accuracy": 0.6840678453445435, + "num_tokens": 41542932.0, + "step": 1609 + }, + { + "epoch": 0.17680650120799474, + "grad_norm": 2.7066283226013184, + "learning_rate": 8.830954994511526e-07, + "loss": 1.0157, + "mean_token_accuracy": 0.6923000812530518, + "num_tokens": 41561673.0, + "step": 1610 + }, + { + "epoch": 0.17691631891060838, + "grad_norm": 2.159583568572998, + "learning_rate": 8.836443468715696e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6822652220726013, + "num_tokens": 41590181.0, + "step": 1611 + }, + { + "epoch": 0.17702613661322206, + "grad_norm": 2.063628911972046, + "learning_rate": 8.841931942919868e-07, + "loss": 0.9742, + "mean_token_accuracy": 0.7033717632293701, + "num_tokens": 41618515.0, + "step": 1612 + }, + { + "epoch": 0.1771359543158357, + "grad_norm": 2.438046932220459, + "learning_rate": 8.84742041712404e-07, + "loss": 0.9704, + "mean_token_accuracy": 0.7048710584640503, + "num_tokens": 41640919.0, + "step": 1613 + }, + { + "epoch": 0.17724577201844938, + "grad_norm": 2.3663899898529053, + "learning_rate": 8.85290889132821e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6858858466148376, + "num_tokens": 41667036.0, + "step": 1614 + }, + { + "epoch": 0.17735558972106302, + "grad_norm": 2.422691583633423, + "learning_rate": 8.858397365532382e-07, + "loss": 1.0126, + "mean_token_accuracy": 0.6979002952575684, + "num_tokens": 41689328.0, + "step": 1615 + }, + { + "epoch": 0.1774654074236767, + "grad_norm": 2.3731184005737305, + "learning_rate": 8.863885839736552e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6796208620071411, + "num_tokens": 41714612.0, + "step": 1616 + }, + { + "epoch": 0.17757522512629037, + "grad_norm": 2.3135409355163574, + "learning_rate": 8.869374313940724e-07, + "loss": 0.9286, + "mean_token_accuracy": 0.7187038660049438, + "num_tokens": 41736777.0, + "step": 1617 + }, + { + "epoch": 0.17768504282890402, + "grad_norm": 2.9877872467041016, + "learning_rate": 8.874862788144896e-07, + "loss": 0.8984, + "mean_token_accuracy": 0.7222148180007935, + "num_tokens": 41753802.0, + "step": 1618 + }, + { + "epoch": 0.1777948605315177, + "grad_norm": 2.394951105117798, + "learning_rate": 8.880351262349066e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6908364295959473, + "num_tokens": 41777849.0, + "step": 1619 + }, + { + "epoch": 0.17790467823413134, + "grad_norm": 2.4119253158569336, + "learning_rate": 8.885839736553238e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.699495792388916, + "num_tokens": 41801041.0, + "step": 1620 + }, + { + "epoch": 0.178014495936745, + "grad_norm": 2.206362247467041, + "learning_rate": 8.89132821075741e-07, + "loss": 1.0574, + "mean_token_accuracy": 0.6844362616539001, + "num_tokens": 41829706.0, + "step": 1621 + }, + { + "epoch": 0.17812431363935866, + "grad_norm": 2.284634590148926, + "learning_rate": 8.89681668496158e-07, + "loss": 1.0792, + "mean_token_accuracy": 0.6770239472389221, + "num_tokens": 41857392.0, + "step": 1622 + }, + { + "epoch": 0.17823413134197233, + "grad_norm": 2.5574238300323486, + "learning_rate": 8.902305159165752e-07, + "loss": 0.999, + "mean_token_accuracy": 0.6958413124084473, + "num_tokens": 41877443.0, + "step": 1623 + }, + { + "epoch": 0.17834394904458598, + "grad_norm": 2.198570966720581, + "learning_rate": 8.907793633369924e-07, + "loss": 1.0686, + "mean_token_accuracy": 0.6870535612106323, + "num_tokens": 41905688.0, + "step": 1624 + }, + { + "epoch": 0.17845376674719965, + "grad_norm": 2.1048641204833984, + "learning_rate": 8.913282107574093e-07, + "loss": 1.0213, + "mean_token_accuracy": 0.6956165432929993, + "num_tokens": 41935276.0, + "step": 1625 + }, + { + "epoch": 0.17856358444981332, + "grad_norm": 2.751657485961914, + "learning_rate": 8.918770581778265e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.6973951458930969, + "num_tokens": 41954763.0, + "step": 1626 + }, + { + "epoch": 0.17867340215242697, + "grad_norm": 2.454104423522949, + "learning_rate": 8.924259055982436e-07, + "loss": 1.0814, + "mean_token_accuracy": 0.6826827526092529, + "num_tokens": 41977894.0, + "step": 1627 + }, + { + "epoch": 0.17878321985504064, + "grad_norm": 2.692263126373291, + "learning_rate": 8.929747530186607e-07, + "loss": 0.9708, + "mean_token_accuracy": 0.7061229348182678, + "num_tokens": 41996461.0, + "step": 1628 + }, + { + "epoch": 0.1788930375576543, + "grad_norm": 2.2068490982055664, + "learning_rate": 8.935236004390779e-07, + "loss": 1.0684, + "mean_token_accuracy": 0.6821140050888062, + "num_tokens": 42021973.0, + "step": 1629 + }, + { + "epoch": 0.17900285526026796, + "grad_norm": 2.2379658222198486, + "learning_rate": 8.94072447859495e-07, + "loss": 1.045, + "mean_token_accuracy": 0.6925119161605835, + "num_tokens": 42050967.0, + "step": 1630 + }, + { + "epoch": 0.1791126729628816, + "grad_norm": 2.285400390625, + "learning_rate": 8.946212952799121e-07, + "loss": 1.067, + "mean_token_accuracy": 0.699396550655365, + "num_tokens": 42076863.0, + "step": 1631 + }, + { + "epoch": 0.17922249066549528, + "grad_norm": 2.587573289871216, + "learning_rate": 8.951701427003293e-07, + "loss": 1.0142, + "mean_token_accuracy": 0.6923685073852539, + "num_tokens": 42097064.0, + "step": 1632 + }, + { + "epoch": 0.17933230836810893, + "grad_norm": 2.122089385986328, + "learning_rate": 8.957189901207464e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.699974775314331, + "num_tokens": 42124216.0, + "step": 1633 + }, + { + "epoch": 0.1794421260707226, + "grad_norm": 2.295403003692627, + "learning_rate": 8.962678375411635e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6800916194915771, + "num_tokens": 42149366.0, + "step": 1634 + }, + { + "epoch": 0.17955194377333628, + "grad_norm": 2.36059832572937, + "learning_rate": 8.968166849615807e-07, + "loss": 0.9527, + "mean_token_accuracy": 0.7126206159591675, + "num_tokens": 42172298.0, + "step": 1635 + }, + { + "epoch": 0.17966176147594992, + "grad_norm": 2.288787603378296, + "learning_rate": 8.973655323819978e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.7026165723800659, + "num_tokens": 42198223.0, + "step": 1636 + }, + { + "epoch": 0.1797715791785636, + "grad_norm": 2.3445661067962646, + "learning_rate": 8.979143798024149e-07, + "loss": 0.9286, + "mean_token_accuracy": 0.7181310653686523, + "num_tokens": 42221880.0, + "step": 1637 + }, + { + "epoch": 0.17988139688117724, + "grad_norm": 2.305954694747925, + "learning_rate": 8.98463227222832e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.7003564834594727, + "num_tokens": 42245260.0, + "step": 1638 + }, + { + "epoch": 0.17999121458379091, + "grad_norm": 2.2379274368286133, + "learning_rate": 8.990120746432492e-07, + "loss": 1.0358, + "mean_token_accuracy": 0.6960079669952393, + "num_tokens": 42270747.0, + "step": 1639 + }, + { + "epoch": 0.18010103228640456, + "grad_norm": 2.313223123550415, + "learning_rate": 8.995609220636663e-07, + "loss": 0.9358, + "mean_token_accuracy": 0.7188229560852051, + "num_tokens": 42296295.0, + "step": 1640 + }, + { + "epoch": 0.18021084998901823, + "grad_norm": 2.231703758239746, + "learning_rate": 9.001097694840834e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6853644847869873, + "num_tokens": 42326320.0, + "step": 1641 + }, + { + "epoch": 0.18032066769163188, + "grad_norm": 2.589890241622925, + "learning_rate": 9.006586169045006e-07, + "loss": 0.8872, + "mean_token_accuracy": 0.7231069803237915, + "num_tokens": 42346733.0, + "step": 1642 + }, + { + "epoch": 0.18043048539424555, + "grad_norm": 1.9553475379943848, + "learning_rate": 9.012074643249177e-07, + "loss": 0.9687, + "mean_token_accuracy": 0.7100467681884766, + "num_tokens": 42380458.0, + "step": 1643 + }, + { + "epoch": 0.1805403030968592, + "grad_norm": 2.1408071517944336, + "learning_rate": 9.017563117453347e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6689395904541016, + "num_tokens": 42411485.0, + "step": 1644 + }, + { + "epoch": 0.18065012079947287, + "grad_norm": 2.283944606781006, + "learning_rate": 9.023051591657519e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6808013319969177, + "num_tokens": 42436245.0, + "step": 1645 + }, + { + "epoch": 0.18075993850208655, + "grad_norm": 2.2473788261413574, + "learning_rate": 9.02854006586169e-07, + "loss": 1.0469, + "mean_token_accuracy": 0.6901454925537109, + "num_tokens": 42462512.0, + "step": 1646 + }, + { + "epoch": 0.1808697562047002, + "grad_norm": 2.4535367488861084, + "learning_rate": 9.034028540065861e-07, + "loss": 1.0861, + "mean_token_accuracy": 0.6724131107330322, + "num_tokens": 42486423.0, + "step": 1647 + }, + { + "epoch": 0.18097957390731387, + "grad_norm": 2.4064340591430664, + "learning_rate": 9.039517014270033e-07, + "loss": 1.1361, + "mean_token_accuracy": 0.6623297333717346, + "num_tokens": 42511737.0, + "step": 1648 + }, + { + "epoch": 0.1810893916099275, + "grad_norm": 2.562270164489746, + "learning_rate": 9.045005488474203e-07, + "loss": 1.0263, + "mean_token_accuracy": 0.6918519139289856, + "num_tokens": 42531443.0, + "step": 1649 + }, + { + "epoch": 0.1811992093125412, + "grad_norm": 2.0750715732574463, + "learning_rate": 9.050493962678375e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.6950163245201111, + "num_tokens": 42561053.0, + "step": 1650 + }, + { + "epoch": 0.18130902701515483, + "grad_norm": 2.5125515460968018, + "learning_rate": 9.055982436882547e-07, + "loss": 1.0011, + "mean_token_accuracy": 0.6933449506759644, + "num_tokens": 42582502.0, + "step": 1651 + }, + { + "epoch": 0.1814188447177685, + "grad_norm": 2.170828104019165, + "learning_rate": 9.061470911086717e-07, + "loss": 1.0713, + "mean_token_accuracy": 0.6799560785293579, + "num_tokens": 42609553.0, + "step": 1652 + }, + { + "epoch": 0.18152866242038215, + "grad_norm": 2.2964284420013428, + "learning_rate": 9.066959385290889e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6793609857559204, + "num_tokens": 42636488.0, + "step": 1653 + }, + { + "epoch": 0.18163848012299583, + "grad_norm": 2.1368024349212646, + "learning_rate": 9.07244785949506e-07, + "loss": 1.0697, + "mean_token_accuracy": 0.6841073632240295, + "num_tokens": 42665730.0, + "step": 1654 + }, + { + "epoch": 0.1817482978256095, + "grad_norm": 2.2508151531219482, + "learning_rate": 9.077936333699231e-07, + "loss": 0.9873, + "mean_token_accuracy": 0.7009097337722778, + "num_tokens": 42692846.0, + "step": 1655 + }, + { + "epoch": 0.18185811552822315, + "grad_norm": 2.210064649581909, + "learning_rate": 9.083424807903403e-07, + "loss": 0.9377, + "mean_token_accuracy": 0.7167496085166931, + "num_tokens": 42718767.0, + "step": 1656 + }, + { + "epoch": 0.18196793323083682, + "grad_norm": 2.500190019607544, + "learning_rate": 9.088913282107573e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6737977862358093, + "num_tokens": 42744701.0, + "step": 1657 + }, + { + "epoch": 0.18207775093345047, + "grad_norm": 2.2938616275787354, + "learning_rate": 9.094401756311745e-07, + "loss": 1.0307, + "mean_token_accuracy": 0.689983606338501, + "num_tokens": 42768831.0, + "step": 1658 + }, + { + "epoch": 0.18218756863606414, + "grad_norm": 2.5533134937286377, + "learning_rate": 9.099890230515917e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6720455884933472, + "num_tokens": 42791344.0, + "step": 1659 + }, + { + "epoch": 0.18229738633867779, + "grad_norm": 2.4255824089050293, + "learning_rate": 9.105378704720087e-07, + "loss": 0.995, + "mean_token_accuracy": 0.7079086899757385, + "num_tokens": 42813585.0, + "step": 1660 + }, + { + "epoch": 0.18240720404129146, + "grad_norm": 2.3461263179779053, + "learning_rate": 9.110867178924259e-07, + "loss": 1.0864, + "mean_token_accuracy": 0.682239294052124, + "num_tokens": 42836722.0, + "step": 1661 + }, + { + "epoch": 0.1825170217439051, + "grad_norm": 2.0129239559173584, + "learning_rate": 9.11635565312843e-07, + "loss": 1.0035, + "mean_token_accuracy": 0.6956268548965454, + "num_tokens": 42865908.0, + "step": 1662 + }, + { + "epoch": 0.18262683944651878, + "grad_norm": 2.6843760013580322, + "learning_rate": 9.1218441273326e-07, + "loss": 0.987, + "mean_token_accuracy": 0.7039322853088379, + "num_tokens": 42885462.0, + "step": 1663 + }, + { + "epoch": 0.18273665714913245, + "grad_norm": 2.3079140186309814, + "learning_rate": 9.127332601536772e-07, + "loss": 0.998, + "mean_token_accuracy": 0.7026190757751465, + "num_tokens": 42909633.0, + "step": 1664 + }, + { + "epoch": 0.1828464748517461, + "grad_norm": 2.4002439975738525, + "learning_rate": 9.132821075740944e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.6941787004470825, + "num_tokens": 42934277.0, + "step": 1665 + }, + { + "epoch": 0.18295629255435977, + "grad_norm": 2.466726541519165, + "learning_rate": 9.138309549945114e-07, + "loss": 1.0438, + "mean_token_accuracy": 0.6898272037506104, + "num_tokens": 42956385.0, + "step": 1666 + }, + { + "epoch": 0.18306611025697342, + "grad_norm": 2.274843215942383, + "learning_rate": 9.143798024149286e-07, + "loss": 0.964, + "mean_token_accuracy": 0.7072206735610962, + "num_tokens": 42982462.0, + "step": 1667 + }, + { + "epoch": 0.1831759279595871, + "grad_norm": 2.2083489894866943, + "learning_rate": 9.149286498353457e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.6993658542633057, + "num_tokens": 43011168.0, + "step": 1668 + }, + { + "epoch": 0.18328574566220074, + "grad_norm": 2.4460582733154297, + "learning_rate": 9.154774972557628e-07, + "loss": 1.0196, + "mean_token_accuracy": 0.6909273862838745, + "num_tokens": 43035302.0, + "step": 1669 + }, + { + "epoch": 0.1833955633648144, + "grad_norm": 2.3667399883270264, + "learning_rate": 9.1602634467618e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6880512237548828, + "num_tokens": 43060561.0, + "step": 1670 + }, + { + "epoch": 0.18350538106742806, + "grad_norm": 2.341306447982788, + "learning_rate": 9.165751920965971e-07, + "loss": 0.9693, + "mean_token_accuracy": 0.698564350605011, + "num_tokens": 43083635.0, + "step": 1671 + }, + { + "epoch": 0.18361519877004173, + "grad_norm": 2.329925537109375, + "learning_rate": 9.171240395170142e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.7005455493927002, + "num_tokens": 43107811.0, + "step": 1672 + }, + { + "epoch": 0.1837250164726554, + "grad_norm": 2.1703219413757324, + "learning_rate": 9.176728869374314e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6779398322105408, + "num_tokens": 43136360.0, + "step": 1673 + }, + { + "epoch": 0.18383483417526905, + "grad_norm": 2.5893242359161377, + "learning_rate": 9.182217343578485e-07, + "loss": 1.0745, + "mean_token_accuracy": 0.6903862357139587, + "num_tokens": 43157334.0, + "step": 1674 + }, + { + "epoch": 0.18394465187788273, + "grad_norm": 2.2413814067840576, + "learning_rate": 9.187705817782656e-07, + "loss": 1.0162, + "mean_token_accuracy": 0.7022289037704468, + "num_tokens": 43182534.0, + "step": 1675 + }, + { + "epoch": 0.18405446958049637, + "grad_norm": 2.32948637008667, + "learning_rate": 9.193194291986828e-07, + "loss": 1.077, + "mean_token_accuracy": 0.6833188533782959, + "num_tokens": 43206832.0, + "step": 1676 + }, + { + "epoch": 0.18416428728311005, + "grad_norm": 2.5963244438171387, + "learning_rate": 9.198682766190999e-07, + "loss": 1.0146, + "mean_token_accuracy": 0.6996656656265259, + "num_tokens": 43228186.0, + "step": 1677 + }, + { + "epoch": 0.1842741049857237, + "grad_norm": 2.1399779319763184, + "learning_rate": 9.20417124039517e-07, + "loss": 1.1293, + "mean_token_accuracy": 0.6676629185676575, + "num_tokens": 43257208.0, + "step": 1678 + }, + { + "epoch": 0.18438392268833736, + "grad_norm": 2.2839603424072266, + "learning_rate": 9.209659714599341e-07, + "loss": 1.077, + "mean_token_accuracy": 0.6819607615470886, + "num_tokens": 43282890.0, + "step": 1679 + }, + { + "epoch": 0.184493740390951, + "grad_norm": 1.954299807548523, + "learning_rate": 9.215148188803513e-07, + "loss": 1.0838, + "mean_token_accuracy": 0.6764792203903198, + "num_tokens": 43314779.0, + "step": 1680 + }, + { + "epoch": 0.18460355809356468, + "grad_norm": 1.8690688610076904, + "learning_rate": 9.220636663007683e-07, + "loss": 1.0209, + "mean_token_accuracy": 0.699183464050293, + "num_tokens": 43351566.0, + "step": 1681 + }, + { + "epoch": 0.18471337579617833, + "grad_norm": 2.465794801712036, + "learning_rate": 9.226125137211854e-07, + "loss": 1.0746, + "mean_token_accuracy": 0.6817312240600586, + "num_tokens": 43375254.0, + "step": 1682 + }, + { + "epoch": 0.184823193498792, + "grad_norm": 1.9885841608047485, + "learning_rate": 9.231613611416026e-07, + "loss": 1.0745, + "mean_token_accuracy": 0.6794755458831787, + "num_tokens": 43406676.0, + "step": 1683 + }, + { + "epoch": 0.18493301120140568, + "grad_norm": 2.441114902496338, + "learning_rate": 9.237102085620197e-07, + "loss": 0.9777, + "mean_token_accuracy": 0.6986331939697266, + "num_tokens": 43429865.0, + "step": 1684 + }, + { + "epoch": 0.18504282890401932, + "grad_norm": 2.242136001586914, + "learning_rate": 9.242590559824368e-07, + "loss": 0.94, + "mean_token_accuracy": 0.7160507440567017, + "num_tokens": 43454990.0, + "step": 1685 + }, + { + "epoch": 0.185152646606633, + "grad_norm": 2.4672136306762695, + "learning_rate": 9.24807903402854e-07, + "loss": 1.0083, + "mean_token_accuracy": 0.705190122127533, + "num_tokens": 43477443.0, + "step": 1686 + }, + { + "epoch": 0.18526246430924664, + "grad_norm": 2.1050221920013428, + "learning_rate": 9.253567508232711e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6896188855171204, + "num_tokens": 43506671.0, + "step": 1687 + }, + { + "epoch": 0.18537228201186032, + "grad_norm": 2.318028211593628, + "learning_rate": 9.259055982436882e-07, + "loss": 1.0418, + "mean_token_accuracy": 0.688679575920105, + "num_tokens": 43530788.0, + "step": 1688 + }, + { + "epoch": 0.18548209971447396, + "grad_norm": 2.3733222484588623, + "learning_rate": 9.264544456641053e-07, + "loss": 0.8882, + "mean_token_accuracy": 0.7310090065002441, + "num_tokens": 43553643.0, + "step": 1689 + }, + { + "epoch": 0.18559191741708764, + "grad_norm": 2.546452045440674, + "learning_rate": 9.270032930845224e-07, + "loss": 1.0138, + "mean_token_accuracy": 0.6955157518386841, + "num_tokens": 43574947.0, + "step": 1690 + }, + { + "epoch": 0.18570173511970128, + "grad_norm": 2.26637601852417, + "learning_rate": 9.275521405049396e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6962921023368835, + "num_tokens": 43599924.0, + "step": 1691 + }, + { + "epoch": 0.18581155282231496, + "grad_norm": 2.5472843647003174, + "learning_rate": 9.281009879253567e-07, + "loss": 1.011, + "mean_token_accuracy": 0.6929038763046265, + "num_tokens": 43622793.0, + "step": 1692 + }, + { + "epoch": 0.18592137052492863, + "grad_norm": 1.9922900199890137, + "learning_rate": 9.286498353457738e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.7093073129653931, + "num_tokens": 43653375.0, + "step": 1693 + }, + { + "epoch": 0.18603118822754228, + "grad_norm": 2.1048364639282227, + "learning_rate": 9.29198682766191e-07, + "loss": 1.0217, + "mean_token_accuracy": 0.6922602653503418, + "num_tokens": 43681645.0, + "step": 1694 + }, + { + "epoch": 0.18614100593015595, + "grad_norm": 2.1812291145324707, + "learning_rate": 9.297475301866081e-07, + "loss": 1.0458, + "mean_token_accuracy": 0.6853310465812683, + "num_tokens": 43709369.0, + "step": 1695 + }, + { + "epoch": 0.1862508236327696, + "grad_norm": 2.517477512359619, + "learning_rate": 9.302963776070252e-07, + "loss": 1.041, + "mean_token_accuracy": 0.687218427658081, + "num_tokens": 43731446.0, + "step": 1696 + }, + { + "epoch": 0.18636064133538327, + "grad_norm": 2.463020086288452, + "learning_rate": 9.308452250274424e-07, + "loss": 1.0434, + "mean_token_accuracy": 0.6914385557174683, + "num_tokens": 43753594.0, + "step": 1697 + }, + { + "epoch": 0.18647045903799692, + "grad_norm": 2.358234167098999, + "learning_rate": 9.313940724478595e-07, + "loss": 1.0787, + "mean_token_accuracy": 0.6759787201881409, + "num_tokens": 43777275.0, + "step": 1698 + }, + { + "epoch": 0.1865802767406106, + "grad_norm": 2.5061421394348145, + "learning_rate": 9.319429198682766e-07, + "loss": 0.9826, + "mean_token_accuracy": 0.7009893655776978, + "num_tokens": 43797983.0, + "step": 1699 + }, + { + "epoch": 0.18669009444322424, + "grad_norm": 2.128175735473633, + "learning_rate": 9.324917672886937e-07, + "loss": 1.0892, + "mean_token_accuracy": 0.6711894273757935, + "num_tokens": 43826522.0, + "step": 1700 + }, + { + "epoch": 0.1867999121458379, + "grad_norm": 2.279371976852417, + "learning_rate": 9.330406147091107e-07, + "loss": 1.0421, + "mean_token_accuracy": 0.6820744276046753, + "num_tokens": 43852848.0, + "step": 1701 + }, + { + "epoch": 0.18690972984845158, + "grad_norm": 2.3649017810821533, + "learning_rate": 9.335894621295279e-07, + "loss": 1.0481, + "mean_token_accuracy": 0.6987453699111938, + "num_tokens": 43877615.0, + "step": 1702 + }, + { + "epoch": 0.18701954755106523, + "grad_norm": 2.1713321208953857, + "learning_rate": 9.341383095499451e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.696895182132721, + "num_tokens": 43905303.0, + "step": 1703 + }, + { + "epoch": 0.1871293652536789, + "grad_norm": 2.4680347442626953, + "learning_rate": 9.346871569703621e-07, + "loss": 0.9765, + "mean_token_accuracy": 0.702614426612854, + "num_tokens": 43927265.0, + "step": 1704 + }, + { + "epoch": 0.18723918295629255, + "grad_norm": 2.339266061782837, + "learning_rate": 9.352360043907793e-07, + "loss": 1.0885, + "mean_token_accuracy": 0.68052077293396, + "num_tokens": 43950804.0, + "step": 1705 + }, + { + "epoch": 0.18734900065890622, + "grad_norm": 2.3745977878570557, + "learning_rate": 9.357848518111965e-07, + "loss": 0.938, + "mean_token_accuracy": 0.7103968262672424, + "num_tokens": 43976172.0, + "step": 1706 + }, + { + "epoch": 0.18745881836151987, + "grad_norm": 2.078930139541626, + "learning_rate": 9.363336992316135e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6825351715087891, + "num_tokens": 44007773.0, + "step": 1707 + }, + { + "epoch": 0.18756863606413354, + "grad_norm": 2.0309290885925293, + "learning_rate": 9.368825466520307e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.7098712921142578, + "num_tokens": 44038701.0, + "step": 1708 + }, + { + "epoch": 0.1876784537667472, + "grad_norm": 2.1676905155181885, + "learning_rate": 9.374313940724479e-07, + "loss": 1.0693, + "mean_token_accuracy": 0.680550217628479, + "num_tokens": 44066225.0, + "step": 1709 + }, + { + "epoch": 0.18778827146936086, + "grad_norm": 2.950502395629883, + "learning_rate": 9.379802414928649e-07, + "loss": 0.9798, + "mean_token_accuracy": 0.7012145519256592, + "num_tokens": 44084790.0, + "step": 1710 + }, + { + "epoch": 0.18789808917197454, + "grad_norm": 2.257826328277588, + "learning_rate": 9.385290889132821e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6681603193283081, + "num_tokens": 44114912.0, + "step": 1711 + }, + { + "epoch": 0.18800790687458818, + "grad_norm": 2.2715747356414795, + "learning_rate": 9.390779363336992e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6848720908164978, + "num_tokens": 44140965.0, + "step": 1712 + }, + { + "epoch": 0.18811772457720186, + "grad_norm": 2.271744966506958, + "learning_rate": 9.396267837541163e-07, + "loss": 1.0436, + "mean_token_accuracy": 0.687821626663208, + "num_tokens": 44166913.0, + "step": 1713 + }, + { + "epoch": 0.1882275422798155, + "grad_norm": 2.0179576873779297, + "learning_rate": 9.401756311745335e-07, + "loss": 1.0465, + "mean_token_accuracy": 0.6934764981269836, + "num_tokens": 44200716.0, + "step": 1714 + }, + { + "epoch": 0.18833735998242918, + "grad_norm": 2.2182328701019287, + "learning_rate": 9.407244785949506e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.6968796253204346, + "num_tokens": 44226943.0, + "step": 1715 + }, + { + "epoch": 0.18844717768504282, + "grad_norm": 2.156992197036743, + "learning_rate": 9.412733260153677e-07, + "loss": 1.0716, + "mean_token_accuracy": 0.6828425526618958, + "num_tokens": 44255191.0, + "step": 1716 + }, + { + "epoch": 0.1885569953876565, + "grad_norm": 2.2611265182495117, + "learning_rate": 9.418221734357849e-07, + "loss": 0.9413, + "mean_token_accuracy": 0.717781662940979, + "num_tokens": 44279809.0, + "step": 1717 + }, + { + "epoch": 0.18866681309027014, + "grad_norm": 2.0364601612091064, + "learning_rate": 9.42371020856202e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.6999962329864502, + "num_tokens": 44309336.0, + "step": 1718 + }, + { + "epoch": 0.18877663079288381, + "grad_norm": 2.654306650161743, + "learning_rate": 9.42919868276619e-07, + "loss": 1.0253, + "mean_token_accuracy": 0.6868139505386353, + "num_tokens": 44328942.0, + "step": 1719 + }, + { + "epoch": 0.18888644849549746, + "grad_norm": 2.0440711975097656, + "learning_rate": 9.434687156970362e-07, + "loss": 0.9718, + "mean_token_accuracy": 0.6936913728713989, + "num_tokens": 44357525.0, + "step": 1720 + }, + { + "epoch": 0.18899626619811113, + "grad_norm": 2.147383689880371, + "learning_rate": 9.440175631174532e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.6975008249282837, + "num_tokens": 44384731.0, + "step": 1721 + }, + { + "epoch": 0.1891060839007248, + "grad_norm": 2.233344793319702, + "learning_rate": 9.445664105378704e-07, + "loss": 1.0567, + "mean_token_accuracy": 0.6803296804428101, + "num_tokens": 44412387.0, + "step": 1722 + }, + { + "epoch": 0.18921590160333845, + "grad_norm": 2.0785701274871826, + "learning_rate": 9.451152579582875e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.6943598985671997, + "num_tokens": 44441943.0, + "step": 1723 + }, + { + "epoch": 0.18932571930595213, + "grad_norm": 2.123408555984497, + "learning_rate": 9.456641053787046e-07, + "loss": 0.9302, + "mean_token_accuracy": 0.7073159217834473, + "num_tokens": 44467853.0, + "step": 1724 + }, + { + "epoch": 0.18943553700856577, + "grad_norm": 2.362886428833008, + "learning_rate": 9.462129527991218e-07, + "loss": 1.0509, + "mean_token_accuracy": 0.6872820854187012, + "num_tokens": 44491601.0, + "step": 1725 + }, + { + "epoch": 0.18954535471117945, + "grad_norm": 2.1635937690734863, + "learning_rate": 9.467618002195389e-07, + "loss": 1.0867, + "mean_token_accuracy": 0.6885945796966553, + "num_tokens": 44521305.0, + "step": 1726 + }, + { + "epoch": 0.1896551724137931, + "grad_norm": 2.307138681411743, + "learning_rate": 9.47310647639956e-07, + "loss": 0.9398, + "mean_token_accuracy": 0.7132804989814758, + "num_tokens": 44547269.0, + "step": 1727 + }, + { + "epoch": 0.18976499011640677, + "grad_norm": 2.189671039581299, + "learning_rate": 9.478594950603732e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6835752129554749, + "num_tokens": 44573687.0, + "step": 1728 + }, + { + "epoch": 0.1898748078190204, + "grad_norm": 2.466930866241455, + "learning_rate": 9.484083424807903e-07, + "loss": 0.984, + "mean_token_accuracy": 0.7003674507141113, + "num_tokens": 44595900.0, + "step": 1729 + }, + { + "epoch": 0.1899846255216341, + "grad_norm": 2.0449717044830322, + "learning_rate": 9.489571899012074e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6893197298049927, + "num_tokens": 44627130.0, + "step": 1730 + }, + { + "epoch": 0.19009444322424776, + "grad_norm": 2.014878749847412, + "learning_rate": 9.495060373216246e-07, + "loss": 0.9708, + "mean_token_accuracy": 0.7039527297019958, + "num_tokens": 44657497.0, + "step": 1731 + }, + { + "epoch": 0.1902042609268614, + "grad_norm": 2.2883903980255127, + "learning_rate": 9.500548847420417e-07, + "loss": 0.9937, + "mean_token_accuracy": 0.6979398727416992, + "num_tokens": 44682194.0, + "step": 1732 + }, + { + "epoch": 0.19031407862947508, + "grad_norm": 2.4096295833587646, + "learning_rate": 9.506037321624588e-07, + "loss": 0.9479, + "mean_token_accuracy": 0.708025336265564, + "num_tokens": 44704394.0, + "step": 1733 + }, + { + "epoch": 0.19042389633208873, + "grad_norm": 2.4798758029937744, + "learning_rate": 9.511525795828759e-07, + "loss": 1.0366, + "mean_token_accuracy": 0.6901649236679077, + "num_tokens": 44727524.0, + "step": 1734 + }, + { + "epoch": 0.1905337140347024, + "grad_norm": 2.005451202392578, + "learning_rate": 9.517014270032931e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6804448366165161, + "num_tokens": 44759351.0, + "step": 1735 + }, + { + "epoch": 0.19064353173731605, + "grad_norm": 2.527045249938965, + "learning_rate": 9.522502744237102e-07, + "loss": 0.9511, + "mean_token_accuracy": 0.7126365303993225, + "num_tokens": 44781075.0, + "step": 1736 + }, + { + "epoch": 0.19075334943992972, + "grad_norm": 1.9982807636260986, + "learning_rate": 9.527991218441273e-07, + "loss": 1.0725, + "mean_token_accuracy": 0.679735541343689, + "num_tokens": 44815107.0, + "step": 1737 + }, + { + "epoch": 0.19086316714254337, + "grad_norm": 2.0696592330932617, + "learning_rate": 9.533479692645444e-07, + "loss": 1.0868, + "mean_token_accuracy": 0.6745407581329346, + "num_tokens": 44848271.0, + "step": 1738 + }, + { + "epoch": 0.19097298484515704, + "grad_norm": 2.410335063934326, + "learning_rate": 9.538968166849616e-07, + "loss": 0.9326, + "mean_token_accuracy": 0.7133879065513611, + "num_tokens": 44871409.0, + "step": 1739 + }, + { + "epoch": 0.1910828025477707, + "grad_norm": 2.568007707595825, + "learning_rate": 9.544456641053787e-07, + "loss": 0.9588, + "mean_token_accuracy": 0.7096320986747742, + "num_tokens": 44890792.0, + "step": 1740 + }, + { + "epoch": 0.19119262025038436, + "grad_norm": 2.240480899810791, + "learning_rate": 9.549945115257958e-07, + "loss": 1.0893, + "mean_token_accuracy": 0.674868106842041, + "num_tokens": 44919403.0, + "step": 1741 + }, + { + "epoch": 0.19130243795299803, + "grad_norm": 2.2698237895965576, + "learning_rate": 9.55543358946213e-07, + "loss": 0.9796, + "mean_token_accuracy": 0.7009139060974121, + "num_tokens": 44945169.0, + "step": 1742 + }, + { + "epoch": 0.19141225565561168, + "grad_norm": 1.999674916267395, + "learning_rate": 9.5609220636663e-07, + "loss": 1.0366, + "mean_token_accuracy": 0.683491587638855, + "num_tokens": 44977204.0, + "step": 1743 + }, + { + "epoch": 0.19152207335822535, + "grad_norm": 2.2697246074676514, + "learning_rate": 9.566410537870472e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6763907670974731, + "num_tokens": 45002524.0, + "step": 1744 + }, + { + "epoch": 0.191631891060839, + "grad_norm": 2.0869522094726562, + "learning_rate": 9.571899012074642e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.6728897094726562, + "num_tokens": 45033611.0, + "step": 1745 + }, + { + "epoch": 0.19174170876345267, + "grad_norm": 2.3082728385925293, + "learning_rate": 9.577387486278815e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.6942746639251709, + "num_tokens": 45057875.0, + "step": 1746 + }, + { + "epoch": 0.19185152646606632, + "grad_norm": 2.362715721130371, + "learning_rate": 9.582875960482986e-07, + "loss": 0.9106, + "mean_token_accuracy": 0.7184244990348816, + "num_tokens": 45080485.0, + "step": 1747 + }, + { + "epoch": 0.19196134416868, + "grad_norm": 2.1690914630889893, + "learning_rate": 9.588364434687156e-07, + "loss": 1.098, + "mean_token_accuracy": 0.6818122863769531, + "num_tokens": 45108899.0, + "step": 1748 + }, + { + "epoch": 0.19207116187129367, + "grad_norm": 2.5197513103485107, + "learning_rate": 9.593852908891327e-07, + "loss": 0.9095, + "mean_token_accuracy": 0.7159754037857056, + "num_tokens": 45128974.0, + "step": 1749 + }, + { + "epoch": 0.1921809795739073, + "grad_norm": 2.2493464946746826, + "learning_rate": 9.5993413830955e-07, + "loss": 0.9868, + "mean_token_accuracy": 0.6991894245147705, + "num_tokens": 45152882.0, + "step": 1750 + }, + { + "epoch": 0.19229079727652099, + "grad_norm": 2.4180941581726074, + "learning_rate": 9.60482985729967e-07, + "loss": 1.0627, + "mean_token_accuracy": 0.6843734979629517, + "num_tokens": 45175466.0, + "step": 1751 + }, + { + "epoch": 0.19240061497913463, + "grad_norm": 2.238053321838379, + "learning_rate": 9.61031833150384e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.6974472999572754, + "num_tokens": 45201907.0, + "step": 1752 + }, + { + "epoch": 0.1925104326817483, + "grad_norm": 2.534090995788574, + "learning_rate": 9.615806805708014e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.6948448419570923, + "num_tokens": 45224701.0, + "step": 1753 + }, + { + "epoch": 0.19262025038436195, + "grad_norm": 2.0736286640167236, + "learning_rate": 9.621295279912184e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6775467395782471, + "num_tokens": 45253739.0, + "step": 1754 + }, + { + "epoch": 0.19273006808697563, + "grad_norm": 2.311882257461548, + "learning_rate": 9.626783754116355e-07, + "loss": 0.9976, + "mean_token_accuracy": 0.7041299343109131, + "num_tokens": 45283637.0, + "step": 1755 + }, + { + "epoch": 0.19283988578958927, + "grad_norm": 2.6508872509002686, + "learning_rate": 9.632272228320525e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.6999092102050781, + "num_tokens": 45302289.0, + "step": 1756 + }, + { + "epoch": 0.19294970349220295, + "grad_norm": 2.0980007648468018, + "learning_rate": 9.637760702524698e-07, + "loss": 1.0615, + "mean_token_accuracy": 0.6825152039527893, + "num_tokens": 45331376.0, + "step": 1757 + }, + { + "epoch": 0.1930595211948166, + "grad_norm": 2.436628818511963, + "learning_rate": 9.643249176728869e-07, + "loss": 1.0333, + "mean_token_accuracy": 0.6893406510353088, + "num_tokens": 45356785.0, + "step": 1758 + }, + { + "epoch": 0.19316933889743026, + "grad_norm": 2.2448630332946777, + "learning_rate": 9.64873765093304e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.6865594387054443, + "num_tokens": 45383799.0, + "step": 1759 + }, + { + "epoch": 0.19327915660004394, + "grad_norm": 2.2276501655578613, + "learning_rate": 9.654226125137212e-07, + "loss": 1.0394, + "mean_token_accuracy": 0.689038872718811, + "num_tokens": 45411235.0, + "step": 1760 + }, + { + "epoch": 0.19338897430265758, + "grad_norm": 2.240757465362549, + "learning_rate": 9.659714599341383e-07, + "loss": 1.1131, + "mean_token_accuracy": 0.6792093515396118, + "num_tokens": 45440392.0, + "step": 1761 + }, + { + "epoch": 0.19349879200527126, + "grad_norm": 2.006976842880249, + "learning_rate": 9.665203073545553e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.6754441857337952, + "num_tokens": 45472615.0, + "step": 1762 + }, + { + "epoch": 0.1936086097078849, + "grad_norm": 2.460022211074829, + "learning_rate": 9.670691547749726e-07, + "loss": 1.0669, + "mean_token_accuracy": 0.6760924458503723, + "num_tokens": 45497962.0, + "step": 1763 + }, + { + "epoch": 0.19371842741049858, + "grad_norm": 2.2798938751220703, + "learning_rate": 9.676180021953897e-07, + "loss": 1.0316, + "mean_token_accuracy": 0.6887174248695374, + "num_tokens": 45526229.0, + "step": 1764 + }, + { + "epoch": 0.19382824511311222, + "grad_norm": 2.309447765350342, + "learning_rate": 9.681668496158067e-07, + "loss": 0.9546, + "mean_token_accuracy": 0.7132388353347778, + "num_tokens": 45550201.0, + "step": 1765 + }, + { + "epoch": 0.1939380628157259, + "grad_norm": 2.658531665802002, + "learning_rate": 9.68715697036224e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.6911799907684326, + "num_tokens": 45569954.0, + "step": 1766 + }, + { + "epoch": 0.19404788051833954, + "grad_norm": 2.7099642753601074, + "learning_rate": 9.692645444566409e-07, + "loss": 1.0693, + "mean_token_accuracy": 0.6838043928146362, + "num_tokens": 45590168.0, + "step": 1767 + }, + { + "epoch": 0.19415769822095322, + "grad_norm": 2.356337308883667, + "learning_rate": 9.698133918770581e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.6939747333526611, + "num_tokens": 45613494.0, + "step": 1768 + }, + { + "epoch": 0.1942675159235669, + "grad_norm": 2.383697748184204, + "learning_rate": 9.703622392974752e-07, + "loss": 1.0466, + "mean_token_accuracy": 0.689549446105957, + "num_tokens": 45636528.0, + "step": 1769 + }, + { + "epoch": 0.19437733362618054, + "grad_norm": 2.305556535720825, + "learning_rate": 9.709110867178923e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.7069674730300903, + "num_tokens": 45659793.0, + "step": 1770 + }, + { + "epoch": 0.1944871513287942, + "grad_norm": 2.4084770679473877, + "learning_rate": 9.714599341383095e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6784930229187012, + "num_tokens": 45683892.0, + "step": 1771 + }, + { + "epoch": 0.19459696903140786, + "grad_norm": 1.952896237373352, + "learning_rate": 9.720087815587266e-07, + "loss": 0.9971, + "mean_token_accuracy": 0.6980413198471069, + "num_tokens": 45716166.0, + "step": 1772 + }, + { + "epoch": 0.19470678673402153, + "grad_norm": 2.5627224445343018, + "learning_rate": 9.725576289791437e-07, + "loss": 1.0639, + "mean_token_accuracy": 0.6790552735328674, + "num_tokens": 45738754.0, + "step": 1773 + }, + { + "epoch": 0.19481660443663518, + "grad_norm": 2.2147610187530518, + "learning_rate": 9.73106476399561e-07, + "loss": 1.0101, + "mean_token_accuracy": 0.6903286576271057, + "num_tokens": 45765621.0, + "step": 1774 + }, + { + "epoch": 0.19492642213924885, + "grad_norm": 2.3608741760253906, + "learning_rate": 9.73655323819978e-07, + "loss": 1.0126, + "mean_token_accuracy": 0.6929747462272644, + "num_tokens": 45791959.0, + "step": 1775 + }, + { + "epoch": 0.1950362398418625, + "grad_norm": 2.516702651977539, + "learning_rate": 9.74204171240395e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7041953802108765, + "num_tokens": 45813770.0, + "step": 1776 + }, + { + "epoch": 0.19514605754447617, + "grad_norm": 1.9597358703613281, + "learning_rate": 9.747530186608123e-07, + "loss": 1.1144, + "mean_token_accuracy": 0.6664426326751709, + "num_tokens": 45849857.0, + "step": 1777 + }, + { + "epoch": 0.19525587524708984, + "grad_norm": 2.1154699325561523, + "learning_rate": 9.753018660812294e-07, + "loss": 1.1064, + "mean_token_accuracy": 0.6679712533950806, + "num_tokens": 45878837.0, + "step": 1778 + }, + { + "epoch": 0.1953656929497035, + "grad_norm": 2.2845077514648438, + "learning_rate": 9.758507135016465e-07, + "loss": 1.0168, + "mean_token_accuracy": 0.6912612915039062, + "num_tokens": 45903485.0, + "step": 1779 + }, + { + "epoch": 0.19547551065231716, + "grad_norm": 2.2405049800872803, + "learning_rate": 9.763995609220637e-07, + "loss": 1.057, + "mean_token_accuracy": 0.681220531463623, + "num_tokens": 45929522.0, + "step": 1780 + }, + { + "epoch": 0.1955853283549308, + "grad_norm": 1.9993876218795776, + "learning_rate": 9.769484083424808e-07, + "loss": 1.063, + "mean_token_accuracy": 0.6847754120826721, + "num_tokens": 45962412.0, + "step": 1781 + }, + { + "epoch": 0.19569514605754448, + "grad_norm": 2.365528106689453, + "learning_rate": 9.774972557628979e-07, + "loss": 1.1447, + "mean_token_accuracy": 0.6623632907867432, + "num_tokens": 45987160.0, + "step": 1782 + }, + { + "epoch": 0.19580496376015813, + "grad_norm": 2.1733856201171875, + "learning_rate": 9.780461031833151e-07, + "loss": 1.0154, + "mean_token_accuracy": 0.6949706077575684, + "num_tokens": 46014500.0, + "step": 1783 + }, + { + "epoch": 0.1959147814627718, + "grad_norm": 2.6423163414001465, + "learning_rate": 9.785949506037322e-07, + "loss": 0.9869, + "mean_token_accuracy": 0.6970547437667847, + "num_tokens": 46034810.0, + "step": 1784 + }, + { + "epoch": 0.19602459916538545, + "grad_norm": 2.0723047256469727, + "learning_rate": 9.791437980241493e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.704620361328125, + "num_tokens": 46063633.0, + "step": 1785 + }, + { + "epoch": 0.19613441686799912, + "grad_norm": 2.532423973083496, + "learning_rate": 9.796926454445663e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6891391277313232, + "num_tokens": 46085733.0, + "step": 1786 + }, + { + "epoch": 0.1962442345706128, + "grad_norm": 2.5595154762268066, + "learning_rate": 9.802414928649834e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.6877597570419312, + "num_tokens": 46108145.0, + "step": 1787 + }, + { + "epoch": 0.19635405227322644, + "grad_norm": 2.2370331287384033, + "learning_rate": 9.807903402854007e-07, + "loss": 0.9498, + "mean_token_accuracy": 0.7197410464286804, + "num_tokens": 46134578.0, + "step": 1788 + }, + { + "epoch": 0.19646386997584012, + "grad_norm": 2.074016809463501, + "learning_rate": 9.813391877058177e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.6853992938995361, + "num_tokens": 46166496.0, + "step": 1789 + }, + { + "epoch": 0.19657368767845376, + "grad_norm": 2.4586727619171143, + "learning_rate": 9.818880351262348e-07, + "loss": 0.9418, + "mean_token_accuracy": 0.7119213342666626, + "num_tokens": 46188278.0, + "step": 1790 + }, + { + "epoch": 0.19668350538106744, + "grad_norm": 2.3738205432891846, + "learning_rate": 9.82436882546652e-07, + "loss": 1.0726, + "mean_token_accuracy": 0.6772580146789551, + "num_tokens": 46213534.0, + "step": 1791 + }, + { + "epoch": 0.19679332308368108, + "grad_norm": 2.2197515964508057, + "learning_rate": 9.829857299670691e-07, + "loss": 1.0389, + "mean_token_accuracy": 0.6945881843566895, + "num_tokens": 46241448.0, + "step": 1792 + }, + { + "epoch": 0.19690314078629476, + "grad_norm": 2.235722780227661, + "learning_rate": 9.835345773874862e-07, + "loss": 0.9728, + "mean_token_accuracy": 0.7046009302139282, + "num_tokens": 46266735.0, + "step": 1793 + }, + { + "epoch": 0.1970129584889084, + "grad_norm": 2.1568596363067627, + "learning_rate": 9.840834248079035e-07, + "loss": 0.9905, + "mean_token_accuracy": 0.7111674547195435, + "num_tokens": 46297094.0, + "step": 1794 + }, + { + "epoch": 0.19712277619152208, + "grad_norm": 2.5511844158172607, + "learning_rate": 9.846322722283205e-07, + "loss": 0.9313, + "mean_token_accuracy": 0.7170208692550659, + "num_tokens": 46318225.0, + "step": 1795 + }, + { + "epoch": 0.19723259389413572, + "grad_norm": 2.3274855613708496, + "learning_rate": 9.851811196487376e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.683369517326355, + "num_tokens": 46342297.0, + "step": 1796 + }, + { + "epoch": 0.1973424115967494, + "grad_norm": 2.521421432495117, + "learning_rate": 9.857299670691546e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.6852854490280151, + "num_tokens": 46362929.0, + "step": 1797 + }, + { + "epoch": 0.19745222929936307, + "grad_norm": 2.1845180988311768, + "learning_rate": 9.86278814489572e-07, + "loss": 0.9782, + "mean_token_accuracy": 0.7049036622047424, + "num_tokens": 46388739.0, + "step": 1798 + }, + { + "epoch": 0.19756204700197671, + "grad_norm": 2.0451455116271973, + "learning_rate": 9.86827661909989e-07, + "loss": 1.0816, + "mean_token_accuracy": 0.678737998008728, + "num_tokens": 46419803.0, + "step": 1799 + }, + { + "epoch": 0.1976718647045904, + "grad_norm": 2.255136728286743, + "learning_rate": 9.87376509330406e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.6953140497207642, + "num_tokens": 46447311.0, + "step": 1800 + }, + { + "epoch": 0.19778168240720403, + "grad_norm": 2.5047364234924316, + "learning_rate": 9.879253567508233e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.6946104764938354, + "num_tokens": 46471371.0, + "step": 1801 + }, + { + "epoch": 0.1978915001098177, + "grad_norm": 2.4929795265197754, + "learning_rate": 9.884742041712404e-07, + "loss": 0.9688, + "mean_token_accuracy": 0.7074103951454163, + "num_tokens": 46491463.0, + "step": 1802 + }, + { + "epoch": 0.19800131781243135, + "grad_norm": 2.271885871887207, + "learning_rate": 9.890230515916574e-07, + "loss": 1.0121, + "mean_token_accuracy": 0.6959260106086731, + "num_tokens": 46517032.0, + "step": 1803 + }, + { + "epoch": 0.19811113551504503, + "grad_norm": 2.3315794467926025, + "learning_rate": 9.895718990120747e-07, + "loss": 0.9737, + "mean_token_accuracy": 0.7055623531341553, + "num_tokens": 46543079.0, + "step": 1804 + }, + { + "epoch": 0.19822095321765867, + "grad_norm": 2.2657716274261475, + "learning_rate": 9.901207464324918e-07, + "loss": 0.9787, + "mean_token_accuracy": 0.7089430093765259, + "num_tokens": 46567825.0, + "step": 1805 + }, + { + "epoch": 0.19833077092027235, + "grad_norm": 2.2918756008148193, + "learning_rate": 9.906695938529088e-07, + "loss": 0.9576, + "mean_token_accuracy": 0.7160919904708862, + "num_tokens": 46592688.0, + "step": 1806 + }, + { + "epoch": 0.19844058862288602, + "grad_norm": 2.0058631896972656, + "learning_rate": 9.91218441273326e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.6944229602813721, + "num_tokens": 46624939.0, + "step": 1807 + }, + { + "epoch": 0.19855040632549967, + "grad_norm": 2.3404932022094727, + "learning_rate": 9.91767288693743e-07, + "loss": 1.1044, + "mean_token_accuracy": 0.691508948802948, + "num_tokens": 46651629.0, + "step": 1808 + }, + { + "epoch": 0.19866022402811334, + "grad_norm": 2.551912546157837, + "learning_rate": 9.923161361141602e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6870614290237427, + "num_tokens": 46673414.0, + "step": 1809 + }, + { + "epoch": 0.198770041730727, + "grad_norm": 2.3313992023468018, + "learning_rate": 9.928649835345773e-07, + "loss": 1.0316, + "mean_token_accuracy": 0.6911368370056152, + "num_tokens": 46700215.0, + "step": 1810 + }, + { + "epoch": 0.19887985943334066, + "grad_norm": 2.368624687194824, + "learning_rate": 9.934138309549944e-07, + "loss": 1.0452, + "mean_token_accuracy": 0.6818879842758179, + "num_tokens": 46726169.0, + "step": 1811 + }, + { + "epoch": 0.1989896771359543, + "grad_norm": 2.4918479919433594, + "learning_rate": 9.939626783754116e-07, + "loss": 0.9639, + "mean_token_accuracy": 0.7100493907928467, + "num_tokens": 46747948.0, + "step": 1812 + }, + { + "epoch": 0.19909949483856798, + "grad_norm": 2.2679171562194824, + "learning_rate": 9.945115257958287e-07, + "loss": 1.0702, + "mean_token_accuracy": 0.6845448017120361, + "num_tokens": 46773619.0, + "step": 1813 + }, + { + "epoch": 0.19920931254118163, + "grad_norm": 2.2616817951202393, + "learning_rate": 9.950603732162458e-07, + "loss": 0.98, + "mean_token_accuracy": 0.7059333920478821, + "num_tokens": 46801619.0, + "step": 1814 + }, + { + "epoch": 0.1993191302437953, + "grad_norm": 2.2442703247070312, + "learning_rate": 9.95609220636663e-07, + "loss": 1.153, + "mean_token_accuracy": 0.660661518573761, + "num_tokens": 46828284.0, + "step": 1815 + }, + { + "epoch": 0.19942894794640897, + "grad_norm": 2.343207836151123, + "learning_rate": 9.9615806805708e-07, + "loss": 1.0357, + "mean_token_accuracy": 0.6909666657447815, + "num_tokens": 46852397.0, + "step": 1816 + }, + { + "epoch": 0.19953876564902262, + "grad_norm": 2.876896858215332, + "learning_rate": 9.967069154774972e-07, + "loss": 0.9969, + "mean_token_accuracy": 0.7047709226608276, + "num_tokens": 46869062.0, + "step": 1817 + }, + { + "epoch": 0.1996485833516363, + "grad_norm": 2.2110681533813477, + "learning_rate": 9.972557628979144e-07, + "loss": 1.0921, + "mean_token_accuracy": 0.6787453293800354, + "num_tokens": 46901505.0, + "step": 1818 + }, + { + "epoch": 0.19975840105424994, + "grad_norm": 2.591646909713745, + "learning_rate": 9.978046103183315e-07, + "loss": 0.9263, + "mean_token_accuracy": 0.7119088172912598, + "num_tokens": 46920429.0, + "step": 1819 + }, + { + "epoch": 0.1998682187568636, + "grad_norm": 2.1674585342407227, + "learning_rate": 9.983534577387486e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6866243481636047, + "num_tokens": 46949821.0, + "step": 1820 + }, + { + "epoch": 0.19997803645947726, + "grad_norm": 2.2692277431488037, + "learning_rate": 9.989023051591658e-07, + "loss": 1.0397, + "mean_token_accuracy": 0.6896665096282959, + "num_tokens": 46974658.0, + "step": 1821 + }, + { + "epoch": 0.20008785416209093, + "grad_norm": 1.9649935960769653, + "learning_rate": 9.994511525795829e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6708452701568604, + "num_tokens": 47006752.0, + "step": 1822 + }, + { + "epoch": 0.20019767186470458, + "grad_norm": 2.307889223098755, + "learning_rate": 1e-06, + "loss": 1.1357, + "mean_token_accuracy": 0.6643370389938354, + "num_tokens": 47031966.0, + "step": 1823 + }, + { + "epoch": 0.20030748956731825, + "grad_norm": 2.469698190689087, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6991109251976013, + "num_tokens": 47053679.0, + "step": 1824 + }, + { + "epoch": 0.2004173072699319, + "grad_norm": 2.5630016326904297, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7189045548439026, + "num_tokens": 47073558.0, + "step": 1825 + }, + { + "epoch": 0.20052712497254557, + "grad_norm": 2.3375680446624756, + "learning_rate": 1e-06, + "loss": 1.085, + "mean_token_accuracy": 0.6800290942192078, + "num_tokens": 47100783.0, + "step": 1826 + }, + { + "epoch": 0.20063694267515925, + "grad_norm": 2.1342713832855225, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6970958709716797, + "num_tokens": 47129372.0, + "step": 1827 + }, + { + "epoch": 0.2007467603777729, + "grad_norm": 2.335787534713745, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6947379112243652, + "num_tokens": 47153496.0, + "step": 1828 + }, + { + "epoch": 0.20085657808038657, + "grad_norm": 2.1313230991363525, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7118631601333618, + "num_tokens": 47180141.0, + "step": 1829 + }, + { + "epoch": 0.2009663957830002, + "grad_norm": 2.548440933227539, + "learning_rate": 1e-06, + "loss": 1.0942, + "mean_token_accuracy": 0.6749528646469116, + "num_tokens": 47203230.0, + "step": 1830 + }, + { + "epoch": 0.20107621348561389, + "grad_norm": 1.9572560787200928, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7067432403564453, + "num_tokens": 47233845.0, + "step": 1831 + }, + { + "epoch": 0.20118603118822753, + "grad_norm": 2.472460985183716, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7183979153633118, + "num_tokens": 47254682.0, + "step": 1832 + }, + { + "epoch": 0.2012958488908412, + "grad_norm": 2.2658889293670654, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6884104013442993, + "num_tokens": 47281270.0, + "step": 1833 + }, + { + "epoch": 0.20140566659345485, + "grad_norm": 2.5829217433929443, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7023110389709473, + "num_tokens": 47302249.0, + "step": 1834 + }, + { + "epoch": 0.20151548429606853, + "grad_norm": 2.2554678916931152, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7009701728820801, + "num_tokens": 47331214.0, + "step": 1835 + }, + { + "epoch": 0.2016253019986822, + "grad_norm": 2.6294350624084473, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6990956664085388, + "num_tokens": 47351165.0, + "step": 1836 + }, + { + "epoch": 0.20173511970129585, + "grad_norm": 2.1693711280822754, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.703086256980896, + "num_tokens": 47379401.0, + "step": 1837 + }, + { + "epoch": 0.20184493740390952, + "grad_norm": 2.1375186443328857, + "learning_rate": 1e-06, + "loss": 1.1117, + "mean_token_accuracy": 0.6748487949371338, + "num_tokens": 47409357.0, + "step": 1838 + }, + { + "epoch": 0.20195475510652316, + "grad_norm": 2.4811909198760986, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7306619882583618, + "num_tokens": 47429632.0, + "step": 1839 + }, + { + "epoch": 0.20206457280913684, + "grad_norm": 2.6499292850494385, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.686881959438324, + "num_tokens": 47450852.0, + "step": 1840 + }, + { + "epoch": 0.20217439051175048, + "grad_norm": 2.4919958114624023, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6930510997772217, + "num_tokens": 47472292.0, + "step": 1841 + }, + { + "epoch": 0.20228420821436416, + "grad_norm": 2.3528175354003906, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.709752082824707, + "num_tokens": 47495841.0, + "step": 1842 + }, + { + "epoch": 0.2023940259169778, + "grad_norm": 2.3199450969696045, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7080215215682983, + "num_tokens": 47520161.0, + "step": 1843 + }, + { + "epoch": 0.20250384361959148, + "grad_norm": 2.28395414352417, + "learning_rate": 1e-06, + "loss": 1.0987, + "mean_token_accuracy": 0.6764238476753235, + "num_tokens": 47546053.0, + "step": 1844 + }, + { + "epoch": 0.20261366132220515, + "grad_norm": 2.4577667713165283, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.686242938041687, + "num_tokens": 47567724.0, + "step": 1845 + }, + { + "epoch": 0.2027234790248188, + "grad_norm": 2.0734777450561523, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6736974716186523, + "num_tokens": 47597316.0, + "step": 1846 + }, + { + "epoch": 0.20283329672743247, + "grad_norm": 2.0778932571411133, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7149277925491333, + "num_tokens": 47623271.0, + "step": 1847 + }, + { + "epoch": 0.20294311443004612, + "grad_norm": 2.5194554328918457, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6898936033248901, + "num_tokens": 47644915.0, + "step": 1848 + }, + { + "epoch": 0.2030529321326598, + "grad_norm": 2.4289188385009766, + "learning_rate": 1e-06, + "loss": 1.1101, + "mean_token_accuracy": 0.6695390343666077, + "num_tokens": 47669351.0, + "step": 1849 + }, + { + "epoch": 0.20316274983527344, + "grad_norm": 2.4817891120910645, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6920251846313477, + "num_tokens": 47690533.0, + "step": 1850 + }, + { + "epoch": 0.2032725675378871, + "grad_norm": 2.5138468742370605, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7203161120414734, + "num_tokens": 47711508.0, + "step": 1851 + }, + { + "epoch": 0.20338238524050076, + "grad_norm": 2.092006206512451, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.681204617023468, + "num_tokens": 47738244.0, + "step": 1852 + }, + { + "epoch": 0.20349220294311443, + "grad_norm": 2.176238775253296, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6971884965896606, + "num_tokens": 47766722.0, + "step": 1853 + }, + { + "epoch": 0.2036020206457281, + "grad_norm": 2.2366139888763428, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6795486211776733, + "num_tokens": 47794251.0, + "step": 1854 + }, + { + "epoch": 0.20371183834834175, + "grad_norm": 2.201089382171631, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6970033645629883, + "num_tokens": 47822218.0, + "step": 1855 + }, + { + "epoch": 0.20382165605095542, + "grad_norm": 2.0988497734069824, + "learning_rate": 1e-06, + "loss": 1.0968, + "mean_token_accuracy": 0.6737321615219116, + "num_tokens": 47851391.0, + "step": 1856 + }, + { + "epoch": 0.20393147375356907, + "grad_norm": 2.0525875091552734, + "learning_rate": 1e-06, + "loss": 1.0923, + "mean_token_accuracy": 0.6853431463241577, + "num_tokens": 47883519.0, + "step": 1857 + }, + { + "epoch": 0.20404129145618274, + "grad_norm": 2.230195999145508, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6989703178405762, + "num_tokens": 47909484.0, + "step": 1858 + }, + { + "epoch": 0.2041511091587964, + "grad_norm": 2.2488601207733154, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.677285373210907, + "num_tokens": 47937091.0, + "step": 1859 + }, + { + "epoch": 0.20426092686141006, + "grad_norm": 2.4568605422973633, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7151458263397217, + "num_tokens": 47959916.0, + "step": 1860 + }, + { + "epoch": 0.2043707445640237, + "grad_norm": 2.439005136489868, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.688546895980835, + "num_tokens": 47982471.0, + "step": 1861 + }, + { + "epoch": 0.20448056226663738, + "grad_norm": 2.426739454269409, + "learning_rate": 1e-06, + "loss": 1.078, + "mean_token_accuracy": 0.6783883571624756, + "num_tokens": 48006577.0, + "step": 1862 + }, + { + "epoch": 0.20459037996925103, + "grad_norm": 2.320990800857544, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7067023515701294, + "num_tokens": 48030455.0, + "step": 1863 + }, + { + "epoch": 0.2047001976718647, + "grad_norm": 2.1055216789245605, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.697187066078186, + "num_tokens": 48061005.0, + "step": 1864 + }, + { + "epoch": 0.20481001537447838, + "grad_norm": 2.0714950561523438, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7101336717605591, + "num_tokens": 48089089.0, + "step": 1865 + }, + { + "epoch": 0.20491983307709202, + "grad_norm": 2.3723928928375244, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6752936840057373, + "num_tokens": 48112880.0, + "step": 1866 + }, + { + "epoch": 0.2050296507797057, + "grad_norm": 2.536839485168457, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6887813806533813, + "num_tokens": 48133014.0, + "step": 1867 + }, + { + "epoch": 0.20513946848231934, + "grad_norm": 2.2663793563842773, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6989652514457703, + "num_tokens": 48159697.0, + "step": 1868 + }, + { + "epoch": 0.20524928618493302, + "grad_norm": 2.236618995666504, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.699825644493103, + "num_tokens": 48183675.0, + "step": 1869 + }, + { + "epoch": 0.20535910388754666, + "grad_norm": 2.1592037677764893, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6881768703460693, + "num_tokens": 48214135.0, + "step": 1870 + }, + { + "epoch": 0.20546892159016034, + "grad_norm": 2.3315176963806152, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6968572735786438, + "num_tokens": 48240499.0, + "step": 1871 + }, + { + "epoch": 0.20557873929277398, + "grad_norm": 2.408090353012085, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6860617995262146, + "num_tokens": 48264390.0, + "step": 1872 + }, + { + "epoch": 0.20568855699538766, + "grad_norm": 2.0780398845672607, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6771674156188965, + "num_tokens": 48296708.0, + "step": 1873 + }, + { + "epoch": 0.20579837469800133, + "grad_norm": 2.01908802986145, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6964594721794128, + "num_tokens": 48329177.0, + "step": 1874 + }, + { + "epoch": 0.20590819240061498, + "grad_norm": 2.161651611328125, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7104798555374146, + "num_tokens": 48357078.0, + "step": 1875 + }, + { + "epoch": 0.20601801010322865, + "grad_norm": 2.1513073444366455, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.6856120824813843, + "num_tokens": 48385537.0, + "step": 1876 + }, + { + "epoch": 0.2061278278058423, + "grad_norm": 2.5246741771698, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7151254415512085, + "num_tokens": 48405713.0, + "step": 1877 + }, + { + "epoch": 0.20623764550845597, + "grad_norm": 2.0252888202667236, + "learning_rate": 1e-06, + "loss": 1.1069, + "mean_token_accuracy": 0.6698153018951416, + "num_tokens": 48436719.0, + "step": 1878 + }, + { + "epoch": 0.20634746321106961, + "grad_norm": 2.509605884552002, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6872165203094482, + "num_tokens": 48461452.0, + "step": 1879 + }, + { + "epoch": 0.2064572809136833, + "grad_norm": 2.2488605976104736, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7022922039031982, + "num_tokens": 48486326.0, + "step": 1880 + }, + { + "epoch": 0.20656709861629693, + "grad_norm": 2.1548068523406982, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6702258586883545, + "num_tokens": 48515160.0, + "step": 1881 + }, + { + "epoch": 0.2066769163189106, + "grad_norm": 2.305488109588623, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6893539428710938, + "num_tokens": 48541174.0, + "step": 1882 + }, + { + "epoch": 0.20678673402152428, + "grad_norm": 2.2494800090789795, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6981503367424011, + "num_tokens": 48568742.0, + "step": 1883 + }, + { + "epoch": 0.20689655172413793, + "grad_norm": 2.377406358718872, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6798393726348877, + "num_tokens": 48591293.0, + "step": 1884 + }, + { + "epoch": 0.2070063694267516, + "grad_norm": 2.189563035964966, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7239277362823486, + "num_tokens": 48616962.0, + "step": 1885 + }, + { + "epoch": 0.20711618712936525, + "grad_norm": 2.1889522075653076, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6883800029754639, + "num_tokens": 48646493.0, + "step": 1886 + }, + { + "epoch": 0.20722600483197892, + "grad_norm": 2.3043506145477295, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6855978965759277, + "num_tokens": 48669626.0, + "step": 1887 + }, + { + "epoch": 0.20733582253459257, + "grad_norm": 2.2803001403808594, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.686901330947876, + "num_tokens": 48695815.0, + "step": 1888 + }, + { + "epoch": 0.20744564023720624, + "grad_norm": 2.660578489303589, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6964772939682007, + "num_tokens": 48717797.0, + "step": 1889 + }, + { + "epoch": 0.2075554579398199, + "grad_norm": 2.1544172763824463, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6899252533912659, + "num_tokens": 48745682.0, + "step": 1890 + }, + { + "epoch": 0.20766527564243356, + "grad_norm": 2.0900774002075195, + "learning_rate": 1e-06, + "loss": 1.1353, + "mean_token_accuracy": 0.6659979820251465, + "num_tokens": 48778420.0, + "step": 1891 + }, + { + "epoch": 0.20777509334504723, + "grad_norm": 2.407820701599121, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7089846730232239, + "num_tokens": 48801159.0, + "step": 1892 + }, + { + "epoch": 0.20788491104766088, + "grad_norm": 2.274601459503174, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6920920610427856, + "num_tokens": 48827530.0, + "step": 1893 + }, + { + "epoch": 0.20799472875027455, + "grad_norm": 2.1363041400909424, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6762123107910156, + "num_tokens": 48854693.0, + "step": 1894 + }, + { + "epoch": 0.2081045464528882, + "grad_norm": 2.2825570106506348, + "learning_rate": 1e-06, + "loss": 1.081, + "mean_token_accuracy": 0.6777942180633545, + "num_tokens": 48879817.0, + "step": 1895 + }, + { + "epoch": 0.20821436415550187, + "grad_norm": 2.293973207473755, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6836770176887512, + "num_tokens": 48904832.0, + "step": 1896 + }, + { + "epoch": 0.20832418185811552, + "grad_norm": 2.810230016708374, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7143701910972595, + "num_tokens": 48921814.0, + "step": 1897 + }, + { + "epoch": 0.2084339995607292, + "grad_norm": 2.7691001892089844, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6839761734008789, + "num_tokens": 48940813.0, + "step": 1898 + }, + { + "epoch": 0.20854381726334284, + "grad_norm": 2.4332756996154785, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6817268133163452, + "num_tokens": 48964813.0, + "step": 1899 + }, + { + "epoch": 0.2086536349659565, + "grad_norm": 2.486701011657715, + "learning_rate": 1e-06, + "loss": 1.0999, + "mean_token_accuracy": 0.685187816619873, + "num_tokens": 48987194.0, + "step": 1900 + }, + { + "epoch": 0.20876345266857016, + "grad_norm": 1.9717936515808105, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6848686933517456, + "num_tokens": 49017952.0, + "step": 1901 + }, + { + "epoch": 0.20887327037118383, + "grad_norm": 2.018864393234253, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7102551460266113, + "num_tokens": 49048069.0, + "step": 1902 + }, + { + "epoch": 0.2089830880737975, + "grad_norm": 2.4940481185913086, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6936544179916382, + "num_tokens": 49068339.0, + "step": 1903 + }, + { + "epoch": 0.20909290577641115, + "grad_norm": 2.1023128032684326, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.696506917476654, + "num_tokens": 49095209.0, + "step": 1904 + }, + { + "epoch": 0.20920272347902483, + "grad_norm": 2.1862621307373047, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.715683102607727, + "num_tokens": 49120549.0, + "step": 1905 + }, + { + "epoch": 0.20931254118163847, + "grad_norm": 2.344644546508789, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.699777364730835, + "num_tokens": 49142760.0, + "step": 1906 + }, + { + "epoch": 0.20942235888425215, + "grad_norm": 2.3967394828796387, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.696273922920227, + "num_tokens": 49164838.0, + "step": 1907 + }, + { + "epoch": 0.2095321765868658, + "grad_norm": 2.4753687381744385, + "learning_rate": 1e-06, + "loss": 1.0879, + "mean_token_accuracy": 0.6821571588516235, + "num_tokens": 49186587.0, + "step": 1908 + }, + { + "epoch": 0.20964199428947947, + "grad_norm": 2.259805917739868, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.690529465675354, + "num_tokens": 49211109.0, + "step": 1909 + }, + { + "epoch": 0.2097518119920931, + "grad_norm": 2.4043455123901367, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6892874836921692, + "num_tokens": 49236174.0, + "step": 1910 + }, + { + "epoch": 0.20986162969470679, + "grad_norm": 2.0070812702178955, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7193841934204102, + "num_tokens": 49264690.0, + "step": 1911 + }, + { + "epoch": 0.20997144739732046, + "grad_norm": 2.5731871128082275, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7277430295944214, + "num_tokens": 49282677.0, + "step": 1912 + }, + { + "epoch": 0.2100812650999341, + "grad_norm": 2.5012571811676025, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6824767589569092, + "num_tokens": 49305116.0, + "step": 1913 + }, + { + "epoch": 0.21019108280254778, + "grad_norm": 2.156167507171631, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6975862979888916, + "num_tokens": 49334003.0, + "step": 1914 + }, + { + "epoch": 0.21030090050516143, + "grad_norm": 2.1726365089416504, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6845719814300537, + "num_tokens": 49363785.0, + "step": 1915 + }, + { + "epoch": 0.2104107182077751, + "grad_norm": 2.5351696014404297, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7269984483718872, + "num_tokens": 49383606.0, + "step": 1916 + }, + { + "epoch": 0.21052053591038875, + "grad_norm": 2.265700340270996, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.675700306892395, + "num_tokens": 49408184.0, + "step": 1917 + }, + { + "epoch": 0.21063035361300242, + "grad_norm": 2.5012145042419434, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6986526846885681, + "num_tokens": 49429723.0, + "step": 1918 + }, + { + "epoch": 0.21074017131561606, + "grad_norm": 2.464399814605713, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6888995170593262, + "num_tokens": 49452169.0, + "step": 1919 + }, + { + "epoch": 0.21084998901822974, + "grad_norm": 2.138547897338867, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6898144483566284, + "num_tokens": 49483334.0, + "step": 1920 + }, + { + "epoch": 0.2109598067208434, + "grad_norm": 2.060312509536743, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6914002895355225, + "num_tokens": 49510964.0, + "step": 1921 + }, + { + "epoch": 0.21106962442345706, + "grad_norm": 2.3183321952819824, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6920852661132812, + "num_tokens": 49537475.0, + "step": 1922 + }, + { + "epoch": 0.21117944212607073, + "grad_norm": 2.0987720489501953, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7164697647094727, + "num_tokens": 49564778.0, + "step": 1923 + }, + { + "epoch": 0.21128925982868438, + "grad_norm": 2.2984776496887207, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.691868782043457, + "num_tokens": 49588472.0, + "step": 1924 + }, + { + "epoch": 0.21139907753129805, + "grad_norm": 2.217088222503662, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.7012676000595093, + "num_tokens": 49617133.0, + "step": 1925 + }, + { + "epoch": 0.2115088952339117, + "grad_norm": 2.3941781520843506, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7031118273735046, + "num_tokens": 49640836.0, + "step": 1926 + }, + { + "epoch": 0.21161871293652537, + "grad_norm": 2.1003994941711426, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.6809670329093933, + "num_tokens": 49671807.0, + "step": 1927 + }, + { + "epoch": 0.21172853063913902, + "grad_norm": 2.2453417778015137, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.688103437423706, + "num_tokens": 49697455.0, + "step": 1928 + }, + { + "epoch": 0.2118383483417527, + "grad_norm": 2.42117977142334, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6803967952728271, + "num_tokens": 49721545.0, + "step": 1929 + }, + { + "epoch": 0.21194816604436637, + "grad_norm": 2.2804982662200928, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7085500955581665, + "num_tokens": 49746111.0, + "step": 1930 + }, + { + "epoch": 0.21205798374698, + "grad_norm": 2.419348955154419, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7340570688247681, + "num_tokens": 49766842.0, + "step": 1931 + }, + { + "epoch": 0.21216780144959368, + "grad_norm": 2.368223190307617, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.697149932384491, + "num_tokens": 49792173.0, + "step": 1932 + }, + { + "epoch": 0.21227761915220733, + "grad_norm": 2.3558874130249023, + "learning_rate": 1e-06, + "loss": 1.097, + "mean_token_accuracy": 0.6759493350982666, + "num_tokens": 49816667.0, + "step": 1933 + }, + { + "epoch": 0.212387436854821, + "grad_norm": 2.038635015487671, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.690165102481842, + "num_tokens": 49848346.0, + "step": 1934 + }, + { + "epoch": 0.21249725455743465, + "grad_norm": 2.5301759243011475, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7319109439849854, + "num_tokens": 49867593.0, + "step": 1935 + }, + { + "epoch": 0.21260707226004832, + "grad_norm": 2.443293333053589, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6989279985427856, + "num_tokens": 49889240.0, + "step": 1936 + }, + { + "epoch": 0.21271688996266197, + "grad_norm": 2.0501792430877686, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6958505511283875, + "num_tokens": 49920703.0, + "step": 1937 + }, + { + "epoch": 0.21282670766527564, + "grad_norm": 2.150432825088501, + "learning_rate": 1e-06, + "loss": 1.1236, + "mean_token_accuracy": 0.6678165197372437, + "num_tokens": 49950607.0, + "step": 1938 + }, + { + "epoch": 0.2129365253678893, + "grad_norm": 1.9092386960983276, + "learning_rate": 1e-06, + "loss": 1.0493, + "mean_token_accuracy": 0.6826077103614807, + "num_tokens": 49987030.0, + "step": 1939 + }, + { + "epoch": 0.21304634307050296, + "grad_norm": 2.4255192279815674, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6866302490234375, + "num_tokens": 50009068.0, + "step": 1940 + }, + { + "epoch": 0.21315616077311664, + "grad_norm": 2.1427197456359863, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6938563585281372, + "num_tokens": 50038890.0, + "step": 1941 + }, + { + "epoch": 0.21326597847573028, + "grad_norm": 2.036513090133667, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6858283281326294, + "num_tokens": 50071873.0, + "step": 1942 + }, + { + "epoch": 0.21337579617834396, + "grad_norm": 2.3296737670898438, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6975035667419434, + "num_tokens": 50096147.0, + "step": 1943 + }, + { + "epoch": 0.2134856138809576, + "grad_norm": 2.0835177898406982, + "learning_rate": 1e-06, + "loss": 1.182, + "mean_token_accuracy": 0.6555435657501221, + "num_tokens": 50126336.0, + "step": 1944 + }, + { + "epoch": 0.21359543158357128, + "grad_norm": 2.1825265884399414, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6741807460784912, + "num_tokens": 50157030.0, + "step": 1945 + }, + { + "epoch": 0.21370524928618492, + "grad_norm": 2.5763466358184814, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7074791193008423, + "num_tokens": 50179206.0, + "step": 1946 + }, + { + "epoch": 0.2138150669887986, + "grad_norm": 2.121547222137451, + "learning_rate": 1e-06, + "loss": 1.1365, + "mean_token_accuracy": 0.6609182357788086, + "num_tokens": 50211495.0, + "step": 1947 + }, + { + "epoch": 0.21392488469141224, + "grad_norm": 2.004615068435669, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7150670289993286, + "num_tokens": 50240676.0, + "step": 1948 + }, + { + "epoch": 0.21403470239402592, + "grad_norm": 2.3735620975494385, + "learning_rate": 1e-06, + "loss": 1.1185, + "mean_token_accuracy": 0.6657366752624512, + "num_tokens": 50263709.0, + "step": 1949 + }, + { + "epoch": 0.2141445200966396, + "grad_norm": 2.0547289848327637, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6976802349090576, + "num_tokens": 50291875.0, + "step": 1950 + }, + { + "epoch": 0.21425433779925324, + "grad_norm": 2.235546350479126, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6891975402832031, + "num_tokens": 50318409.0, + "step": 1951 + }, + { + "epoch": 0.2143641555018669, + "grad_norm": 2.397538900375366, + "learning_rate": 1e-06, + "loss": 1.0807, + "mean_token_accuracy": 0.6750262975692749, + "num_tokens": 50342654.0, + "step": 1952 + }, + { + "epoch": 0.21447397320448056, + "grad_norm": 2.152358055114746, + "learning_rate": 1e-06, + "loss": 1.0657, + "mean_token_accuracy": 0.6880247592926025, + "num_tokens": 50368414.0, + "step": 1953 + }, + { + "epoch": 0.21458379090709423, + "grad_norm": 2.5853121280670166, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7001596689224243, + "num_tokens": 50387083.0, + "step": 1954 + }, + { + "epoch": 0.21469360860970788, + "grad_norm": 2.681633472442627, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7133309841156006, + "num_tokens": 50406621.0, + "step": 1955 + }, + { + "epoch": 0.21480342631232155, + "grad_norm": 2.3061001300811768, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7205066680908203, + "num_tokens": 50431956.0, + "step": 1956 + }, + { + "epoch": 0.2149132440149352, + "grad_norm": 2.1312031745910645, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6774511337280273, + "num_tokens": 50460705.0, + "step": 1957 + }, + { + "epoch": 0.21502306171754887, + "grad_norm": 2.346876621246338, + "learning_rate": 1e-06, + "loss": 1.1204, + "mean_token_accuracy": 0.6651735901832581, + "num_tokens": 50484818.0, + "step": 1958 + }, + { + "epoch": 0.21513287942016254, + "grad_norm": 2.4579086303710938, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.7022106051445007, + "num_tokens": 50507428.0, + "step": 1959 + }, + { + "epoch": 0.2152426971227762, + "grad_norm": 2.088623285293579, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6873458027839661, + "num_tokens": 50536098.0, + "step": 1960 + }, + { + "epoch": 0.21535251482538986, + "grad_norm": 2.1509578227996826, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.675479531288147, + "num_tokens": 50565280.0, + "step": 1961 + }, + { + "epoch": 0.2154623325280035, + "grad_norm": 2.1170809268951416, + "learning_rate": 1e-06, + "loss": 1.0755, + "mean_token_accuracy": 0.6791170835494995, + "num_tokens": 50594142.0, + "step": 1962 + }, + { + "epoch": 0.21557215023061718, + "grad_norm": 2.5238492488861084, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6823158860206604, + "num_tokens": 50617184.0, + "step": 1963 + }, + { + "epoch": 0.21568196793323083, + "grad_norm": 2.0235648155212402, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.707868754863739, + "num_tokens": 50644832.0, + "step": 1964 + }, + { + "epoch": 0.2157917856358445, + "grad_norm": 2.057553768157959, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.693603515625, + "num_tokens": 50673146.0, + "step": 1965 + }, + { + "epoch": 0.21590160333845815, + "grad_norm": 2.101991891860962, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6934962272644043, + "num_tokens": 50700712.0, + "step": 1966 + }, + { + "epoch": 0.21601142104107182, + "grad_norm": 2.3096585273742676, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7031317949295044, + "num_tokens": 50723978.0, + "step": 1967 + }, + { + "epoch": 0.2161212387436855, + "grad_norm": 2.4806957244873047, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7049955129623413, + "num_tokens": 50745221.0, + "step": 1968 + }, + { + "epoch": 0.21623105644629914, + "grad_norm": 2.2249786853790283, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6975527405738831, + "num_tokens": 50771672.0, + "step": 1969 + }, + { + "epoch": 0.21634087414891282, + "grad_norm": 2.3493337631225586, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6860683560371399, + "num_tokens": 50795329.0, + "step": 1970 + }, + { + "epoch": 0.21645069185152646, + "grad_norm": 2.3962483406066895, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6946808099746704, + "num_tokens": 50817883.0, + "step": 1971 + }, + { + "epoch": 0.21656050955414013, + "grad_norm": 2.1914432048797607, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6887646913528442, + "num_tokens": 50843602.0, + "step": 1972 + }, + { + "epoch": 0.21667032725675378, + "grad_norm": 2.5093319416046143, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6994876861572266, + "num_tokens": 50864605.0, + "step": 1973 + }, + { + "epoch": 0.21678014495936745, + "grad_norm": 2.183929681777954, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6758536696434021, + "num_tokens": 50891339.0, + "step": 1974 + }, + { + "epoch": 0.2168899626619811, + "grad_norm": 2.202979803085327, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.688660740852356, + "num_tokens": 50916736.0, + "step": 1975 + }, + { + "epoch": 0.21699978036459477, + "grad_norm": 2.0905888080596924, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7151705622673035, + "num_tokens": 50942262.0, + "step": 1976 + }, + { + "epoch": 0.21710959806720842, + "grad_norm": 1.9778714179992676, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6834086775779724, + "num_tokens": 50976710.0, + "step": 1977 + }, + { + "epoch": 0.2172194157698221, + "grad_norm": 2.0964698791503906, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6873266100883484, + "num_tokens": 51004703.0, + "step": 1978 + }, + { + "epoch": 0.21732923347243577, + "grad_norm": 2.1559524536132812, + "learning_rate": 1e-06, + "loss": 1.0703, + "mean_token_accuracy": 0.6779730319976807, + "num_tokens": 51034580.0, + "step": 1979 + }, + { + "epoch": 0.2174390511750494, + "grad_norm": 2.2023422718048096, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6989998817443848, + "num_tokens": 51059530.0, + "step": 1980 + }, + { + "epoch": 0.2175488688776631, + "grad_norm": 2.2927334308624268, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6887468695640564, + "num_tokens": 51086966.0, + "step": 1981 + }, + { + "epoch": 0.21765868658027673, + "grad_norm": 2.2862160205841064, + "learning_rate": 1e-06, + "loss": 1.0818, + "mean_token_accuracy": 0.6744215488433838, + "num_tokens": 51112895.0, + "step": 1982 + }, + { + "epoch": 0.2177685042828904, + "grad_norm": 2.148014545440674, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.712171733379364, + "num_tokens": 51144468.0, + "step": 1983 + }, + { + "epoch": 0.21787832198550405, + "grad_norm": 2.4518561363220215, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6959818601608276, + "num_tokens": 51167490.0, + "step": 1984 + }, + { + "epoch": 0.21798813968811773, + "grad_norm": 1.9508395195007324, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6905363202095032, + "num_tokens": 51200762.0, + "step": 1985 + }, + { + "epoch": 0.21809795739073137, + "grad_norm": 2.7437524795532227, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7141036987304688, + "num_tokens": 51217965.0, + "step": 1986 + }, + { + "epoch": 0.21820777509334505, + "grad_norm": 2.3125598430633545, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6804224252700806, + "num_tokens": 51242779.0, + "step": 1987 + }, + { + "epoch": 0.21831759279595872, + "grad_norm": 2.50113844871521, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6922855973243713, + "num_tokens": 51266150.0, + "step": 1988 + }, + { + "epoch": 0.21842741049857237, + "grad_norm": 2.391871690750122, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6921185851097107, + "num_tokens": 51292586.0, + "step": 1989 + }, + { + "epoch": 0.21853722820118604, + "grad_norm": 2.763786554336548, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7042215466499329, + "num_tokens": 51311798.0, + "step": 1990 + }, + { + "epoch": 0.21864704590379969, + "grad_norm": 2.0070688724517822, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6936018466949463, + "num_tokens": 51343122.0, + "step": 1991 + }, + { + "epoch": 0.21875686360641336, + "grad_norm": 2.4923343658447266, + "learning_rate": 1e-06, + "loss": 1.063, + "mean_token_accuracy": 0.6836159229278564, + "num_tokens": 51366633.0, + "step": 1992 + }, + { + "epoch": 0.218866681309027, + "grad_norm": 2.5267856121063232, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7066759467124939, + "num_tokens": 51386762.0, + "step": 1993 + }, + { + "epoch": 0.21897649901164068, + "grad_norm": 2.210613489151001, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.708017110824585, + "num_tokens": 51411904.0, + "step": 1994 + }, + { + "epoch": 0.21908631671425433, + "grad_norm": 1.9896042346954346, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6993438005447388, + "num_tokens": 51443189.0, + "step": 1995 + }, + { + "epoch": 0.219196134416868, + "grad_norm": 2.2535996437072754, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6815817356109619, + "num_tokens": 51470002.0, + "step": 1996 + }, + { + "epoch": 0.21930595211948167, + "grad_norm": 2.2535860538482666, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.689070463180542, + "num_tokens": 51496221.0, + "step": 1997 + }, + { + "epoch": 0.21941576982209532, + "grad_norm": 2.2650434970855713, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7168316841125488, + "num_tokens": 51522047.0, + "step": 1998 + }, + { + "epoch": 0.219525587524709, + "grad_norm": 2.503796339035034, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7043272852897644, + "num_tokens": 51543968.0, + "step": 1999 + }, + { + "epoch": 0.21963540522732264, + "grad_norm": 1.9360029697418213, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7283309698104858, + "num_tokens": 51574860.0, + "step": 2000 + }, + { + "epoch": 0.2197452229299363, + "grad_norm": 2.4913580417633057, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6884080171585083, + "num_tokens": 51597281.0, + "step": 2001 + }, + { + "epoch": 0.21985504063254996, + "grad_norm": 2.4393420219421387, + "learning_rate": 1e-06, + "loss": 1.1088, + "mean_token_accuracy": 0.6783907413482666, + "num_tokens": 51621929.0, + "step": 2002 + }, + { + "epoch": 0.21996485833516363, + "grad_norm": 2.1955487728118896, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7132848501205444, + "num_tokens": 51648309.0, + "step": 2003 + }, + { + "epoch": 0.22007467603777728, + "grad_norm": 2.3509833812713623, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7077527642250061, + "num_tokens": 51671475.0, + "step": 2004 + }, + { + "epoch": 0.22018449374039095, + "grad_norm": 2.089629650115967, + "learning_rate": 1e-06, + "loss": 1.1311, + "mean_token_accuracy": 0.6702072620391846, + "num_tokens": 51704408.0, + "step": 2005 + }, + { + "epoch": 0.22029431144300463, + "grad_norm": 2.623277187347412, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7156935334205627, + "num_tokens": 51725442.0, + "step": 2006 + }, + { + "epoch": 0.22040412914561827, + "grad_norm": 2.5945138931274414, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7052671909332275, + "num_tokens": 51745442.0, + "step": 2007 + }, + { + "epoch": 0.22051394684823195, + "grad_norm": 2.3193936347961426, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7018911838531494, + "num_tokens": 51768359.0, + "step": 2008 + }, + { + "epoch": 0.2206237645508456, + "grad_norm": 2.3500494956970215, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7143868207931519, + "num_tokens": 51792080.0, + "step": 2009 + }, + { + "epoch": 0.22073358225345927, + "grad_norm": 2.508650064468384, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6861566305160522, + "num_tokens": 51813779.0, + "step": 2010 + }, + { + "epoch": 0.2208433999560729, + "grad_norm": 2.383606433868408, + "learning_rate": 1e-06, + "loss": 1.0755, + "mean_token_accuracy": 0.675670862197876, + "num_tokens": 51838770.0, + "step": 2011 + }, + { + "epoch": 0.22095321765868658, + "grad_norm": 2.2503316402435303, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7106043100357056, + "num_tokens": 51865554.0, + "step": 2012 + }, + { + "epoch": 0.22106303536130023, + "grad_norm": 2.4900104999542236, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6983499526977539, + "num_tokens": 51888757.0, + "step": 2013 + }, + { + "epoch": 0.2211728530639139, + "grad_norm": 1.989453673362732, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7029899954795837, + "num_tokens": 51918599.0, + "step": 2014 + }, + { + "epoch": 0.22128267076652755, + "grad_norm": 1.9797781705856323, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6939705610275269, + "num_tokens": 51952504.0, + "step": 2015 + }, + { + "epoch": 0.22139248846914122, + "grad_norm": 2.1946585178375244, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6899868249893188, + "num_tokens": 51981691.0, + "step": 2016 + }, + { + "epoch": 0.2215023061717549, + "grad_norm": 2.262589454650879, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7158289551734924, + "num_tokens": 52007012.0, + "step": 2017 + }, + { + "epoch": 0.22161212387436854, + "grad_norm": 2.500941753387451, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6891971826553345, + "num_tokens": 52029883.0, + "step": 2018 + }, + { + "epoch": 0.22172194157698222, + "grad_norm": 2.239351511001587, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.708087682723999, + "num_tokens": 52055909.0, + "step": 2019 + }, + { + "epoch": 0.22183175927959586, + "grad_norm": 2.2200400829315186, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6966500282287598, + "num_tokens": 52080214.0, + "step": 2020 + }, + { + "epoch": 0.22194157698220954, + "grad_norm": 2.35465407371521, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.689105212688446, + "num_tokens": 52104396.0, + "step": 2021 + }, + { + "epoch": 0.22205139468482318, + "grad_norm": 2.4693686962127686, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7002556920051575, + "num_tokens": 52127059.0, + "step": 2022 + }, + { + "epoch": 0.22216121238743686, + "grad_norm": 2.328630208969116, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7174548506736755, + "num_tokens": 52152065.0, + "step": 2023 + }, + { + "epoch": 0.2222710300900505, + "grad_norm": 2.5196924209594727, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6840376853942871, + "num_tokens": 52174122.0, + "step": 2024 + }, + { + "epoch": 0.22238084779266418, + "grad_norm": 2.4212119579315186, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6921921372413635, + "num_tokens": 52198874.0, + "step": 2025 + }, + { + "epoch": 0.22249066549527785, + "grad_norm": 2.246354341506958, + "learning_rate": 1e-06, + "loss": 1.0906, + "mean_token_accuracy": 0.6834812760353088, + "num_tokens": 52225257.0, + "step": 2026 + }, + { + "epoch": 0.2226004831978915, + "grad_norm": 2.438734292984009, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7035894989967346, + "num_tokens": 52248284.0, + "step": 2027 + }, + { + "epoch": 0.22271030090050517, + "grad_norm": 2.404784917831421, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6900989413261414, + "num_tokens": 52273226.0, + "step": 2028 + }, + { + "epoch": 0.22282011860311882, + "grad_norm": 2.3888368606567383, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7086814641952515, + "num_tokens": 52295396.0, + "step": 2029 + }, + { + "epoch": 0.2229299363057325, + "grad_norm": 2.5422487258911133, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7219879627227783, + "num_tokens": 52315542.0, + "step": 2030 + }, + { + "epoch": 0.22303975400834614, + "grad_norm": 2.162820816040039, + "learning_rate": 1e-06, + "loss": 1.1816, + "mean_token_accuracy": 0.6586207151412964, + "num_tokens": 52346091.0, + "step": 2031 + }, + { + "epoch": 0.2231495717109598, + "grad_norm": 2.2456963062286377, + "learning_rate": 1e-06, + "loss": 1.0956, + "mean_token_accuracy": 0.6688408851623535, + "num_tokens": 52375098.0, + "step": 2032 + }, + { + "epoch": 0.22325938941357346, + "grad_norm": 2.2636704444885254, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.6928312182426453, + "num_tokens": 52401152.0, + "step": 2033 + }, + { + "epoch": 0.22336920711618713, + "grad_norm": 2.1768388748168945, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6964225769042969, + "num_tokens": 52429634.0, + "step": 2034 + }, + { + "epoch": 0.2234790248188008, + "grad_norm": 2.0557053089141846, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7072370052337646, + "num_tokens": 52460839.0, + "step": 2035 + }, + { + "epoch": 0.22358884252141445, + "grad_norm": 2.0743494033813477, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7026038765907288, + "num_tokens": 52489248.0, + "step": 2036 + }, + { + "epoch": 0.22369866022402812, + "grad_norm": 2.149608850479126, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7047668099403381, + "num_tokens": 52516517.0, + "step": 2037 + }, + { + "epoch": 0.22380847792664177, + "grad_norm": 2.1062376499176025, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6947094202041626, + "num_tokens": 52544695.0, + "step": 2038 + }, + { + "epoch": 0.22391829562925544, + "grad_norm": 2.036336660385132, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7143869400024414, + "num_tokens": 52572773.0, + "step": 2039 + }, + { + "epoch": 0.2240281133318691, + "grad_norm": 2.2870264053344727, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7209149599075317, + "num_tokens": 52597343.0, + "step": 2040 + }, + { + "epoch": 0.22413793103448276, + "grad_norm": 2.2844462394714355, + "learning_rate": 1e-06, + "loss": 1.0719, + "mean_token_accuracy": 0.6784175634384155, + "num_tokens": 52621060.0, + "step": 2041 + }, + { + "epoch": 0.2242477487370964, + "grad_norm": 2.126678705215454, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6982303857803345, + "num_tokens": 52649123.0, + "step": 2042 + }, + { + "epoch": 0.22435756643971008, + "grad_norm": 2.494357109069824, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7261292934417725, + "num_tokens": 52671520.0, + "step": 2043 + }, + { + "epoch": 0.22446738414232376, + "grad_norm": 2.4692084789276123, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7098029255867004, + "num_tokens": 52694326.0, + "step": 2044 + }, + { + "epoch": 0.2245772018449374, + "grad_norm": 2.160059690475464, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7047584652900696, + "num_tokens": 52721273.0, + "step": 2045 + }, + { + "epoch": 0.22468701954755108, + "grad_norm": 2.275557518005371, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6911457777023315, + "num_tokens": 52746399.0, + "step": 2046 + }, + { + "epoch": 0.22479683725016472, + "grad_norm": 2.184753656387329, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6895255446434021, + "num_tokens": 52772857.0, + "step": 2047 + }, + { + "epoch": 0.2249066549527784, + "grad_norm": 2.3506438732147217, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7118210792541504, + "num_tokens": 52796662.0, + "step": 2048 + }, + { + "epoch": 0.22501647265539204, + "grad_norm": 2.791766405105591, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7100281119346619, + "num_tokens": 52815102.0, + "step": 2049 + }, + { + "epoch": 0.22512629035800572, + "grad_norm": 2.0526466369628906, + "learning_rate": 1e-06, + "loss": 1.0833, + "mean_token_accuracy": 0.6786413788795471, + "num_tokens": 52847158.0, + "step": 2050 + }, + { + "epoch": 0.22523610806061936, + "grad_norm": 2.3040356636047363, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6920868158340454, + "num_tokens": 52870370.0, + "step": 2051 + }, + { + "epoch": 0.22534592576323303, + "grad_norm": 2.0756847858428955, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6860775351524353, + "num_tokens": 52898495.0, + "step": 2052 + }, + { + "epoch": 0.22545574346584668, + "grad_norm": 2.572211980819702, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6835129261016846, + "num_tokens": 52919164.0, + "step": 2053 + }, + { + "epoch": 0.22556556116846035, + "grad_norm": 2.6036064624786377, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7017735242843628, + "num_tokens": 52939777.0, + "step": 2054 + }, + { + "epoch": 0.22567537887107403, + "grad_norm": 1.9949966669082642, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.723812460899353, + "num_tokens": 52968933.0, + "step": 2055 + }, + { + "epoch": 0.22578519657368767, + "grad_norm": 2.264819860458374, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6777077913284302, + "num_tokens": 52994801.0, + "step": 2056 + }, + { + "epoch": 0.22589501427630135, + "grad_norm": 2.3942275047302246, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6993923187255859, + "num_tokens": 53016203.0, + "step": 2057 + }, + { + "epoch": 0.226004831978915, + "grad_norm": 2.5590574741363525, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6954149007797241, + "num_tokens": 53038265.0, + "step": 2058 + }, + { + "epoch": 0.22611464968152867, + "grad_norm": 2.0657241344451904, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6939489245414734, + "num_tokens": 53069234.0, + "step": 2059 + }, + { + "epoch": 0.2262244673841423, + "grad_norm": 2.7056124210357666, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6884029507637024, + "num_tokens": 53090392.0, + "step": 2060 + }, + { + "epoch": 0.226334285086756, + "grad_norm": 2.4730794429779053, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7126097679138184, + "num_tokens": 53111744.0, + "step": 2061 + }, + { + "epoch": 0.22644410278936963, + "grad_norm": 2.1446549892425537, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6897145509719849, + "num_tokens": 53141849.0, + "step": 2062 + }, + { + "epoch": 0.2265539204919833, + "grad_norm": 2.1509463787078857, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6980983018875122, + "num_tokens": 53170091.0, + "step": 2063 + }, + { + "epoch": 0.22666373819459698, + "grad_norm": 2.159041404724121, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6889088749885559, + "num_tokens": 53198276.0, + "step": 2064 + }, + { + "epoch": 0.22677355589721063, + "grad_norm": 2.3580338954925537, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6879932880401611, + "num_tokens": 53223349.0, + "step": 2065 + }, + { + "epoch": 0.2268833735998243, + "grad_norm": 2.1838200092315674, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7107489109039307, + "num_tokens": 53250641.0, + "step": 2066 + }, + { + "epoch": 0.22699319130243795, + "grad_norm": 2.5386011600494385, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6823906898498535, + "num_tokens": 53275537.0, + "step": 2067 + }, + { + "epoch": 0.22710300900505162, + "grad_norm": 2.4914889335632324, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6979907751083374, + "num_tokens": 53299513.0, + "step": 2068 + }, + { + "epoch": 0.22721282670766527, + "grad_norm": 2.1385364532470703, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6859594583511353, + "num_tokens": 53327712.0, + "step": 2069 + }, + { + "epoch": 0.22732264441027894, + "grad_norm": 2.296421766281128, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6912260055541992, + "num_tokens": 53353636.0, + "step": 2070 + }, + { + "epoch": 0.22743246211289259, + "grad_norm": 2.476461410522461, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6990870237350464, + "num_tokens": 53372611.0, + "step": 2071 + }, + { + "epoch": 0.22754227981550626, + "grad_norm": 2.1974899768829346, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6998175382614136, + "num_tokens": 53399562.0, + "step": 2072 + }, + { + "epoch": 0.22765209751811993, + "grad_norm": 2.2847867012023926, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7214905023574829, + "num_tokens": 53422269.0, + "step": 2073 + }, + { + "epoch": 0.22776191522073358, + "grad_norm": 2.187835454940796, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6952679753303528, + "num_tokens": 53449078.0, + "step": 2074 + }, + { + "epoch": 0.22787173292334725, + "grad_norm": 2.081381320953369, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.714687705039978, + "num_tokens": 53479939.0, + "step": 2075 + }, + { + "epoch": 0.2279815506259609, + "grad_norm": 2.08559513092041, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6948668956756592, + "num_tokens": 53510351.0, + "step": 2076 + }, + { + "epoch": 0.22809136832857457, + "grad_norm": 2.298936128616333, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6831876039505005, + "num_tokens": 53538974.0, + "step": 2077 + }, + { + "epoch": 0.22820118603118822, + "grad_norm": 2.820230007171631, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7191442847251892, + "num_tokens": 53558060.0, + "step": 2078 + }, + { + "epoch": 0.2283110037338019, + "grad_norm": 2.534717321395874, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6853632926940918, + "num_tokens": 53579260.0, + "step": 2079 + }, + { + "epoch": 0.22842082143641554, + "grad_norm": 2.419581651687622, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6858792304992676, + "num_tokens": 53601801.0, + "step": 2080 + }, + { + "epoch": 0.2285306391390292, + "grad_norm": 2.3602957725524902, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6992571353912354, + "num_tokens": 53625215.0, + "step": 2081 + }, + { + "epoch": 0.2286404568416429, + "grad_norm": 2.008044719696045, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.698645830154419, + "num_tokens": 53657463.0, + "step": 2082 + }, + { + "epoch": 0.22875027454425653, + "grad_norm": 2.3367738723754883, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.695094883441925, + "num_tokens": 53680530.0, + "step": 2083 + }, + { + "epoch": 0.2288600922468702, + "grad_norm": 2.3907456398010254, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6966037154197693, + "num_tokens": 53702331.0, + "step": 2084 + }, + { + "epoch": 0.22896990994948385, + "grad_norm": 2.368527889251709, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7180232405662537, + "num_tokens": 53726066.0, + "step": 2085 + }, + { + "epoch": 0.22907972765209753, + "grad_norm": 2.396026372909546, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7055748701095581, + "num_tokens": 53746827.0, + "step": 2086 + }, + { + "epoch": 0.22918954535471117, + "grad_norm": 2.227108955383301, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7093086242675781, + "num_tokens": 53772487.0, + "step": 2087 + }, + { + "epoch": 0.22929936305732485, + "grad_norm": 2.4080967903137207, + "learning_rate": 1e-06, + "loss": 1.0899, + "mean_token_accuracy": 0.6838428974151611, + "num_tokens": 53796322.0, + "step": 2088 + }, + { + "epoch": 0.2294091807599385, + "grad_norm": 2.5124247074127197, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7043680548667908, + "num_tokens": 53819149.0, + "step": 2089 + }, + { + "epoch": 0.22951899846255217, + "grad_norm": 1.9527032375335693, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6928250789642334, + "num_tokens": 53853577.0, + "step": 2090 + }, + { + "epoch": 0.2296288161651658, + "grad_norm": 2.42451548576355, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.7053332328796387, + "num_tokens": 53877189.0, + "step": 2091 + }, + { + "epoch": 0.22973863386777948, + "grad_norm": 2.188154935836792, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6789007186889648, + "num_tokens": 53905812.0, + "step": 2092 + }, + { + "epoch": 0.22984845157039316, + "grad_norm": 1.9857285022735596, + "learning_rate": 1e-06, + "loss": 1.098, + "mean_token_accuracy": 0.6749659776687622, + "num_tokens": 53936948.0, + "step": 2093 + }, + { + "epoch": 0.2299582692730068, + "grad_norm": 2.3873519897460938, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.726202130317688, + "num_tokens": 53958236.0, + "step": 2094 + }, + { + "epoch": 0.23006808697562048, + "grad_norm": 2.2460103034973145, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6969115734100342, + "num_tokens": 53984217.0, + "step": 2095 + }, + { + "epoch": 0.23017790467823412, + "grad_norm": 2.7295095920562744, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6932855844497681, + "num_tokens": 54003397.0, + "step": 2096 + }, + { + "epoch": 0.2302877223808478, + "grad_norm": 1.995022177696228, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7363467216491699, + "num_tokens": 54031478.0, + "step": 2097 + }, + { + "epoch": 0.23039754008346144, + "grad_norm": 2.768292188644409, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7151480913162231, + "num_tokens": 54049390.0, + "step": 2098 + }, + { + "epoch": 0.23050735778607512, + "grad_norm": 2.166372299194336, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7036985158920288, + "num_tokens": 54078084.0, + "step": 2099 + }, + { + "epoch": 0.23061717548868876, + "grad_norm": 2.130510091781616, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.6931948661804199, + "num_tokens": 54108508.0, + "step": 2100 + }, + { + "epoch": 0.23072699319130244, + "grad_norm": 2.400660753250122, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6929774284362793, + "num_tokens": 54133948.0, + "step": 2101 + }, + { + "epoch": 0.2308368108939161, + "grad_norm": 2.124513626098633, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6950633525848389, + "num_tokens": 54163097.0, + "step": 2102 + }, + { + "epoch": 0.23094662859652976, + "grad_norm": 2.668822765350342, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6908900737762451, + "num_tokens": 54183715.0, + "step": 2103 + }, + { + "epoch": 0.23105644629914343, + "grad_norm": 2.206543445587158, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.688143789768219, + "num_tokens": 54210041.0, + "step": 2104 + }, + { + "epoch": 0.23116626400175708, + "grad_norm": 2.5073113441467285, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6990000009536743, + "num_tokens": 54230894.0, + "step": 2105 + }, + { + "epoch": 0.23127608170437075, + "grad_norm": 2.1166269779205322, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6937832236289978, + "num_tokens": 54257984.0, + "step": 2106 + }, + { + "epoch": 0.2313858994069844, + "grad_norm": 2.2569098472595215, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6922789812088013, + "num_tokens": 54285641.0, + "step": 2107 + }, + { + "epoch": 0.23149571710959807, + "grad_norm": 2.140119791030884, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6783647537231445, + "num_tokens": 54316697.0, + "step": 2108 + }, + { + "epoch": 0.23160553481221172, + "grad_norm": 2.4915573596954346, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6876266002655029, + "num_tokens": 54336842.0, + "step": 2109 + }, + { + "epoch": 0.2317153525148254, + "grad_norm": 2.264228582382202, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7105323076248169, + "num_tokens": 54359913.0, + "step": 2110 + }, + { + "epoch": 0.23182517021743906, + "grad_norm": 2.5434393882751465, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7017104029655457, + "num_tokens": 54380886.0, + "step": 2111 + }, + { + "epoch": 0.2319349879200527, + "grad_norm": 2.428682565689087, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7161683440208435, + "num_tokens": 54404840.0, + "step": 2112 + }, + { + "epoch": 0.23204480562266638, + "grad_norm": 2.4105136394500732, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.703189492225647, + "num_tokens": 54427542.0, + "step": 2113 + }, + { + "epoch": 0.23215462332528003, + "grad_norm": 2.487488269805908, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6921287178993225, + "num_tokens": 54449988.0, + "step": 2114 + }, + { + "epoch": 0.2322644410278937, + "grad_norm": 2.3272781372070312, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7002586722373962, + "num_tokens": 54475300.0, + "step": 2115 + }, + { + "epoch": 0.23237425873050735, + "grad_norm": 2.223875045776367, + "learning_rate": 1e-06, + "loss": 1.0718, + "mean_token_accuracy": 0.6857264041900635, + "num_tokens": 54502669.0, + "step": 2116 + }, + { + "epoch": 0.23248407643312102, + "grad_norm": 2.1063008308410645, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7092489004135132, + "num_tokens": 54531160.0, + "step": 2117 + }, + { + "epoch": 0.23259389413573467, + "grad_norm": 2.3986074924468994, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6926884651184082, + "num_tokens": 54556316.0, + "step": 2118 + }, + { + "epoch": 0.23270371183834834, + "grad_norm": 2.3231396675109863, + "learning_rate": 1e-06, + "loss": 1.0689, + "mean_token_accuracy": 0.682580828666687, + "num_tokens": 54582909.0, + "step": 2119 + }, + { + "epoch": 0.23281352954096202, + "grad_norm": 2.485222578048706, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6783390045166016, + "num_tokens": 54605478.0, + "step": 2120 + }, + { + "epoch": 0.23292334724357566, + "grad_norm": 2.297211170196533, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7298014163970947, + "num_tokens": 54627021.0, + "step": 2121 + }, + { + "epoch": 0.23303316494618934, + "grad_norm": 2.3277454376220703, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7006120681762695, + "num_tokens": 54651208.0, + "step": 2122 + }, + { + "epoch": 0.23314298264880298, + "grad_norm": 2.4952025413513184, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6949772238731384, + "num_tokens": 54675392.0, + "step": 2123 + }, + { + "epoch": 0.23325280035141666, + "grad_norm": 2.5350635051727295, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6931915283203125, + "num_tokens": 54695086.0, + "step": 2124 + }, + { + "epoch": 0.2333626180540303, + "grad_norm": 2.3673765659332275, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6949383020401001, + "num_tokens": 54719821.0, + "step": 2125 + }, + { + "epoch": 0.23347243575664398, + "grad_norm": 2.184892416000366, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6994779109954834, + "num_tokens": 54746382.0, + "step": 2126 + }, + { + "epoch": 0.23358225345925762, + "grad_norm": 2.4738609790802, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6928903460502625, + "num_tokens": 54767780.0, + "step": 2127 + }, + { + "epoch": 0.2336920711618713, + "grad_norm": 2.1164445877075195, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6903035640716553, + "num_tokens": 54797862.0, + "step": 2128 + }, + { + "epoch": 0.23380188886448494, + "grad_norm": 2.0166640281677246, + "learning_rate": 1e-06, + "loss": 1.1337, + "mean_token_accuracy": 0.6614212393760681, + "num_tokens": 54832121.0, + "step": 2129 + }, + { + "epoch": 0.23391170656709862, + "grad_norm": 2.0374045372009277, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7003329992294312, + "num_tokens": 54864382.0, + "step": 2130 + }, + { + "epoch": 0.2340215242697123, + "grad_norm": 2.33148193359375, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7079757452011108, + "num_tokens": 54887056.0, + "step": 2131 + }, + { + "epoch": 0.23413134197232593, + "grad_norm": 2.276557445526123, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.695924699306488, + "num_tokens": 54912345.0, + "step": 2132 + }, + { + "epoch": 0.2342411596749396, + "grad_norm": 2.236643075942993, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7045960426330566, + "num_tokens": 54938091.0, + "step": 2133 + }, + { + "epoch": 0.23435097737755325, + "grad_norm": 2.2450191974639893, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.688156008720398, + "num_tokens": 54965428.0, + "step": 2134 + }, + { + "epoch": 0.23446079508016693, + "grad_norm": 2.7996208667755127, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6997302770614624, + "num_tokens": 54988030.0, + "step": 2135 + }, + { + "epoch": 0.23457061278278057, + "grad_norm": 2.821115255355835, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.706524133682251, + "num_tokens": 55004383.0, + "step": 2136 + }, + { + "epoch": 0.23468043048539425, + "grad_norm": 2.2038824558258057, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6880936026573181, + "num_tokens": 55030621.0, + "step": 2137 + }, + { + "epoch": 0.2347902481880079, + "grad_norm": 2.5124471187591553, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6787312030792236, + "num_tokens": 55054575.0, + "step": 2138 + }, + { + "epoch": 0.23490006589062157, + "grad_norm": 2.3049099445343018, + "learning_rate": 1e-06, + "loss": 1.0918, + "mean_token_accuracy": 0.6731084585189819, + "num_tokens": 55081851.0, + "step": 2139 + }, + { + "epoch": 0.23500988359323524, + "grad_norm": 2.411586046218872, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7298396825790405, + "num_tokens": 55103161.0, + "step": 2140 + }, + { + "epoch": 0.2351197012958489, + "grad_norm": 2.0702927112579346, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.699982762336731, + "num_tokens": 55133473.0, + "step": 2141 + }, + { + "epoch": 0.23522951899846256, + "grad_norm": 2.2389273643493652, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.681550145149231, + "num_tokens": 55160174.0, + "step": 2142 + }, + { + "epoch": 0.2353393367010762, + "grad_norm": 1.93410062789917, + "learning_rate": 1e-06, + "loss": 1.0993, + "mean_token_accuracy": 0.6780738830566406, + "num_tokens": 55194613.0, + "step": 2143 + }, + { + "epoch": 0.23544915440368988, + "grad_norm": 2.032820463180542, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7209882736206055, + "num_tokens": 55224346.0, + "step": 2144 + }, + { + "epoch": 0.23555897210630353, + "grad_norm": 2.255669116973877, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6902844905853271, + "num_tokens": 55250034.0, + "step": 2145 + }, + { + "epoch": 0.2356687898089172, + "grad_norm": 2.325131893157959, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7034099698066711, + "num_tokens": 55275200.0, + "step": 2146 + }, + { + "epoch": 0.23577860751153085, + "grad_norm": 2.268108606338501, + "learning_rate": 1e-06, + "loss": 1.1283, + "mean_token_accuracy": 0.667184591293335, + "num_tokens": 55302514.0, + "step": 2147 + }, + { + "epoch": 0.23588842521414452, + "grad_norm": 1.9785939455032349, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7102693319320679, + "num_tokens": 55334426.0, + "step": 2148 + }, + { + "epoch": 0.2359982429167582, + "grad_norm": 2.5073423385620117, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7203879356384277, + "num_tokens": 55354262.0, + "step": 2149 + }, + { + "epoch": 0.23610806061937184, + "grad_norm": 2.3450868129730225, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6939976811408997, + "num_tokens": 55378500.0, + "step": 2150 + }, + { + "epoch": 0.2362178783219855, + "grad_norm": 2.2386066913604736, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.70057612657547, + "num_tokens": 55404273.0, + "step": 2151 + }, + { + "epoch": 0.23632769602459916, + "grad_norm": 2.2009196281433105, + "learning_rate": 1e-06, + "loss": 1.0971, + "mean_token_accuracy": 0.6809767484664917, + "num_tokens": 55434053.0, + "step": 2152 + }, + { + "epoch": 0.23643751372721283, + "grad_norm": 2.190314769744873, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7184138298034668, + "num_tokens": 55460378.0, + "step": 2153 + }, + { + "epoch": 0.23654733142982648, + "grad_norm": 2.281353712081909, + "learning_rate": 1e-06, + "loss": 1.1005, + "mean_token_accuracy": 0.6738834977149963, + "num_tokens": 55486136.0, + "step": 2154 + }, + { + "epoch": 0.23665714913244015, + "grad_norm": 1.9902458190917969, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.72215735912323, + "num_tokens": 55516600.0, + "step": 2155 + }, + { + "epoch": 0.2367669668350538, + "grad_norm": 2.376535177230835, + "learning_rate": 1e-06, + "loss": 1.0788, + "mean_token_accuracy": 0.6820389032363892, + "num_tokens": 55542042.0, + "step": 2156 + }, + { + "epoch": 0.23687678453766747, + "grad_norm": 2.578279733657837, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7125227451324463, + "num_tokens": 55561766.0, + "step": 2157 + }, + { + "epoch": 0.23698660224028115, + "grad_norm": 2.2573156356811523, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6992217302322388, + "num_tokens": 55587044.0, + "step": 2158 + }, + { + "epoch": 0.2370964199428948, + "grad_norm": 2.3076822757720947, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6993801593780518, + "num_tokens": 55610880.0, + "step": 2159 + }, + { + "epoch": 0.23720623764550847, + "grad_norm": 2.461411952972412, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7039207220077515, + "num_tokens": 55634676.0, + "step": 2160 + }, + { + "epoch": 0.2373160553481221, + "grad_norm": 2.458811044692993, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7093227505683899, + "num_tokens": 55656224.0, + "step": 2161 + }, + { + "epoch": 0.2374258730507358, + "grad_norm": 2.199308156967163, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7059910297393799, + "num_tokens": 55683081.0, + "step": 2162 + }, + { + "epoch": 0.23753569075334943, + "grad_norm": 2.5013036727905273, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7043479084968567, + "num_tokens": 55705178.0, + "step": 2163 + }, + { + "epoch": 0.2376455084559631, + "grad_norm": 2.4302399158477783, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6952697038650513, + "num_tokens": 55728186.0, + "step": 2164 + }, + { + "epoch": 0.23775532615857675, + "grad_norm": 1.9990432262420654, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7042351961135864, + "num_tokens": 55758971.0, + "step": 2165 + }, + { + "epoch": 0.23786514386119043, + "grad_norm": 2.225715398788452, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6957783699035645, + "num_tokens": 55784731.0, + "step": 2166 + }, + { + "epoch": 0.23797496156380407, + "grad_norm": 2.2113165855407715, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.68604576587677, + "num_tokens": 55812510.0, + "step": 2167 + }, + { + "epoch": 0.23808477926641775, + "grad_norm": 2.121854543685913, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7060288786888123, + "num_tokens": 55840202.0, + "step": 2168 + }, + { + "epoch": 0.23819459696903142, + "grad_norm": 2.231642961502075, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6862412691116333, + "num_tokens": 55865961.0, + "step": 2169 + }, + { + "epoch": 0.23830441467164507, + "grad_norm": 2.260030746459961, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.71098393201828, + "num_tokens": 55892346.0, + "step": 2170 + }, + { + "epoch": 0.23841423237425874, + "grad_norm": 2.303894519805908, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6977165937423706, + "num_tokens": 55918580.0, + "step": 2171 + }, + { + "epoch": 0.23852405007687238, + "grad_norm": 2.382042646408081, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7110454440116882, + "num_tokens": 55940615.0, + "step": 2172 + }, + { + "epoch": 0.23863386777948606, + "grad_norm": 2.125396251678467, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6918449401855469, + "num_tokens": 55968339.0, + "step": 2173 + }, + { + "epoch": 0.2387436854820997, + "grad_norm": 1.9779876470565796, + "learning_rate": 1e-06, + "loss": 1.0779, + "mean_token_accuracy": 0.6791164875030518, + "num_tokens": 56002061.0, + "step": 2174 + }, + { + "epoch": 0.23885350318471338, + "grad_norm": 2.236577272415161, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6994472742080688, + "num_tokens": 56029109.0, + "step": 2175 + }, + { + "epoch": 0.23896332088732702, + "grad_norm": 2.0144717693328857, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6801524758338928, + "num_tokens": 56062167.0, + "step": 2176 + }, + { + "epoch": 0.2390731385899407, + "grad_norm": 2.5663251876831055, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.708981990814209, + "num_tokens": 56081011.0, + "step": 2177 + }, + { + "epoch": 0.23918295629255437, + "grad_norm": 2.5041778087615967, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6979352235794067, + "num_tokens": 56102702.0, + "step": 2178 + }, + { + "epoch": 0.23929277399516802, + "grad_norm": 2.023951530456543, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6792406439781189, + "num_tokens": 56133729.0, + "step": 2179 + }, + { + "epoch": 0.2394025916977817, + "grad_norm": 2.3170275688171387, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6912540793418884, + "num_tokens": 56159205.0, + "step": 2180 + }, + { + "epoch": 0.23951240940039534, + "grad_norm": 2.3473050594329834, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6997413635253906, + "num_tokens": 56183491.0, + "step": 2181 + }, + { + "epoch": 0.239622227103009, + "grad_norm": 2.2681643962860107, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7003469467163086, + "num_tokens": 56208091.0, + "step": 2182 + }, + { + "epoch": 0.23973204480562266, + "grad_norm": 2.256693124771118, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6853723526000977, + "num_tokens": 56234398.0, + "step": 2183 + }, + { + "epoch": 0.23984186250823633, + "grad_norm": 2.131232976913452, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6990726590156555, + "num_tokens": 56259605.0, + "step": 2184 + }, + { + "epoch": 0.23995168021084998, + "grad_norm": 2.0824899673461914, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.7012615203857422, + "num_tokens": 56287472.0, + "step": 2185 + }, + { + "epoch": 0.24006149791346365, + "grad_norm": 2.3579561710357666, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7166745662689209, + "num_tokens": 56311608.0, + "step": 2186 + }, + { + "epoch": 0.24017131561607732, + "grad_norm": 2.4191229343414307, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6943540573120117, + "num_tokens": 56333653.0, + "step": 2187 + }, + { + "epoch": 0.24028113331869097, + "grad_norm": 1.8914930820465088, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.690350353717804, + "num_tokens": 56368749.0, + "step": 2188 + }, + { + "epoch": 0.24039095102130464, + "grad_norm": 2.1534478664398193, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6971381902694702, + "num_tokens": 56396570.0, + "step": 2189 + }, + { + "epoch": 0.2405007687239183, + "grad_norm": 2.2888946533203125, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6902493834495544, + "num_tokens": 56421987.0, + "step": 2190 + }, + { + "epoch": 0.24061058642653196, + "grad_norm": 2.108001470565796, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7043412923812866, + "num_tokens": 56451045.0, + "step": 2191 + }, + { + "epoch": 0.2407204041291456, + "grad_norm": 2.4025392532348633, + "learning_rate": 1e-06, + "loss": 1.0747, + "mean_token_accuracy": 0.6774041652679443, + "num_tokens": 56474696.0, + "step": 2192 + }, + { + "epoch": 0.24083022183175928, + "grad_norm": 2.2654592990875244, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7119809985160828, + "num_tokens": 56498952.0, + "step": 2193 + }, + { + "epoch": 0.24094003953437293, + "grad_norm": 2.407863140106201, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7095125317573547, + "num_tokens": 56524164.0, + "step": 2194 + }, + { + "epoch": 0.2410498572369866, + "grad_norm": 2.383402109146118, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.705757737159729, + "num_tokens": 56547152.0, + "step": 2195 + }, + { + "epoch": 0.24115967493960028, + "grad_norm": 2.2016844749450684, + "learning_rate": 1e-06, + "loss": 1.0769, + "mean_token_accuracy": 0.6796762347221375, + "num_tokens": 56575922.0, + "step": 2196 + }, + { + "epoch": 0.24126949264221392, + "grad_norm": 2.03590726852417, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6910178661346436, + "num_tokens": 56607385.0, + "step": 2197 + }, + { + "epoch": 0.2413793103448276, + "grad_norm": 2.2010419368743896, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6884381175041199, + "num_tokens": 56634368.0, + "step": 2198 + }, + { + "epoch": 0.24148912804744124, + "grad_norm": 2.2438457012176514, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7013024687767029, + "num_tokens": 56659567.0, + "step": 2199 + }, + { + "epoch": 0.24159894575005492, + "grad_norm": 2.6108481884002686, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.6913957595825195, + "num_tokens": 56681934.0, + "step": 2200 + }, + { + "epoch": 0.24170876345266856, + "grad_norm": 2.1962506771087646, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7274971008300781, + "num_tokens": 56706640.0, + "step": 2201 + }, + { + "epoch": 0.24181858115528224, + "grad_norm": 2.062504529953003, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6848769783973694, + "num_tokens": 56735792.0, + "step": 2202 + }, + { + "epoch": 0.24192839885789588, + "grad_norm": 2.0803542137145996, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6853041052818298, + "num_tokens": 56764243.0, + "step": 2203 + }, + { + "epoch": 0.24203821656050956, + "grad_norm": 2.118759870529175, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6779468655586243, + "num_tokens": 56793600.0, + "step": 2204 + }, + { + "epoch": 0.2421480342631232, + "grad_norm": 2.2832202911376953, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6942631602287292, + "num_tokens": 56819184.0, + "step": 2205 + }, + { + "epoch": 0.24225785196573688, + "grad_norm": 2.5484397411346436, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6956099271774292, + "num_tokens": 56841179.0, + "step": 2206 + }, + { + "epoch": 0.24236766966835055, + "grad_norm": 2.3358514308929443, + "learning_rate": 1e-06, + "loss": 1.0861, + "mean_token_accuracy": 0.6724059581756592, + "num_tokens": 56866356.0, + "step": 2207 + }, + { + "epoch": 0.2424774873709642, + "grad_norm": 2.093752861022949, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7156305313110352, + "num_tokens": 56895082.0, + "step": 2208 + }, + { + "epoch": 0.24258730507357787, + "grad_norm": 2.042813777923584, + "learning_rate": 1e-06, + "loss": 1.0865, + "mean_token_accuracy": 0.6799126267433167, + "num_tokens": 56927630.0, + "step": 2209 + }, + { + "epoch": 0.24269712277619152, + "grad_norm": 2.5292723178863525, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6940513849258423, + "num_tokens": 56949632.0, + "step": 2210 + }, + { + "epoch": 0.2428069404788052, + "grad_norm": 2.2207789421081543, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7065564393997192, + "num_tokens": 56975850.0, + "step": 2211 + }, + { + "epoch": 0.24291675818141883, + "grad_norm": 2.217960834503174, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7067294120788574, + "num_tokens": 57001710.0, + "step": 2212 + }, + { + "epoch": 0.2430265758840325, + "grad_norm": 2.3046061992645264, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6875332593917847, + "num_tokens": 57026527.0, + "step": 2213 + }, + { + "epoch": 0.24313639358664615, + "grad_norm": 2.1283106803894043, + "learning_rate": 1e-06, + "loss": 1.1054, + "mean_token_accuracy": 0.6702584624290466, + "num_tokens": 57056550.0, + "step": 2214 + }, + { + "epoch": 0.24324621128925983, + "grad_norm": 2.1485607624053955, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7023865580558777, + "num_tokens": 57084976.0, + "step": 2215 + }, + { + "epoch": 0.2433560289918735, + "grad_norm": 2.036139488220215, + "learning_rate": 1e-06, + "loss": 1.1296, + "mean_token_accuracy": 0.674862265586853, + "num_tokens": 57117283.0, + "step": 2216 + }, + { + "epoch": 0.24346584669448715, + "grad_norm": 2.1661376953125, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7082311511039734, + "num_tokens": 57145405.0, + "step": 2217 + }, + { + "epoch": 0.24357566439710082, + "grad_norm": 2.366379499435425, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6927218437194824, + "num_tokens": 57169739.0, + "step": 2218 + }, + { + "epoch": 0.24368548209971447, + "grad_norm": 2.5446314811706543, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.674657940864563, + "num_tokens": 57191820.0, + "step": 2219 + }, + { + "epoch": 0.24379529980232814, + "grad_norm": 2.1093881130218506, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6900228261947632, + "num_tokens": 57222318.0, + "step": 2220 + }, + { + "epoch": 0.2439051175049418, + "grad_norm": 2.39888596534729, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.713141679763794, + "num_tokens": 57244770.0, + "step": 2221 + }, + { + "epoch": 0.24401493520755546, + "grad_norm": 2.030921220779419, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6860338449478149, + "num_tokens": 57274761.0, + "step": 2222 + }, + { + "epoch": 0.2441247529101691, + "grad_norm": 2.288703680038452, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7099180221557617, + "num_tokens": 57297789.0, + "step": 2223 + }, + { + "epoch": 0.24423457061278278, + "grad_norm": 2.014528274536133, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6987953782081604, + "num_tokens": 57327021.0, + "step": 2224 + }, + { + "epoch": 0.24434438831539645, + "grad_norm": 2.07092547416687, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7003864645957947, + "num_tokens": 57356404.0, + "step": 2225 + }, + { + "epoch": 0.2444542060180101, + "grad_norm": 2.304746627807617, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7077668905258179, + "num_tokens": 57381325.0, + "step": 2226 + }, + { + "epoch": 0.24456402372062377, + "grad_norm": 2.348062753677368, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.702606201171875, + "num_tokens": 57404025.0, + "step": 2227 + }, + { + "epoch": 0.24467384142323742, + "grad_norm": 2.402388095855713, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.717951774597168, + "num_tokens": 57424935.0, + "step": 2228 + }, + { + "epoch": 0.2447836591258511, + "grad_norm": 2.203272819519043, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6839902400970459, + "num_tokens": 57453884.0, + "step": 2229 + }, + { + "epoch": 0.24489347682846474, + "grad_norm": 2.2744178771972656, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6977531909942627, + "num_tokens": 57477618.0, + "step": 2230 + }, + { + "epoch": 0.2450032945310784, + "grad_norm": 2.0748074054718018, + "learning_rate": 1e-06, + "loss": 1.0864, + "mean_token_accuracy": 0.6837489604949951, + "num_tokens": 57508714.0, + "step": 2231 + }, + { + "epoch": 0.24511311223369206, + "grad_norm": 2.3174123764038086, + "learning_rate": 1e-06, + "loss": 1.0658, + "mean_token_accuracy": 0.6826990842819214, + "num_tokens": 57533554.0, + "step": 2232 + }, + { + "epoch": 0.24522292993630573, + "grad_norm": 2.3052151203155518, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6919515132904053, + "num_tokens": 57559149.0, + "step": 2233 + }, + { + "epoch": 0.2453327476389194, + "grad_norm": 2.4782025814056396, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6869140863418579, + "num_tokens": 57582162.0, + "step": 2234 + }, + { + "epoch": 0.24544256534153305, + "grad_norm": 2.631674289703369, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7086073160171509, + "num_tokens": 57601139.0, + "step": 2235 + }, + { + "epoch": 0.24555238304414673, + "grad_norm": 2.306539297103882, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7128427624702454, + "num_tokens": 57624809.0, + "step": 2236 + }, + { + "epoch": 0.24566220074676037, + "grad_norm": 2.1692733764648438, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6963045597076416, + "num_tokens": 57652461.0, + "step": 2237 + }, + { + "epoch": 0.24577201844937405, + "grad_norm": 2.1204447746276855, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6766924262046814, + "num_tokens": 57680572.0, + "step": 2238 + }, + { + "epoch": 0.2458818361519877, + "grad_norm": 2.1881258487701416, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6981337070465088, + "num_tokens": 57708803.0, + "step": 2239 + }, + { + "epoch": 0.24599165385460137, + "grad_norm": 2.2023158073425293, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6886061429977417, + "num_tokens": 57738086.0, + "step": 2240 + }, + { + "epoch": 0.246101471557215, + "grad_norm": 2.337677240371704, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6876160502433777, + "num_tokens": 57761344.0, + "step": 2241 + }, + { + "epoch": 0.2462112892598287, + "grad_norm": 2.562899112701416, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7165243625640869, + "num_tokens": 57780447.0, + "step": 2242 + }, + { + "epoch": 0.24632110696244233, + "grad_norm": 2.2357349395751953, + "learning_rate": 1e-06, + "loss": 1.0897, + "mean_token_accuracy": 0.6729722023010254, + "num_tokens": 57806840.0, + "step": 2243 + }, + { + "epoch": 0.246430924665056, + "grad_norm": 2.1594223976135254, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7051665782928467, + "num_tokens": 57837058.0, + "step": 2244 + }, + { + "epoch": 0.24654074236766968, + "grad_norm": 2.293804883956909, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7145751714706421, + "num_tokens": 57862011.0, + "step": 2245 + }, + { + "epoch": 0.24665056007028333, + "grad_norm": 2.2995665073394775, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7152601480484009, + "num_tokens": 57885265.0, + "step": 2246 + }, + { + "epoch": 0.246760377772897, + "grad_norm": 1.9384360313415527, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7196659445762634, + "num_tokens": 57917653.0, + "step": 2247 + }, + { + "epoch": 0.24687019547551065, + "grad_norm": 2.3514492511749268, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7042169570922852, + "num_tokens": 57940393.0, + "step": 2248 + }, + { + "epoch": 0.24698001317812432, + "grad_norm": 2.143839120864868, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6865828037261963, + "num_tokens": 57967468.0, + "step": 2249 + }, + { + "epoch": 0.24708983088073797, + "grad_norm": 2.0267727375030518, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6764915585517883, + "num_tokens": 57997833.0, + "step": 2250 + }, + { + "epoch": 0.24719964858335164, + "grad_norm": 2.1803815364837646, + "learning_rate": 1e-06, + "loss": 1.0995, + "mean_token_accuracy": 0.6776083707809448, + "num_tokens": 58027554.0, + "step": 2251 + }, + { + "epoch": 0.24730946628596528, + "grad_norm": 2.2141621112823486, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7100117206573486, + "num_tokens": 58054779.0, + "step": 2252 + }, + { + "epoch": 0.24741928398857896, + "grad_norm": 2.2928755283355713, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7183126211166382, + "num_tokens": 58079173.0, + "step": 2253 + }, + { + "epoch": 0.24752910169119263, + "grad_norm": 2.295788288116455, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7087022662162781, + "num_tokens": 58102942.0, + "step": 2254 + }, + { + "epoch": 0.24763891939380628, + "grad_norm": 2.0917181968688965, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.7028392553329468, + "num_tokens": 58130411.0, + "step": 2255 + }, + { + "epoch": 0.24774873709641995, + "grad_norm": 2.2053959369659424, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6962723731994629, + "num_tokens": 58156607.0, + "step": 2256 + }, + { + "epoch": 0.2478585547990336, + "grad_norm": 1.9042922258377075, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6834160685539246, + "num_tokens": 58192594.0, + "step": 2257 + }, + { + "epoch": 0.24796837250164727, + "grad_norm": 2.124673843383789, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.696653425693512, + "num_tokens": 58218544.0, + "step": 2258 + }, + { + "epoch": 0.24807819020426092, + "grad_norm": 2.2220661640167236, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6736269593238831, + "num_tokens": 58244762.0, + "step": 2259 + }, + { + "epoch": 0.2481880079068746, + "grad_norm": 2.11498761177063, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6943759918212891, + "num_tokens": 58272170.0, + "step": 2260 + }, + { + "epoch": 0.24829782560948824, + "grad_norm": 2.171441078186035, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7068612575531006, + "num_tokens": 58298491.0, + "step": 2261 + }, + { + "epoch": 0.2484076433121019, + "grad_norm": 2.2786495685577393, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6990083456039429, + "num_tokens": 58323390.0, + "step": 2262 + }, + { + "epoch": 0.24851746101471558, + "grad_norm": 2.314069986343384, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.710808515548706, + "num_tokens": 58346356.0, + "step": 2263 + }, + { + "epoch": 0.24862727871732923, + "grad_norm": 2.0796966552734375, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6818473935127258, + "num_tokens": 58378169.0, + "step": 2264 + }, + { + "epoch": 0.2487370964199429, + "grad_norm": 2.398177146911621, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6883213520050049, + "num_tokens": 58402748.0, + "step": 2265 + }, + { + "epoch": 0.24884691412255655, + "grad_norm": 2.556624412536621, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6969926953315735, + "num_tokens": 58423011.0, + "step": 2266 + }, + { + "epoch": 0.24895673182517022, + "grad_norm": 2.4852514266967773, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6959193348884583, + "num_tokens": 58444823.0, + "step": 2267 + }, + { + "epoch": 0.24906654952778387, + "grad_norm": 2.219801902770996, + "learning_rate": 1e-06, + "loss": 1.1082, + "mean_token_accuracy": 0.6762192249298096, + "num_tokens": 58472478.0, + "step": 2268 + }, + { + "epoch": 0.24917636723039754, + "grad_norm": 2.2360174655914307, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.700989842414856, + "num_tokens": 58497113.0, + "step": 2269 + }, + { + "epoch": 0.2492861849330112, + "grad_norm": 2.408843517303467, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7091608047485352, + "num_tokens": 58518222.0, + "step": 2270 + }, + { + "epoch": 0.24939600263562486, + "grad_norm": 2.077819585800171, + "learning_rate": 1e-06, + "loss": 1.1059, + "mean_token_accuracy": 0.6734042167663574, + "num_tokens": 58549467.0, + "step": 2271 + }, + { + "epoch": 0.24950582033823854, + "grad_norm": 2.0566487312316895, + "learning_rate": 1e-06, + "loss": 1.072, + "mean_token_accuracy": 0.6878093481063843, + "num_tokens": 58578747.0, + "step": 2272 + }, + { + "epoch": 0.24961563804085218, + "grad_norm": 2.1855661869049072, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7152503132820129, + "num_tokens": 58605184.0, + "step": 2273 + }, + { + "epoch": 0.24972545574346586, + "grad_norm": 2.2694480419158936, + "learning_rate": 1e-06, + "loss": 1.1185, + "mean_token_accuracy": 0.6837178468704224, + "num_tokens": 58631057.0, + "step": 2274 + }, + { + "epoch": 0.2498352734460795, + "grad_norm": 2.404259204864502, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6774614453315735, + "num_tokens": 58654556.0, + "step": 2275 + }, + { + "epoch": 0.24994509114869318, + "grad_norm": 2.3578925132751465, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6784099340438843, + "num_tokens": 58679295.0, + "step": 2276 + }, + { + "epoch": 0.2500549088513068, + "grad_norm": 2.1977505683898926, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6934809684753418, + "num_tokens": 58705273.0, + "step": 2277 + }, + { + "epoch": 0.25016472655392047, + "grad_norm": 2.152202844619751, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7239773869514465, + "num_tokens": 58731333.0, + "step": 2278 + }, + { + "epoch": 0.25027454425653417, + "grad_norm": 2.316138505935669, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7146216630935669, + "num_tokens": 58753114.0, + "step": 2279 + }, + { + "epoch": 0.2503843619591478, + "grad_norm": 2.2255797386169434, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6934901475906372, + "num_tokens": 58778908.0, + "step": 2280 + }, + { + "epoch": 0.25049417966176146, + "grad_norm": 2.294020891189575, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6977447867393494, + "num_tokens": 58803946.0, + "step": 2281 + }, + { + "epoch": 0.25060399736437516, + "grad_norm": 2.4150772094726562, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7086244821548462, + "num_tokens": 58826826.0, + "step": 2282 + }, + { + "epoch": 0.2507138150669888, + "grad_norm": 2.45986008644104, + "learning_rate": 1e-06, + "loss": 1.1123, + "mean_token_accuracy": 0.6683562994003296, + "num_tokens": 58850337.0, + "step": 2283 + }, + { + "epoch": 0.25082363276960246, + "grad_norm": 2.1840415000915527, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7052375078201294, + "num_tokens": 58877082.0, + "step": 2284 + }, + { + "epoch": 0.2509334504722161, + "grad_norm": 2.1614832878112793, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7184958457946777, + "num_tokens": 58903450.0, + "step": 2285 + }, + { + "epoch": 0.2510432681748298, + "grad_norm": 2.2191410064697266, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7050470113754272, + "num_tokens": 58926906.0, + "step": 2286 + }, + { + "epoch": 0.25115308587744345, + "grad_norm": 2.1294968128204346, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7087357640266418, + "num_tokens": 58952459.0, + "step": 2287 + }, + { + "epoch": 0.2512629035800571, + "grad_norm": 2.2044520378112793, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6853725910186768, + "num_tokens": 58980453.0, + "step": 2288 + }, + { + "epoch": 0.25137272128267074, + "grad_norm": 2.0148885250091553, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6945468187332153, + "num_tokens": 59012195.0, + "step": 2289 + }, + { + "epoch": 0.25148253898528444, + "grad_norm": 2.2372233867645264, + "learning_rate": 1e-06, + "loss": 1.0981, + "mean_token_accuracy": 0.6752973794937134, + "num_tokens": 59039230.0, + "step": 2290 + }, + { + "epoch": 0.2515923566878981, + "grad_norm": 2.2395591735839844, + "learning_rate": 1e-06, + "loss": 1.0789, + "mean_token_accuracy": 0.6834766268730164, + "num_tokens": 59065466.0, + "step": 2291 + }, + { + "epoch": 0.25170217439051173, + "grad_norm": 2.3271758556365967, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7048296332359314, + "num_tokens": 59088205.0, + "step": 2292 + }, + { + "epoch": 0.25181199209312544, + "grad_norm": 2.4299938678741455, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.714540958404541, + "num_tokens": 59108993.0, + "step": 2293 + }, + { + "epoch": 0.2519218097957391, + "grad_norm": 2.460501194000244, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6943663954734802, + "num_tokens": 59131124.0, + "step": 2294 + }, + { + "epoch": 0.25203162749835273, + "grad_norm": 2.511068344116211, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7034578919410706, + "num_tokens": 59153311.0, + "step": 2295 + }, + { + "epoch": 0.2521414452009664, + "grad_norm": 2.2202162742614746, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6984696984291077, + "num_tokens": 59180226.0, + "step": 2296 + }, + { + "epoch": 0.2522512629035801, + "grad_norm": 2.3496179580688477, + "learning_rate": 1e-06, + "loss": 1.1148, + "mean_token_accuracy": 0.6697860956192017, + "num_tokens": 59208861.0, + "step": 2297 + }, + { + "epoch": 0.2523610806061937, + "grad_norm": 2.065042495727539, + "learning_rate": 1e-06, + "loss": 1.1261, + "mean_token_accuracy": 0.6623475551605225, + "num_tokens": 59240200.0, + "step": 2298 + }, + { + "epoch": 0.25247089830880737, + "grad_norm": 2.1455166339874268, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6873801350593567, + "num_tokens": 59268427.0, + "step": 2299 + }, + { + "epoch": 0.252580716011421, + "grad_norm": 2.1815929412841797, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6899536848068237, + "num_tokens": 59297564.0, + "step": 2300 + }, + { + "epoch": 0.2526905337140347, + "grad_norm": 2.173105001449585, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.688382625579834, + "num_tokens": 59324628.0, + "step": 2301 + }, + { + "epoch": 0.25280035141664836, + "grad_norm": 2.091824531555176, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7100462913513184, + "num_tokens": 59352077.0, + "step": 2302 + }, + { + "epoch": 0.252910169119262, + "grad_norm": 2.2574424743652344, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7309545278549194, + "num_tokens": 59376694.0, + "step": 2303 + }, + { + "epoch": 0.2530199868218757, + "grad_norm": 1.9737167358398438, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7024343013763428, + "num_tokens": 59407108.0, + "step": 2304 + }, + { + "epoch": 0.25312980452448935, + "grad_norm": 2.167945623397827, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.700465738773346, + "num_tokens": 59432338.0, + "step": 2305 + }, + { + "epoch": 0.253239622227103, + "grad_norm": 2.26161527633667, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.677458643913269, + "num_tokens": 59460173.0, + "step": 2306 + }, + { + "epoch": 0.25334943992971665, + "grad_norm": 2.3444955348968506, + "learning_rate": 1e-06, + "loss": 1.0491, + "mean_token_accuracy": 0.6956762671470642, + "num_tokens": 59483519.0, + "step": 2307 + }, + { + "epoch": 0.25345925763233035, + "grad_norm": 2.225778102874756, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6974275708198547, + "num_tokens": 59511468.0, + "step": 2308 + }, + { + "epoch": 0.253569075334944, + "grad_norm": 2.1622989177703857, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6980560421943665, + "num_tokens": 59539075.0, + "step": 2309 + }, + { + "epoch": 0.25367889303755764, + "grad_norm": 1.9912890195846558, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7085791826248169, + "num_tokens": 59568283.0, + "step": 2310 + }, + { + "epoch": 0.25378871074017134, + "grad_norm": 2.4044597148895264, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.703292727470398, + "num_tokens": 59595818.0, + "step": 2311 + }, + { + "epoch": 0.253898528442785, + "grad_norm": 2.12199330329895, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7096213102340698, + "num_tokens": 59621935.0, + "step": 2312 + }, + { + "epoch": 0.25400834614539863, + "grad_norm": 2.0051136016845703, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6965787410736084, + "num_tokens": 59652693.0, + "step": 2313 + }, + { + "epoch": 0.2541181638480123, + "grad_norm": 2.398155927658081, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7140582799911499, + "num_tokens": 59675135.0, + "step": 2314 + }, + { + "epoch": 0.254227981550626, + "grad_norm": 2.214160680770874, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6898049116134644, + "num_tokens": 59701092.0, + "step": 2315 + }, + { + "epoch": 0.2543377992532396, + "grad_norm": 2.325782060623169, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7015606164932251, + "num_tokens": 59725603.0, + "step": 2316 + }, + { + "epoch": 0.2544476169558533, + "grad_norm": 2.2986843585968018, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6927133798599243, + "num_tokens": 59749765.0, + "step": 2317 + }, + { + "epoch": 0.2545574346584669, + "grad_norm": 2.248033285140991, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6989070773124695, + "num_tokens": 59774182.0, + "step": 2318 + }, + { + "epoch": 0.2546672523610806, + "grad_norm": 2.4669909477233887, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7227603197097778, + "num_tokens": 59795453.0, + "step": 2319 + }, + { + "epoch": 0.25477707006369427, + "grad_norm": 2.194521188735962, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6969100832939148, + "num_tokens": 59820299.0, + "step": 2320 + }, + { + "epoch": 0.2548868877663079, + "grad_norm": 2.058344602584839, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6904593706130981, + "num_tokens": 59850019.0, + "step": 2321 + }, + { + "epoch": 0.2549967054689216, + "grad_norm": 2.4623756408691406, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6989400386810303, + "num_tokens": 59870866.0, + "step": 2322 + }, + { + "epoch": 0.25510652317153526, + "grad_norm": 2.302438497543335, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.6725720167160034, + "num_tokens": 59894144.0, + "step": 2323 + }, + { + "epoch": 0.2552163408741489, + "grad_norm": 2.241426467895508, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7030642032623291, + "num_tokens": 59918623.0, + "step": 2324 + }, + { + "epoch": 0.25532615857676255, + "grad_norm": 2.103468179702759, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6931123733520508, + "num_tokens": 59946443.0, + "step": 2325 + }, + { + "epoch": 0.25543597627937625, + "grad_norm": 2.4156486988067627, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7230675220489502, + "num_tokens": 59967145.0, + "step": 2326 + }, + { + "epoch": 0.2555457939819899, + "grad_norm": 2.105206251144409, + "learning_rate": 1e-06, + "loss": 1.1309, + "mean_token_accuracy": 0.6645684242248535, + "num_tokens": 59996378.0, + "step": 2327 + }, + { + "epoch": 0.25565561168460355, + "grad_norm": 2.143615484237671, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6968787908554077, + "num_tokens": 60023205.0, + "step": 2328 + }, + { + "epoch": 0.25576542938721725, + "grad_norm": 2.1782801151275635, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7265186309814453, + "num_tokens": 60047466.0, + "step": 2329 + }, + { + "epoch": 0.2558752470898309, + "grad_norm": 2.2333333492279053, + "learning_rate": 1e-06, + "loss": 1.0897, + "mean_token_accuracy": 0.6739892959594727, + "num_tokens": 60075304.0, + "step": 2330 + }, + { + "epoch": 0.25598506479244454, + "grad_norm": 2.305206537246704, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.709172785282135, + "num_tokens": 60099669.0, + "step": 2331 + }, + { + "epoch": 0.2560948824950582, + "grad_norm": 2.051302433013916, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6826152205467224, + "num_tokens": 60130142.0, + "step": 2332 + }, + { + "epoch": 0.2562047001976719, + "grad_norm": 2.380502939224243, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7195308208465576, + "num_tokens": 60152191.0, + "step": 2333 + }, + { + "epoch": 0.25631451790028553, + "grad_norm": 2.3686938285827637, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6888371109962463, + "num_tokens": 60175592.0, + "step": 2334 + }, + { + "epoch": 0.2564243356028992, + "grad_norm": 2.314145088195801, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.6810258626937866, + "num_tokens": 60201430.0, + "step": 2335 + }, + { + "epoch": 0.2565341533055128, + "grad_norm": 2.1580309867858887, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7050719261169434, + "num_tokens": 60230913.0, + "step": 2336 + }, + { + "epoch": 0.2566439710081265, + "grad_norm": 2.429830312728882, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7038530111312866, + "num_tokens": 60252795.0, + "step": 2337 + }, + { + "epoch": 0.25675378871074017, + "grad_norm": 2.151506185531616, + "learning_rate": 1e-06, + "loss": 1.0825, + "mean_token_accuracy": 0.680259108543396, + "num_tokens": 60282696.0, + "step": 2338 + }, + { + "epoch": 0.2568636064133538, + "grad_norm": 2.2460455894470215, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7078168392181396, + "num_tokens": 60308309.0, + "step": 2339 + }, + { + "epoch": 0.2569734241159675, + "grad_norm": 2.4815125465393066, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6885876655578613, + "num_tokens": 60331928.0, + "step": 2340 + }, + { + "epoch": 0.25708324181858117, + "grad_norm": 2.16119122505188, + "learning_rate": 1e-06, + "loss": 1.0952, + "mean_token_accuracy": 0.6721712946891785, + "num_tokens": 60360283.0, + "step": 2341 + }, + { + "epoch": 0.2571930595211948, + "grad_norm": 2.2947442531585693, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6900813579559326, + "num_tokens": 60384422.0, + "step": 2342 + }, + { + "epoch": 0.25730287722380846, + "grad_norm": 2.2144405841827393, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7119655609130859, + "num_tokens": 60408646.0, + "step": 2343 + }, + { + "epoch": 0.25741269492642216, + "grad_norm": 2.3139142990112305, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6827059984207153, + "num_tokens": 60432837.0, + "step": 2344 + }, + { + "epoch": 0.2575225126290358, + "grad_norm": 2.182640314102173, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.710576057434082, + "num_tokens": 60458819.0, + "step": 2345 + }, + { + "epoch": 0.25763233033164945, + "grad_norm": 2.3397462368011475, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6842654943466187, + "num_tokens": 60481856.0, + "step": 2346 + }, + { + "epoch": 0.2577421480342631, + "grad_norm": 2.581267833709717, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7112085819244385, + "num_tokens": 60501035.0, + "step": 2347 + }, + { + "epoch": 0.2578519657368768, + "grad_norm": 2.34092378616333, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.684576153755188, + "num_tokens": 60525232.0, + "step": 2348 + }, + { + "epoch": 0.25796178343949044, + "grad_norm": 2.081696033477783, + "learning_rate": 1e-06, + "loss": 1.0943, + "mean_token_accuracy": 0.6729553937911987, + "num_tokens": 60555027.0, + "step": 2349 + }, + { + "epoch": 0.2580716011421041, + "grad_norm": 2.314484119415283, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7000977993011475, + "num_tokens": 60578556.0, + "step": 2350 + }, + { + "epoch": 0.2581814188447178, + "grad_norm": 2.2315685749053955, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6919021010398865, + "num_tokens": 60604266.0, + "step": 2351 + }, + { + "epoch": 0.25829123654733144, + "grad_norm": 2.2257964611053467, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7030009627342224, + "num_tokens": 60631338.0, + "step": 2352 + }, + { + "epoch": 0.2584010542499451, + "grad_norm": 2.149346351623535, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.7058635950088501, + "num_tokens": 60661336.0, + "step": 2353 + }, + { + "epoch": 0.25851087195255873, + "grad_norm": 2.608092784881592, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7345874309539795, + "num_tokens": 60679807.0, + "step": 2354 + }, + { + "epoch": 0.25862068965517243, + "grad_norm": 2.204697370529175, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.7000237703323364, + "num_tokens": 60706368.0, + "step": 2355 + }, + { + "epoch": 0.2587305073577861, + "grad_norm": 2.381192207336426, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6954978704452515, + "num_tokens": 60727404.0, + "step": 2356 + }, + { + "epoch": 0.2588403250603997, + "grad_norm": 2.5847725868225098, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7082551121711731, + "num_tokens": 60747979.0, + "step": 2357 + }, + { + "epoch": 0.2589501427630134, + "grad_norm": 2.1475718021392822, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6834545135498047, + "num_tokens": 60774822.0, + "step": 2358 + }, + { + "epoch": 0.25905996046562707, + "grad_norm": 2.2274746894836426, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6972710490226746, + "num_tokens": 60801362.0, + "step": 2359 + }, + { + "epoch": 0.2591697781682407, + "grad_norm": 1.9946887493133545, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6886155605316162, + "num_tokens": 60831058.0, + "step": 2360 + }, + { + "epoch": 0.25927959587085436, + "grad_norm": 2.596308469772339, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7153612971305847, + "num_tokens": 60849313.0, + "step": 2361 + }, + { + "epoch": 0.25938941357346806, + "grad_norm": 2.1570992469787598, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7142976522445679, + "num_tokens": 60874030.0, + "step": 2362 + }, + { + "epoch": 0.2594992312760817, + "grad_norm": 2.3212599754333496, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6978163719177246, + "num_tokens": 60898319.0, + "step": 2363 + }, + { + "epoch": 0.25960904897869536, + "grad_norm": 2.2489140033721924, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6939928531646729, + "num_tokens": 60924059.0, + "step": 2364 + }, + { + "epoch": 0.259718866681309, + "grad_norm": 2.270472526550293, + "learning_rate": 1e-06, + "loss": 1.0814, + "mean_token_accuracy": 0.6728757619857788, + "num_tokens": 60952955.0, + "step": 2365 + }, + { + "epoch": 0.2598286843839227, + "grad_norm": 2.084599733352661, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6852295398712158, + "num_tokens": 60981415.0, + "step": 2366 + }, + { + "epoch": 0.25993850208653635, + "grad_norm": 2.3155765533447266, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7020155191421509, + "num_tokens": 61005323.0, + "step": 2367 + }, + { + "epoch": 0.26004831978915, + "grad_norm": 2.3045754432678223, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6827334761619568, + "num_tokens": 61030364.0, + "step": 2368 + }, + { + "epoch": 0.2601581374917637, + "grad_norm": 2.193743944168091, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6937443614006042, + "num_tokens": 61055718.0, + "step": 2369 + }, + { + "epoch": 0.26026795519437734, + "grad_norm": 2.383273124694824, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6915386915206909, + "num_tokens": 61079371.0, + "step": 2370 + }, + { + "epoch": 0.260377772896991, + "grad_norm": 2.5375967025756836, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6958944797515869, + "num_tokens": 61100903.0, + "step": 2371 + }, + { + "epoch": 0.26048759059960463, + "grad_norm": 1.9880121946334839, + "learning_rate": 1e-06, + "loss": 1.1083, + "mean_token_accuracy": 0.6763067841529846, + "num_tokens": 61135706.0, + "step": 2372 + }, + { + "epoch": 0.26059740830221834, + "grad_norm": 2.1533701419830322, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7216750383377075, + "num_tokens": 61158959.0, + "step": 2373 + }, + { + "epoch": 0.260707226004832, + "grad_norm": 2.3291032314300537, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6895884275436401, + "num_tokens": 61183096.0, + "step": 2374 + }, + { + "epoch": 0.26081704370744563, + "grad_norm": 2.3385677337646484, + "learning_rate": 1e-06, + "loss": 1.1128, + "mean_token_accuracy": 0.6674073338508606, + "num_tokens": 61206980.0, + "step": 2375 + }, + { + "epoch": 0.2609268614100593, + "grad_norm": 2.444565534591675, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7133809924125671, + "num_tokens": 61227817.0, + "step": 2376 + }, + { + "epoch": 0.261036679112673, + "grad_norm": 2.250948429107666, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7000384330749512, + "num_tokens": 61253874.0, + "step": 2377 + }, + { + "epoch": 0.2611464968152866, + "grad_norm": 2.454929828643799, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7010238170623779, + "num_tokens": 61275751.0, + "step": 2378 + }, + { + "epoch": 0.26125631451790027, + "grad_norm": 2.25224232673645, + "learning_rate": 1e-06, + "loss": 1.0662, + "mean_token_accuracy": 0.6816251277923584, + "num_tokens": 61302214.0, + "step": 2379 + }, + { + "epoch": 0.26136613222051397, + "grad_norm": 2.2174012660980225, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6875658631324768, + "num_tokens": 61330072.0, + "step": 2380 + }, + { + "epoch": 0.2614759499231276, + "grad_norm": 2.8299312591552734, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7068087458610535, + "num_tokens": 61350810.0, + "step": 2381 + }, + { + "epoch": 0.26158576762574126, + "grad_norm": 2.281403064727783, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.7031064033508301, + "num_tokens": 61378210.0, + "step": 2382 + }, + { + "epoch": 0.2616955853283549, + "grad_norm": 2.5681135654449463, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6961883306503296, + "num_tokens": 61398863.0, + "step": 2383 + }, + { + "epoch": 0.2618054030309686, + "grad_norm": 1.9679926633834839, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7013302445411682, + "num_tokens": 61432029.0, + "step": 2384 + }, + { + "epoch": 0.26191522073358225, + "grad_norm": 2.112786054611206, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6861597299575806, + "num_tokens": 61459776.0, + "step": 2385 + }, + { + "epoch": 0.2620250384361959, + "grad_norm": 2.2472100257873535, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7093523740768433, + "num_tokens": 61485835.0, + "step": 2386 + }, + { + "epoch": 0.2621348561388096, + "grad_norm": 2.4124248027801514, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7033455967903137, + "num_tokens": 61508019.0, + "step": 2387 + }, + { + "epoch": 0.26224467384142325, + "grad_norm": 2.278527021408081, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.687737762928009, + "num_tokens": 61533468.0, + "step": 2388 + }, + { + "epoch": 0.2623544915440369, + "grad_norm": 2.058152914047241, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6905518770217896, + "num_tokens": 61565050.0, + "step": 2389 + }, + { + "epoch": 0.26246430924665054, + "grad_norm": 2.5463316440582275, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6937084794044495, + "num_tokens": 61587451.0, + "step": 2390 + }, + { + "epoch": 0.26257412694926424, + "grad_norm": 2.254774332046509, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6885201930999756, + "num_tokens": 61611977.0, + "step": 2391 + }, + { + "epoch": 0.2626839446518779, + "grad_norm": 2.132108688354492, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6855847239494324, + "num_tokens": 61640958.0, + "step": 2392 + }, + { + "epoch": 0.26279376235449153, + "grad_norm": 2.313469648361206, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6876204013824463, + "num_tokens": 61665024.0, + "step": 2393 + }, + { + "epoch": 0.2629035800571052, + "grad_norm": 2.2049312591552734, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6999752521514893, + "num_tokens": 61691851.0, + "step": 2394 + }, + { + "epoch": 0.2630133977597189, + "grad_norm": 2.2997848987579346, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7126950025558472, + "num_tokens": 61715019.0, + "step": 2395 + }, + { + "epoch": 0.2631232154623325, + "grad_norm": 1.887342929840088, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6927878856658936, + "num_tokens": 61747915.0, + "step": 2396 + }, + { + "epoch": 0.2632330331649462, + "grad_norm": 2.1397902965545654, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6884219646453857, + "num_tokens": 61776134.0, + "step": 2397 + }, + { + "epoch": 0.2633428508675599, + "grad_norm": 2.1775527000427246, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.7020241618156433, + "num_tokens": 61802629.0, + "step": 2398 + }, + { + "epoch": 0.2634526685701735, + "grad_norm": 2.217792510986328, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6920677423477173, + "num_tokens": 61830458.0, + "step": 2399 + }, + { + "epoch": 0.26356248627278717, + "grad_norm": 2.040715217590332, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.6727391481399536, + "num_tokens": 61862460.0, + "step": 2400 + }, + { + "epoch": 0.2636723039754008, + "grad_norm": 2.0209991931915283, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6763803362846375, + "num_tokens": 61892769.0, + "step": 2401 + }, + { + "epoch": 0.2637821216780145, + "grad_norm": 2.1396279335021973, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6927618980407715, + "num_tokens": 61920203.0, + "step": 2402 + }, + { + "epoch": 0.26389193938062816, + "grad_norm": 2.086731433868408, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7161372900009155, + "num_tokens": 61947810.0, + "step": 2403 + }, + { + "epoch": 0.2640017570832418, + "grad_norm": 2.4046425819396973, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7056118249893188, + "num_tokens": 61970027.0, + "step": 2404 + }, + { + "epoch": 0.2641115747858555, + "grad_norm": 2.3287127017974854, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7084925174713135, + "num_tokens": 61994808.0, + "step": 2405 + }, + { + "epoch": 0.26422139248846915, + "grad_norm": 2.3097007274627686, + "learning_rate": 1e-06, + "loss": 1.059, + "mean_token_accuracy": 0.6841346025466919, + "num_tokens": 62020913.0, + "step": 2406 + }, + { + "epoch": 0.2643312101910828, + "grad_norm": 2.597778558731079, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7002298831939697, + "num_tokens": 62043514.0, + "step": 2407 + }, + { + "epoch": 0.26444102789369645, + "grad_norm": 2.2431018352508545, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7232473492622375, + "num_tokens": 62068028.0, + "step": 2408 + }, + { + "epoch": 0.26455084559631015, + "grad_norm": 2.142399311065674, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6893038153648376, + "num_tokens": 62095409.0, + "step": 2409 + }, + { + "epoch": 0.2646606632989238, + "grad_norm": 2.0673816204071045, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7119133472442627, + "num_tokens": 62124227.0, + "step": 2410 + }, + { + "epoch": 0.26477048100153744, + "grad_norm": 2.3491859436035156, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.692610502243042, + "num_tokens": 62149303.0, + "step": 2411 + }, + { + "epoch": 0.2648802987041511, + "grad_norm": 2.752455949783325, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6791214942932129, + "num_tokens": 62169978.0, + "step": 2412 + }, + { + "epoch": 0.2649901164067648, + "grad_norm": 2.3292200565338135, + "learning_rate": 1e-06, + "loss": 1.0936, + "mean_token_accuracy": 0.6713675260543823, + "num_tokens": 62197634.0, + "step": 2413 + }, + { + "epoch": 0.26509993410937843, + "grad_norm": 2.4059855937957764, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.695127010345459, + "num_tokens": 62220896.0, + "step": 2414 + }, + { + "epoch": 0.2652097518119921, + "grad_norm": 2.6587133407592773, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6989980936050415, + "num_tokens": 62239613.0, + "step": 2415 + }, + { + "epoch": 0.2653195695146058, + "grad_norm": 2.3793575763702393, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7072275280952454, + "num_tokens": 62263789.0, + "step": 2416 + }, + { + "epoch": 0.2654293872172194, + "grad_norm": 2.1195077896118164, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.675748348236084, + "num_tokens": 62293169.0, + "step": 2417 + }, + { + "epoch": 0.26553920491983307, + "grad_norm": 1.9020183086395264, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6919571161270142, + "num_tokens": 62328061.0, + "step": 2418 + }, + { + "epoch": 0.2656490226224467, + "grad_norm": 2.181330442428589, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6922547817230225, + "num_tokens": 62355589.0, + "step": 2419 + }, + { + "epoch": 0.2657588403250604, + "grad_norm": 2.165111780166626, + "learning_rate": 1e-06, + "loss": 1.1649, + "mean_token_accuracy": 0.6658527255058289, + "num_tokens": 62384442.0, + "step": 2420 + }, + { + "epoch": 0.26586865802767407, + "grad_norm": 2.132317543029785, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6981627941131592, + "num_tokens": 62410673.0, + "step": 2421 + }, + { + "epoch": 0.2659784757302877, + "grad_norm": 2.2035131454467773, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.694855809211731, + "num_tokens": 62440840.0, + "step": 2422 + }, + { + "epoch": 0.26608829343290136, + "grad_norm": 2.5527470111846924, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.714651882648468, + "num_tokens": 62459300.0, + "step": 2423 + }, + { + "epoch": 0.26619811113551506, + "grad_norm": 2.2066993713378906, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6848543882369995, + "num_tokens": 62485697.0, + "step": 2424 + }, + { + "epoch": 0.2663079288381287, + "grad_norm": 2.5151779651641846, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7123947143554688, + "num_tokens": 62507730.0, + "step": 2425 + }, + { + "epoch": 0.26641774654074235, + "grad_norm": 2.098019599914551, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6747089624404907, + "num_tokens": 62539762.0, + "step": 2426 + }, + { + "epoch": 0.26652756424335605, + "grad_norm": 2.2856757640838623, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.70691978931427, + "num_tokens": 62563723.0, + "step": 2427 + }, + { + "epoch": 0.2666373819459697, + "grad_norm": 1.946834921836853, + "learning_rate": 1e-06, + "loss": 1.1127, + "mean_token_accuracy": 0.6739058494567871, + "num_tokens": 62599150.0, + "step": 2428 + }, + { + "epoch": 0.26674719964858334, + "grad_norm": 2.334836721420288, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.7002507448196411, + "num_tokens": 62622381.0, + "step": 2429 + }, + { + "epoch": 0.266857017351197, + "grad_norm": 2.193298101425171, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6923487782478333, + "num_tokens": 62647150.0, + "step": 2430 + }, + { + "epoch": 0.2669668350538107, + "grad_norm": 2.3364572525024414, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6952354311943054, + "num_tokens": 62671900.0, + "step": 2431 + }, + { + "epoch": 0.26707665275642434, + "grad_norm": 2.123255968093872, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7058600187301636, + "num_tokens": 62697209.0, + "step": 2432 + }, + { + "epoch": 0.267186470459038, + "grad_norm": 2.3566884994506836, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7312735319137573, + "num_tokens": 62717647.0, + "step": 2433 + }, + { + "epoch": 0.2672962881616517, + "grad_norm": 2.008117198944092, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6857869625091553, + "num_tokens": 62751210.0, + "step": 2434 + }, + { + "epoch": 0.26740610586426533, + "grad_norm": 2.242990016937256, + "learning_rate": 1e-06, + "loss": 1.113, + "mean_token_accuracy": 0.6701549291610718, + "num_tokens": 62779601.0, + "step": 2435 + }, + { + "epoch": 0.267515923566879, + "grad_norm": 2.339524030685425, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6819548010826111, + "num_tokens": 62805909.0, + "step": 2436 + }, + { + "epoch": 0.2676257412694926, + "grad_norm": 2.301335573196411, + "learning_rate": 1e-06, + "loss": 1.0785, + "mean_token_accuracy": 0.677040696144104, + "num_tokens": 62830339.0, + "step": 2437 + }, + { + "epoch": 0.2677355589721063, + "grad_norm": 2.082033395767212, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6847957372665405, + "num_tokens": 62859337.0, + "step": 2438 + }, + { + "epoch": 0.26784537667471997, + "grad_norm": 2.368384838104248, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7157360315322876, + "num_tokens": 62883142.0, + "step": 2439 + }, + { + "epoch": 0.2679551943773336, + "grad_norm": 2.4943687915802, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7056269645690918, + "num_tokens": 62904668.0, + "step": 2440 + }, + { + "epoch": 0.26806501207994726, + "grad_norm": 2.2709341049194336, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6875804662704468, + "num_tokens": 62930072.0, + "step": 2441 + }, + { + "epoch": 0.26817482978256096, + "grad_norm": 2.0578129291534424, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7110954523086548, + "num_tokens": 62958587.0, + "step": 2442 + }, + { + "epoch": 0.2682846474851746, + "grad_norm": 2.178656578063965, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6999646425247192, + "num_tokens": 62984573.0, + "step": 2443 + }, + { + "epoch": 0.26839446518778826, + "grad_norm": 2.195566177368164, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7145522236824036, + "num_tokens": 63011131.0, + "step": 2444 + }, + { + "epoch": 0.26850428289040196, + "grad_norm": 2.123659133911133, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6876838207244873, + "num_tokens": 63039796.0, + "step": 2445 + }, + { + "epoch": 0.2686141005930156, + "grad_norm": 2.6430001258850098, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7255294322967529, + "num_tokens": 63059168.0, + "step": 2446 + }, + { + "epoch": 0.26872391829562925, + "grad_norm": 2.0155301094055176, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6950040459632874, + "num_tokens": 63089517.0, + "step": 2447 + }, + { + "epoch": 0.2688337359982429, + "grad_norm": 2.1148886680603027, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6963800191879272, + "num_tokens": 63116264.0, + "step": 2448 + }, + { + "epoch": 0.2689435537008566, + "grad_norm": 2.40700626373291, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7099527716636658, + "num_tokens": 63138681.0, + "step": 2449 + }, + { + "epoch": 0.26905337140347024, + "grad_norm": 2.2915890216827393, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6818913221359253, + "num_tokens": 63164853.0, + "step": 2450 + }, + { + "epoch": 0.2691631891060839, + "grad_norm": 2.274502754211426, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.704440176486969, + "num_tokens": 63190783.0, + "step": 2451 + }, + { + "epoch": 0.26927300680869753, + "grad_norm": 2.151503562927246, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.700786828994751, + "num_tokens": 63218137.0, + "step": 2452 + }, + { + "epoch": 0.26938282451131124, + "grad_norm": 2.205049753189087, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.7017472982406616, + "num_tokens": 63244978.0, + "step": 2453 + }, + { + "epoch": 0.2694926422139249, + "grad_norm": 2.1447324752807617, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7082600593566895, + "num_tokens": 63270339.0, + "step": 2454 + }, + { + "epoch": 0.26960245991653853, + "grad_norm": 2.535259962081909, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7078946828842163, + "num_tokens": 63290842.0, + "step": 2455 + }, + { + "epoch": 0.26971227761915223, + "grad_norm": 2.0170395374298096, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6984652876853943, + "num_tokens": 63321106.0, + "step": 2456 + }, + { + "epoch": 0.2698220953217659, + "grad_norm": 2.3507213592529297, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6993135213851929, + "num_tokens": 63344410.0, + "step": 2457 + }, + { + "epoch": 0.2699319130243795, + "grad_norm": 2.442560911178589, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7153676152229309, + "num_tokens": 63366025.0, + "step": 2458 + }, + { + "epoch": 0.27004173072699317, + "grad_norm": 2.5248215198516846, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7009409666061401, + "num_tokens": 63385447.0, + "step": 2459 + }, + { + "epoch": 0.27015154842960687, + "grad_norm": 1.9149088859558105, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6927985548973083, + "num_tokens": 63416607.0, + "step": 2460 + }, + { + "epoch": 0.2702613661322205, + "grad_norm": 2.2136900424957275, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6878419518470764, + "num_tokens": 63443573.0, + "step": 2461 + }, + { + "epoch": 0.27037118383483416, + "grad_norm": 2.2414908409118652, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7061024904251099, + "num_tokens": 63468821.0, + "step": 2462 + }, + { + "epoch": 0.27048100153744786, + "grad_norm": 2.1911354064941406, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6921229362487793, + "num_tokens": 63494149.0, + "step": 2463 + }, + { + "epoch": 0.2705908192400615, + "grad_norm": 2.2160837650299072, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6940187215805054, + "num_tokens": 63520396.0, + "step": 2464 + }, + { + "epoch": 0.27070063694267515, + "grad_norm": 2.436204671859741, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6861563324928284, + "num_tokens": 63541365.0, + "step": 2465 + }, + { + "epoch": 0.2708104546452888, + "grad_norm": 2.5494191646575928, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6986499428749084, + "num_tokens": 63561415.0, + "step": 2466 + }, + { + "epoch": 0.2709202723479025, + "grad_norm": 2.2404890060424805, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6900635361671448, + "num_tokens": 63586710.0, + "step": 2467 + }, + { + "epoch": 0.27103009005051615, + "grad_norm": 2.178927421569824, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6961262226104736, + "num_tokens": 63614719.0, + "step": 2468 + }, + { + "epoch": 0.2711399077531298, + "grad_norm": 2.08481764793396, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7128319144248962, + "num_tokens": 63641627.0, + "step": 2469 + }, + { + "epoch": 0.27124972545574344, + "grad_norm": 2.112224578857422, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6884497404098511, + "num_tokens": 63671223.0, + "step": 2470 + }, + { + "epoch": 0.27135954315835714, + "grad_norm": 2.4102416038513184, + "learning_rate": 1e-06, + "loss": 1.1009, + "mean_token_accuracy": 0.6635973453521729, + "num_tokens": 63694965.0, + "step": 2471 + }, + { + "epoch": 0.2714693608609708, + "grad_norm": 2.12813401222229, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7060344219207764, + "num_tokens": 63722554.0, + "step": 2472 + }, + { + "epoch": 0.27157917856358443, + "grad_norm": 2.0266144275665283, + "learning_rate": 1e-06, + "loss": 1.1047, + "mean_token_accuracy": 0.6655116677284241, + "num_tokens": 63755682.0, + "step": 2473 + }, + { + "epoch": 0.27168899626619814, + "grad_norm": 2.432957887649536, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6951382160186768, + "num_tokens": 63779615.0, + "step": 2474 + }, + { + "epoch": 0.2717988139688118, + "grad_norm": 2.122173309326172, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6957736015319824, + "num_tokens": 63807524.0, + "step": 2475 + }, + { + "epoch": 0.2719086316714254, + "grad_norm": 2.399336338043213, + "learning_rate": 1e-06, + "loss": 1.0612, + "mean_token_accuracy": 0.6856611967086792, + "num_tokens": 63831343.0, + "step": 2476 + }, + { + "epoch": 0.2720184493740391, + "grad_norm": 2.214840888977051, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7032305598258972, + "num_tokens": 63858821.0, + "step": 2477 + }, + { + "epoch": 0.2721282670766528, + "grad_norm": 2.2263381481170654, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.7039663791656494, + "num_tokens": 63882253.0, + "step": 2478 + }, + { + "epoch": 0.2722380847792664, + "grad_norm": 2.156364679336548, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6965645551681519, + "num_tokens": 63910339.0, + "step": 2479 + }, + { + "epoch": 0.27234790248188007, + "grad_norm": 2.336836099624634, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6981568336486816, + "num_tokens": 63934875.0, + "step": 2480 + }, + { + "epoch": 0.27245772018449377, + "grad_norm": 2.190034866333008, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6986656188964844, + "num_tokens": 63962987.0, + "step": 2481 + }, + { + "epoch": 0.2725675378871074, + "grad_norm": 1.9502544403076172, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6965420842170715, + "num_tokens": 63996130.0, + "step": 2482 + }, + { + "epoch": 0.27267735558972106, + "grad_norm": 2.469482183456421, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6941948533058167, + "num_tokens": 64018057.0, + "step": 2483 + }, + { + "epoch": 0.2727871732923347, + "grad_norm": 1.9973474740982056, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6976919174194336, + "num_tokens": 64048265.0, + "step": 2484 + }, + { + "epoch": 0.2728969909949484, + "grad_norm": 2.4588701725006104, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7128528356552124, + "num_tokens": 64068898.0, + "step": 2485 + }, + { + "epoch": 0.27300680869756205, + "grad_norm": 2.233975887298584, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6843540668487549, + "num_tokens": 64093939.0, + "step": 2486 + }, + { + "epoch": 0.2731166264001757, + "grad_norm": 1.934016227722168, + "learning_rate": 1e-06, + "loss": 1.0816, + "mean_token_accuracy": 0.6777840256690979, + "num_tokens": 64128715.0, + "step": 2487 + }, + { + "epoch": 0.27322644410278935, + "grad_norm": 2.244601011276245, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6870938539505005, + "num_tokens": 64156263.0, + "step": 2488 + }, + { + "epoch": 0.27333626180540305, + "grad_norm": 2.0922842025756836, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7033801078796387, + "num_tokens": 64184630.0, + "step": 2489 + }, + { + "epoch": 0.2734460795080167, + "grad_norm": 2.128615140914917, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6816380620002747, + "num_tokens": 64213343.0, + "step": 2490 + }, + { + "epoch": 0.27355589721063034, + "grad_norm": 2.6193697452545166, + "learning_rate": 1e-06, + "loss": 1.086, + "mean_token_accuracy": 0.6692317128181458, + "num_tokens": 64233366.0, + "step": 2491 + }, + { + "epoch": 0.27366571491324404, + "grad_norm": 2.4326133728027344, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7316970825195312, + "num_tokens": 64255687.0, + "step": 2492 + }, + { + "epoch": 0.2737755326158577, + "grad_norm": 2.658177137374878, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.6981869339942932, + "num_tokens": 64276389.0, + "step": 2493 + }, + { + "epoch": 0.27388535031847133, + "grad_norm": 2.277977466583252, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7210761308670044, + "num_tokens": 64300730.0, + "step": 2494 + }, + { + "epoch": 0.273995168021085, + "grad_norm": 2.6451408863067627, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7118227481842041, + "num_tokens": 64320382.0, + "step": 2495 + }, + { + "epoch": 0.2741049857236987, + "grad_norm": 2.1717982292175293, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6875706911087036, + "num_tokens": 64350133.0, + "step": 2496 + }, + { + "epoch": 0.2742148034263123, + "grad_norm": 2.2732741832733154, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7133586406707764, + "num_tokens": 64375176.0, + "step": 2497 + }, + { + "epoch": 0.27432462112892597, + "grad_norm": 2.085378646850586, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.7019253969192505, + "num_tokens": 64406267.0, + "step": 2498 + }, + { + "epoch": 0.2744344388315396, + "grad_norm": 2.2286641597747803, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.68742835521698, + "num_tokens": 64434463.0, + "step": 2499 + }, + { + "epoch": 0.2745442565341533, + "grad_norm": 2.2902958393096924, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7053949236869812, + "num_tokens": 64459114.0, + "step": 2500 + }, + { + "epoch": 0.27465407423676697, + "grad_norm": 2.3403050899505615, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7055253386497498, + "num_tokens": 64484811.0, + "step": 2501 + }, + { + "epoch": 0.2747638919393806, + "grad_norm": 2.487638235092163, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7087430953979492, + "num_tokens": 64504748.0, + "step": 2502 + }, + { + "epoch": 0.2748737096419943, + "grad_norm": 2.452723979949951, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6825305223464966, + "num_tokens": 64528167.0, + "step": 2503 + }, + { + "epoch": 0.27498352734460796, + "grad_norm": 2.1707661151885986, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6929940581321716, + "num_tokens": 64555549.0, + "step": 2504 + }, + { + "epoch": 0.2750933450472216, + "grad_norm": 2.1717705726623535, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7030565142631531, + "num_tokens": 64581332.0, + "step": 2505 + }, + { + "epoch": 0.27520316274983525, + "grad_norm": 2.1910369396209717, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.702184796333313, + "num_tokens": 64609496.0, + "step": 2506 + }, + { + "epoch": 0.27531298045244895, + "grad_norm": 2.368706703186035, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7026753425598145, + "num_tokens": 64633857.0, + "step": 2507 + }, + { + "epoch": 0.2754227981550626, + "grad_norm": 2.5529544353485107, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.69484943151474, + "num_tokens": 64654908.0, + "step": 2508 + }, + { + "epoch": 0.27553261585767624, + "grad_norm": 2.295565128326416, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6980401277542114, + "num_tokens": 64679705.0, + "step": 2509 + }, + { + "epoch": 0.27564243356028995, + "grad_norm": 2.3961641788482666, + "learning_rate": 1e-06, + "loss": 1.085, + "mean_token_accuracy": 0.6725595593452454, + "num_tokens": 64701981.0, + "step": 2510 + }, + { + "epoch": 0.2757522512629036, + "grad_norm": 1.9392738342285156, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6814804077148438, + "num_tokens": 64733667.0, + "step": 2511 + }, + { + "epoch": 0.27586206896551724, + "grad_norm": 2.296274185180664, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6978039145469666, + "num_tokens": 64757965.0, + "step": 2512 + }, + { + "epoch": 0.2759718866681309, + "grad_norm": 2.275235652923584, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6981817483901978, + "num_tokens": 64783346.0, + "step": 2513 + }, + { + "epoch": 0.2760817043707446, + "grad_norm": 2.3967626094818115, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6974854469299316, + "num_tokens": 64808349.0, + "step": 2514 + }, + { + "epoch": 0.27619152207335823, + "grad_norm": 2.6255152225494385, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.707582950592041, + "num_tokens": 64827894.0, + "step": 2515 + }, + { + "epoch": 0.2763013397759719, + "grad_norm": 2.2213006019592285, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6927250623703003, + "num_tokens": 64854836.0, + "step": 2516 + }, + { + "epoch": 0.2764111574785855, + "grad_norm": 2.5080807209014893, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6888161897659302, + "num_tokens": 64876862.0, + "step": 2517 + }, + { + "epoch": 0.2765209751811992, + "grad_norm": 1.959905982017517, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6998422145843506, + "num_tokens": 64911057.0, + "step": 2518 + }, + { + "epoch": 0.27663079288381287, + "grad_norm": 2.3013246059417725, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7003630995750427, + "num_tokens": 64936443.0, + "step": 2519 + }, + { + "epoch": 0.2767406105864265, + "grad_norm": 2.219412088394165, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6903688907623291, + "num_tokens": 64964673.0, + "step": 2520 + }, + { + "epoch": 0.2768504282890402, + "grad_norm": 2.352195978164673, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6947768926620483, + "num_tokens": 64988825.0, + "step": 2521 + }, + { + "epoch": 0.27696024599165386, + "grad_norm": 2.3331193923950195, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7153981924057007, + "num_tokens": 65011625.0, + "step": 2522 + }, + { + "epoch": 0.2770700636942675, + "grad_norm": 2.1482772827148438, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.695518970489502, + "num_tokens": 65039960.0, + "step": 2523 + }, + { + "epoch": 0.27717988139688116, + "grad_norm": 2.474256753921509, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7034685611724854, + "num_tokens": 65060985.0, + "step": 2524 + }, + { + "epoch": 0.27728969909949486, + "grad_norm": 2.282569646835327, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7082047462463379, + "num_tokens": 65084878.0, + "step": 2525 + }, + { + "epoch": 0.2773995168021085, + "grad_norm": 2.4878461360931396, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6912424564361572, + "num_tokens": 65105699.0, + "step": 2526 + }, + { + "epoch": 0.27750933450472215, + "grad_norm": 2.037295341491699, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6981704831123352, + "num_tokens": 65137599.0, + "step": 2527 + }, + { + "epoch": 0.2776191522073358, + "grad_norm": 2.0249552726745605, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6960660815238953, + "num_tokens": 65167967.0, + "step": 2528 + }, + { + "epoch": 0.2777289699099495, + "grad_norm": 2.154282808303833, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7044427990913391, + "num_tokens": 65196322.0, + "step": 2529 + }, + { + "epoch": 0.27783878761256314, + "grad_norm": 2.1613526344299316, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7123574018478394, + "num_tokens": 65221667.0, + "step": 2530 + }, + { + "epoch": 0.2779486053151768, + "grad_norm": 2.568434238433838, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7123340368270874, + "num_tokens": 65240699.0, + "step": 2531 + }, + { + "epoch": 0.2780584230177905, + "grad_norm": 2.1955814361572266, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7039154171943665, + "num_tokens": 65266340.0, + "step": 2532 + }, + { + "epoch": 0.27816824072040414, + "grad_norm": 2.3097708225250244, + "learning_rate": 1e-06, + "loss": 1.1154, + "mean_token_accuracy": 0.663765013217926, + "num_tokens": 65292524.0, + "step": 2533 + }, + { + "epoch": 0.2782780584230178, + "grad_norm": 2.595731496810913, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7053881883621216, + "num_tokens": 65315025.0, + "step": 2534 + }, + { + "epoch": 0.27838787612563143, + "grad_norm": 2.0775809288024902, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7256376147270203, + "num_tokens": 65342296.0, + "step": 2535 + }, + { + "epoch": 0.27849769382824513, + "grad_norm": 2.1798930168151855, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7203707695007324, + "num_tokens": 65367060.0, + "step": 2536 + }, + { + "epoch": 0.2786075115308588, + "grad_norm": 2.2412819862365723, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7062882781028748, + "num_tokens": 65390375.0, + "step": 2537 + }, + { + "epoch": 0.2787173292334724, + "grad_norm": 2.3247742652893066, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.7029619216918945, + "num_tokens": 65414900.0, + "step": 2538 + }, + { + "epoch": 0.2788271469360861, + "grad_norm": 2.0282421112060547, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6883099675178528, + "num_tokens": 65446822.0, + "step": 2539 + }, + { + "epoch": 0.27893696463869977, + "grad_norm": 2.4277987480163574, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6939659118652344, + "num_tokens": 65469927.0, + "step": 2540 + }, + { + "epoch": 0.2790467823413134, + "grad_norm": 2.046320915222168, + "learning_rate": 1e-06, + "loss": 1.0744, + "mean_token_accuracy": 0.684501051902771, + "num_tokens": 65500416.0, + "step": 2541 + }, + { + "epoch": 0.27915660004392706, + "grad_norm": 2.3503503799438477, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6824218034744263, + "num_tokens": 65524308.0, + "step": 2542 + }, + { + "epoch": 0.27926641774654076, + "grad_norm": 2.097973108291626, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7092021703720093, + "num_tokens": 65551703.0, + "step": 2543 + }, + { + "epoch": 0.2793762354491544, + "grad_norm": 2.4162299633026123, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6901425123214722, + "num_tokens": 65574999.0, + "step": 2544 + }, + { + "epoch": 0.27948605315176805, + "grad_norm": 2.0690865516662598, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6974827647209167, + "num_tokens": 65603752.0, + "step": 2545 + }, + { + "epoch": 0.2795958708543817, + "grad_norm": 2.4258835315704346, + "learning_rate": 1e-06, + "loss": 1.066, + "mean_token_accuracy": 0.6814447641372681, + "num_tokens": 65625826.0, + "step": 2546 + }, + { + "epoch": 0.2797056885569954, + "grad_norm": 2.3509583473205566, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6858166456222534, + "num_tokens": 65649265.0, + "step": 2547 + }, + { + "epoch": 0.27981550625960905, + "grad_norm": 2.0554964542388916, + "learning_rate": 1e-06, + "loss": 1.1035, + "mean_token_accuracy": 0.6698739528656006, + "num_tokens": 65681384.0, + "step": 2548 + }, + { + "epoch": 0.2799253239622227, + "grad_norm": 2.2728898525238037, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7077915072441101, + "num_tokens": 65707321.0, + "step": 2549 + }, + { + "epoch": 0.2800351416648364, + "grad_norm": 2.307542085647583, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7002393007278442, + "num_tokens": 65732859.0, + "step": 2550 + }, + { + "epoch": 0.28014495936745004, + "grad_norm": 2.5723929405212402, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6881735324859619, + "num_tokens": 65753691.0, + "step": 2551 + }, + { + "epoch": 0.2802547770700637, + "grad_norm": 2.3275322914123535, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6870082020759583, + "num_tokens": 65777808.0, + "step": 2552 + }, + { + "epoch": 0.28036459477267733, + "grad_norm": 2.1308958530426025, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.6723392009735107, + "num_tokens": 65806787.0, + "step": 2553 + }, + { + "epoch": 0.28047441247529104, + "grad_norm": 2.0366828441619873, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6862590909004211, + "num_tokens": 65837649.0, + "step": 2554 + }, + { + "epoch": 0.2805842301779047, + "grad_norm": 2.3476157188415527, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7082025408744812, + "num_tokens": 65860646.0, + "step": 2555 + }, + { + "epoch": 0.2806940478805183, + "grad_norm": 2.4230010509490967, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.690748929977417, + "num_tokens": 65883155.0, + "step": 2556 + }, + { + "epoch": 0.28080386558313203, + "grad_norm": 2.14322829246521, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7017859816551208, + "num_tokens": 65911667.0, + "step": 2557 + }, + { + "epoch": 0.2809136832857457, + "grad_norm": 2.482548952102661, + "learning_rate": 1e-06, + "loss": 1.08, + "mean_token_accuracy": 0.6737629175186157, + "num_tokens": 65934586.0, + "step": 2558 + }, + { + "epoch": 0.2810235009883593, + "grad_norm": 2.171205997467041, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7138870358467102, + "num_tokens": 65959052.0, + "step": 2559 + }, + { + "epoch": 0.28113331869097297, + "grad_norm": 2.4319145679473877, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7080183625221252, + "num_tokens": 65981333.0, + "step": 2560 + }, + { + "epoch": 0.28124313639358667, + "grad_norm": 2.2036750316619873, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7055867910385132, + "num_tokens": 66006590.0, + "step": 2561 + }, + { + "epoch": 0.2813529540962003, + "grad_norm": 2.1549768447875977, + "learning_rate": 1e-06, + "loss": 1.0662, + "mean_token_accuracy": 0.6829162836074829, + "num_tokens": 66033695.0, + "step": 2562 + }, + { + "epoch": 0.28146277179881396, + "grad_norm": 2.6836907863616943, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7103482484817505, + "num_tokens": 66052842.0, + "step": 2563 + }, + { + "epoch": 0.2815725895014276, + "grad_norm": 2.1056406497955322, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6804526448249817, + "num_tokens": 66083272.0, + "step": 2564 + }, + { + "epoch": 0.2816824072040413, + "grad_norm": 2.3158178329467773, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7097023129463196, + "num_tokens": 66106479.0, + "step": 2565 + }, + { + "epoch": 0.28179222490665495, + "grad_norm": 2.449061155319214, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.7069598436355591, + "num_tokens": 66127569.0, + "step": 2566 + }, + { + "epoch": 0.2819020426092686, + "grad_norm": 2.3645124435424805, + "learning_rate": 1e-06, + "loss": 1.098, + "mean_token_accuracy": 0.6807683110237122, + "num_tokens": 66151263.0, + "step": 2567 + }, + { + "epoch": 0.2820118603118823, + "grad_norm": 1.9373810291290283, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6890888214111328, + "num_tokens": 66181396.0, + "step": 2568 + }, + { + "epoch": 0.28212167801449595, + "grad_norm": 2.4392194747924805, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7154510617256165, + "num_tokens": 66202315.0, + "step": 2569 + }, + { + "epoch": 0.2822314957171096, + "grad_norm": 2.2968366146087646, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6973019242286682, + "num_tokens": 66226791.0, + "step": 2570 + }, + { + "epoch": 0.28234131341972324, + "grad_norm": 2.2745795249938965, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7062147259712219, + "num_tokens": 66250927.0, + "step": 2571 + }, + { + "epoch": 0.28245113112233694, + "grad_norm": 2.3846821784973145, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.700741171836853, + "num_tokens": 66274019.0, + "step": 2572 + }, + { + "epoch": 0.2825609488249506, + "grad_norm": 2.3351025581359863, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7294186949729919, + "num_tokens": 66296912.0, + "step": 2573 + }, + { + "epoch": 0.28267076652756423, + "grad_norm": 2.576192855834961, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6835660338401794, + "num_tokens": 66319149.0, + "step": 2574 + }, + { + "epoch": 0.2827805842301779, + "grad_norm": 2.1215312480926514, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6891358494758606, + "num_tokens": 66345853.0, + "step": 2575 + }, + { + "epoch": 0.2828904019327916, + "grad_norm": 2.3633391857147217, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7028868198394775, + "num_tokens": 66368772.0, + "step": 2576 + }, + { + "epoch": 0.2830002196354052, + "grad_norm": 2.400235652923584, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.706188440322876, + "num_tokens": 66391522.0, + "step": 2577 + }, + { + "epoch": 0.28311003733801887, + "grad_norm": 2.1855602264404297, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.7023723125457764, + "num_tokens": 66418689.0, + "step": 2578 + }, + { + "epoch": 0.2832198550406326, + "grad_norm": 2.372091293334961, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7053477764129639, + "num_tokens": 66442714.0, + "step": 2579 + }, + { + "epoch": 0.2833296727432462, + "grad_norm": 2.2955715656280518, + "learning_rate": 1e-06, + "loss": 1.0721, + "mean_token_accuracy": 0.6822805404663086, + "num_tokens": 66469238.0, + "step": 2580 + }, + { + "epoch": 0.28343949044585987, + "grad_norm": 1.9592667818069458, + "learning_rate": 1e-06, + "loss": 1.0833, + "mean_token_accuracy": 0.6791400909423828, + "num_tokens": 66500975.0, + "step": 2581 + }, + { + "epoch": 0.2835493081484735, + "grad_norm": 2.1094396114349365, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7084767818450928, + "num_tokens": 66530612.0, + "step": 2582 + }, + { + "epoch": 0.2836591258510872, + "grad_norm": 2.260061025619507, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7169342637062073, + "num_tokens": 66554485.0, + "step": 2583 + }, + { + "epoch": 0.28376894355370086, + "grad_norm": 2.083505392074585, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7066947221755981, + "num_tokens": 66581818.0, + "step": 2584 + }, + { + "epoch": 0.2838787612563145, + "grad_norm": 2.5696284770965576, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6938437223434448, + "num_tokens": 66602572.0, + "step": 2585 + }, + { + "epoch": 0.2839885789589282, + "grad_norm": 2.1867411136627197, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7136363387107849, + "num_tokens": 66630653.0, + "step": 2586 + }, + { + "epoch": 0.28409839666154185, + "grad_norm": 1.8658699989318848, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.709773600101471, + "num_tokens": 66664450.0, + "step": 2587 + }, + { + "epoch": 0.2842082143641555, + "grad_norm": 2.1535897254943848, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7055758237838745, + "num_tokens": 66690703.0, + "step": 2588 + }, + { + "epoch": 0.28431803206676914, + "grad_norm": 2.421586275100708, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7002471685409546, + "num_tokens": 66712782.0, + "step": 2589 + }, + { + "epoch": 0.28442784976938285, + "grad_norm": 2.089174747467041, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.7008624076843262, + "num_tokens": 66742984.0, + "step": 2590 + }, + { + "epoch": 0.2845376674719965, + "grad_norm": 2.1922521591186523, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6966692209243774, + "num_tokens": 66768299.0, + "step": 2591 + }, + { + "epoch": 0.28464748517461014, + "grad_norm": 2.2467832565307617, + "learning_rate": 1e-06, + "loss": 1.0548, + "mean_token_accuracy": 0.6848447322845459, + "num_tokens": 66794569.0, + "step": 2592 + }, + { + "epoch": 0.2847573028772238, + "grad_norm": 2.2047979831695557, + "learning_rate": 1e-06, + "loss": 1.0947, + "mean_token_accuracy": 0.6754070520401001, + "num_tokens": 66824159.0, + "step": 2593 + }, + { + "epoch": 0.2848671205798375, + "grad_norm": 2.349292278289795, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6977270841598511, + "num_tokens": 66847419.0, + "step": 2594 + }, + { + "epoch": 0.28497693828245113, + "grad_norm": 2.3849196434020996, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6926895380020142, + "num_tokens": 66871350.0, + "step": 2595 + }, + { + "epoch": 0.2850867559850648, + "grad_norm": 2.348729372024536, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6945819854736328, + "num_tokens": 66895892.0, + "step": 2596 + }, + { + "epoch": 0.2851965736876785, + "grad_norm": 2.276054620742798, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7127465605735779, + "num_tokens": 66921161.0, + "step": 2597 + }, + { + "epoch": 0.2853063913902921, + "grad_norm": 2.117274045944214, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6996836066246033, + "num_tokens": 66949308.0, + "step": 2598 + }, + { + "epoch": 0.28541620909290577, + "grad_norm": 2.3559463024139404, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7130382061004639, + "num_tokens": 66972711.0, + "step": 2599 + }, + { + "epoch": 0.2855260267955194, + "grad_norm": 2.219754219055176, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6902238130569458, + "num_tokens": 67000821.0, + "step": 2600 + }, + { + "epoch": 0.2856358444981331, + "grad_norm": 2.167898416519165, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7098683714866638, + "num_tokens": 67026994.0, + "step": 2601 + }, + { + "epoch": 0.28574566220074676, + "grad_norm": 2.210122585296631, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7011468410491943, + "num_tokens": 67051567.0, + "step": 2602 + }, + { + "epoch": 0.2858554799033604, + "grad_norm": 2.067992925643921, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7261867523193359, + "num_tokens": 67078606.0, + "step": 2603 + }, + { + "epoch": 0.28596529760597406, + "grad_norm": 2.466945171356201, + "learning_rate": 1e-06, + "loss": 1.0944, + "mean_token_accuracy": 0.6786444783210754, + "num_tokens": 67101219.0, + "step": 2604 + }, + { + "epoch": 0.28607511530858776, + "grad_norm": 2.121521472930908, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.7018017172813416, + "num_tokens": 67128999.0, + "step": 2605 + }, + { + "epoch": 0.2861849330112014, + "grad_norm": 2.329848527908325, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7062006592750549, + "num_tokens": 67154528.0, + "step": 2606 + }, + { + "epoch": 0.28629475071381505, + "grad_norm": 2.2946088314056396, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7107075452804565, + "num_tokens": 67177621.0, + "step": 2607 + }, + { + "epoch": 0.28640456841642875, + "grad_norm": 2.315650463104248, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6923024654388428, + "num_tokens": 67202651.0, + "step": 2608 + }, + { + "epoch": 0.2865143861190424, + "grad_norm": 2.41737961769104, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7202430963516235, + "num_tokens": 67224132.0, + "step": 2609 + }, + { + "epoch": 0.28662420382165604, + "grad_norm": 2.485595703125, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7070082426071167, + "num_tokens": 67245034.0, + "step": 2610 + }, + { + "epoch": 0.2867340215242697, + "grad_norm": 2.1558287143707275, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6953692436218262, + "num_tokens": 67273482.0, + "step": 2611 + }, + { + "epoch": 0.2868438392268834, + "grad_norm": 1.9603155851364136, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7084747552871704, + "num_tokens": 67302837.0, + "step": 2612 + }, + { + "epoch": 0.28695365692949704, + "grad_norm": 2.535658597946167, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.708544909954071, + "num_tokens": 67323986.0, + "step": 2613 + }, + { + "epoch": 0.2870634746321107, + "grad_norm": 2.3136932849884033, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6910320520401001, + "num_tokens": 67349426.0, + "step": 2614 + }, + { + "epoch": 0.2871732923347244, + "grad_norm": 2.3423852920532227, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7035231590270996, + "num_tokens": 67374174.0, + "step": 2615 + }, + { + "epoch": 0.28728311003733803, + "grad_norm": 2.2556405067443848, + "learning_rate": 1e-06, + "loss": 1.1016, + "mean_token_accuracy": 0.6795907020568848, + "num_tokens": 67400937.0, + "step": 2616 + }, + { + "epoch": 0.2873929277399517, + "grad_norm": 2.5699164867401123, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6879321336746216, + "num_tokens": 67420887.0, + "step": 2617 + }, + { + "epoch": 0.2875027454425653, + "grad_norm": 2.516286611557007, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6893255710601807, + "num_tokens": 67442394.0, + "step": 2618 + }, + { + "epoch": 0.287612563145179, + "grad_norm": 2.5666234493255615, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6952031850814819, + "num_tokens": 67465076.0, + "step": 2619 + }, + { + "epoch": 0.28772238084779267, + "grad_norm": 2.239072322845459, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6865472793579102, + "num_tokens": 67489901.0, + "step": 2620 + }, + { + "epoch": 0.2878321985504063, + "grad_norm": 2.340552568435669, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.715632975101471, + "num_tokens": 67513110.0, + "step": 2621 + }, + { + "epoch": 0.28794201625301996, + "grad_norm": 2.406254768371582, + "learning_rate": 1e-06, + "loss": 1.0908, + "mean_token_accuracy": 0.6795657873153687, + "num_tokens": 67539227.0, + "step": 2622 + }, + { + "epoch": 0.28805183395563366, + "grad_norm": 2.404466152191162, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7224688529968262, + "num_tokens": 67561250.0, + "step": 2623 + }, + { + "epoch": 0.2881616516582473, + "grad_norm": 2.5102643966674805, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6921378374099731, + "num_tokens": 67584283.0, + "step": 2624 + }, + { + "epoch": 0.28827146936086095, + "grad_norm": 2.1216049194335938, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7015079855918884, + "num_tokens": 67611794.0, + "step": 2625 + }, + { + "epoch": 0.28838128706347466, + "grad_norm": 2.2104454040527344, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6839243769645691, + "num_tokens": 67639046.0, + "step": 2626 + }, + { + "epoch": 0.2884911047660883, + "grad_norm": 2.3898820877075195, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7074528932571411, + "num_tokens": 67662146.0, + "step": 2627 + }, + { + "epoch": 0.28860092246870195, + "grad_norm": 2.2067949771881104, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.7056635618209839, + "num_tokens": 67687544.0, + "step": 2628 + }, + { + "epoch": 0.2887107401713156, + "grad_norm": 2.1209425926208496, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6833679676055908, + "num_tokens": 67715681.0, + "step": 2629 + }, + { + "epoch": 0.2888205578739293, + "grad_norm": 2.3226356506347656, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7007969617843628, + "num_tokens": 67738188.0, + "step": 2630 + }, + { + "epoch": 0.28893037557654294, + "grad_norm": 2.1889896392822266, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6809622049331665, + "num_tokens": 67764325.0, + "step": 2631 + }, + { + "epoch": 0.2890401932791566, + "grad_norm": 2.193185329437256, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6884021162986755, + "num_tokens": 67792996.0, + "step": 2632 + }, + { + "epoch": 0.2891500109817703, + "grad_norm": 2.23537278175354, + "learning_rate": 1e-06, + "loss": 1.1073, + "mean_token_accuracy": 0.6751440167427063, + "num_tokens": 67820882.0, + "step": 2633 + }, + { + "epoch": 0.28925982868438394, + "grad_norm": 2.063652753829956, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7008010149002075, + "num_tokens": 67851675.0, + "step": 2634 + }, + { + "epoch": 0.2893696463869976, + "grad_norm": 2.2233011722564697, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6978543996810913, + "num_tokens": 67877372.0, + "step": 2635 + }, + { + "epoch": 0.2894794640896112, + "grad_norm": 2.341679811477661, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6975319385528564, + "num_tokens": 67900400.0, + "step": 2636 + }, + { + "epoch": 0.28958928179222493, + "grad_norm": 2.3358492851257324, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.7011162042617798, + "num_tokens": 67924759.0, + "step": 2637 + }, + { + "epoch": 0.2896990994948386, + "grad_norm": 2.4694669246673584, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7005358338356018, + "num_tokens": 67948351.0, + "step": 2638 + }, + { + "epoch": 0.2898089171974522, + "grad_norm": 2.1967501640319824, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.680224597454071, + "num_tokens": 67976137.0, + "step": 2639 + }, + { + "epoch": 0.28991873490006587, + "grad_norm": 2.2132186889648438, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7014248967170715, + "num_tokens": 68002112.0, + "step": 2640 + }, + { + "epoch": 0.29002855260267957, + "grad_norm": 2.2222957611083984, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7005710601806641, + "num_tokens": 68027473.0, + "step": 2641 + }, + { + "epoch": 0.2901383703052932, + "grad_norm": 1.9210683107376099, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7214070558547974, + "num_tokens": 68059745.0, + "step": 2642 + }, + { + "epoch": 0.29024818800790686, + "grad_norm": 2.2200567722320557, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.719908595085144, + "num_tokens": 68084908.0, + "step": 2643 + }, + { + "epoch": 0.29035800571052056, + "grad_norm": 2.347045660018921, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7119396924972534, + "num_tokens": 68107069.0, + "step": 2644 + }, + { + "epoch": 0.2904678234131342, + "grad_norm": 2.1683759689331055, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6966978311538696, + "num_tokens": 68134852.0, + "step": 2645 + }, + { + "epoch": 0.29057764111574785, + "grad_norm": 2.333005428314209, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6817578077316284, + "num_tokens": 68158512.0, + "step": 2646 + }, + { + "epoch": 0.2906874588183615, + "grad_norm": 2.0159590244293213, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.68877774477005, + "num_tokens": 68190202.0, + "step": 2647 + }, + { + "epoch": 0.2907972765209752, + "grad_norm": 2.1542856693267822, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7088709473609924, + "num_tokens": 68216030.0, + "step": 2648 + }, + { + "epoch": 0.29090709422358885, + "grad_norm": 2.4754998683929443, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6934827566146851, + "num_tokens": 68239601.0, + "step": 2649 + }, + { + "epoch": 0.2910169119262025, + "grad_norm": 2.1368601322174072, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6898916959762573, + "num_tokens": 68267099.0, + "step": 2650 + }, + { + "epoch": 0.29112672962881614, + "grad_norm": 2.3960189819335938, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6896718740463257, + "num_tokens": 68292599.0, + "step": 2651 + }, + { + "epoch": 0.29123654733142984, + "grad_norm": 2.3718619346618652, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6985980272293091, + "num_tokens": 68317039.0, + "step": 2652 + }, + { + "epoch": 0.2913463650340435, + "grad_norm": 2.429044246673584, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7011837959289551, + "num_tokens": 68338721.0, + "step": 2653 + }, + { + "epoch": 0.29145618273665713, + "grad_norm": 2.543757200241089, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7230169177055359, + "num_tokens": 68358767.0, + "step": 2654 + }, + { + "epoch": 0.29156600043927083, + "grad_norm": 2.169332504272461, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7035753726959229, + "num_tokens": 68384919.0, + "step": 2655 + }, + { + "epoch": 0.2916758181418845, + "grad_norm": 2.636874198913574, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7054871320724487, + "num_tokens": 68405024.0, + "step": 2656 + }, + { + "epoch": 0.2917856358444981, + "grad_norm": 2.3140921592712402, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6934827566146851, + "num_tokens": 68430417.0, + "step": 2657 + }, + { + "epoch": 0.29189545354711177, + "grad_norm": 2.3964908123016357, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7014433145523071, + "num_tokens": 68452508.0, + "step": 2658 + }, + { + "epoch": 0.2920052712497255, + "grad_norm": 2.3664464950561523, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6967670917510986, + "num_tokens": 68475177.0, + "step": 2659 + }, + { + "epoch": 0.2921150889523391, + "grad_norm": 2.311232805252075, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.720401406288147, + "num_tokens": 68497518.0, + "step": 2660 + }, + { + "epoch": 0.29222490665495277, + "grad_norm": 2.1939165592193604, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6866386532783508, + "num_tokens": 68520967.0, + "step": 2661 + }, + { + "epoch": 0.29233472435756647, + "grad_norm": 2.2876620292663574, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7108986973762512, + "num_tokens": 68545941.0, + "step": 2662 + }, + { + "epoch": 0.2924445420601801, + "grad_norm": 2.3318166732788086, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6904714107513428, + "num_tokens": 68569937.0, + "step": 2663 + }, + { + "epoch": 0.29255435976279376, + "grad_norm": 2.2208468914031982, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7244960069656372, + "num_tokens": 68594954.0, + "step": 2664 + }, + { + "epoch": 0.2926641774654074, + "grad_norm": 2.3758163452148438, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6987898945808411, + "num_tokens": 68618646.0, + "step": 2665 + }, + { + "epoch": 0.2927739951680211, + "grad_norm": 2.384162425994873, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7118120193481445, + "num_tokens": 68639915.0, + "step": 2666 + }, + { + "epoch": 0.29288381287063475, + "grad_norm": 1.9738718271255493, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.6851301789283752, + "num_tokens": 68670952.0, + "step": 2667 + }, + { + "epoch": 0.2929936305732484, + "grad_norm": 2.1528377532958984, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6917315125465393, + "num_tokens": 68699559.0, + "step": 2668 + }, + { + "epoch": 0.29310344827586204, + "grad_norm": 2.034367799758911, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6991955041885376, + "num_tokens": 68729291.0, + "step": 2669 + }, + { + "epoch": 0.29321326597847575, + "grad_norm": 2.732783794403076, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.709099292755127, + "num_tokens": 68747131.0, + "step": 2670 + }, + { + "epoch": 0.2933230836810894, + "grad_norm": 2.175955295562744, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.703777551651001, + "num_tokens": 68772623.0, + "step": 2671 + }, + { + "epoch": 0.29343290138370304, + "grad_norm": 2.0902628898620605, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.687125563621521, + "num_tokens": 68801208.0, + "step": 2672 + }, + { + "epoch": 0.29354271908631674, + "grad_norm": 2.5394556522369385, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7043339014053345, + "num_tokens": 68819564.0, + "step": 2673 + }, + { + "epoch": 0.2936525367889304, + "grad_norm": 2.1205148696899414, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7012342810630798, + "num_tokens": 68846893.0, + "step": 2674 + }, + { + "epoch": 0.29376235449154403, + "grad_norm": 2.2987687587738037, + "learning_rate": 1e-06, + "loss": 1.1113, + "mean_token_accuracy": 0.6722084283828735, + "num_tokens": 68873454.0, + "step": 2675 + }, + { + "epoch": 0.2938721721941577, + "grad_norm": 2.4106342792510986, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6943323016166687, + "num_tokens": 68896952.0, + "step": 2676 + }, + { + "epoch": 0.2939819898967714, + "grad_norm": 2.0715701580047607, + "learning_rate": 1e-06, + "loss": 1.0784, + "mean_token_accuracy": 0.6843626499176025, + "num_tokens": 68929138.0, + "step": 2677 + }, + { + "epoch": 0.294091807599385, + "grad_norm": 2.5403099060058594, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7057529091835022, + "num_tokens": 68949654.0, + "step": 2678 + }, + { + "epoch": 0.29420162530199867, + "grad_norm": 2.4045028686523438, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7028494477272034, + "num_tokens": 68970474.0, + "step": 2679 + }, + { + "epoch": 0.2943114430046123, + "grad_norm": 2.423816442489624, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6815993785858154, + "num_tokens": 68993909.0, + "step": 2680 + }, + { + "epoch": 0.294421260707226, + "grad_norm": 2.1768109798431396, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6894240975379944, + "num_tokens": 69022187.0, + "step": 2681 + }, + { + "epoch": 0.29453107840983966, + "grad_norm": 2.4574694633483887, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6913690567016602, + "num_tokens": 69042134.0, + "step": 2682 + }, + { + "epoch": 0.2946408961124533, + "grad_norm": 2.2138185501098633, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7083742618560791, + "num_tokens": 69068210.0, + "step": 2683 + }, + { + "epoch": 0.294750713815067, + "grad_norm": 2.2900614738464355, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6959548592567444, + "num_tokens": 69092508.0, + "step": 2684 + }, + { + "epoch": 0.29486053151768066, + "grad_norm": 2.397944450378418, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6745392084121704, + "num_tokens": 69114422.0, + "step": 2685 + }, + { + "epoch": 0.2949703492202943, + "grad_norm": 2.371976137161255, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7215259075164795, + "num_tokens": 69137193.0, + "step": 2686 + }, + { + "epoch": 0.29508016692290795, + "grad_norm": 2.3750579357147217, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7057396173477173, + "num_tokens": 69160323.0, + "step": 2687 + }, + { + "epoch": 0.29518998462552165, + "grad_norm": 2.232050895690918, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6968324184417725, + "num_tokens": 69185994.0, + "step": 2688 + }, + { + "epoch": 0.2952998023281353, + "grad_norm": 2.1268343925476074, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7017027139663696, + "num_tokens": 69212598.0, + "step": 2689 + }, + { + "epoch": 0.29540962003074894, + "grad_norm": 2.751400947570801, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.694165050983429, + "num_tokens": 69230120.0, + "step": 2690 + }, + { + "epoch": 0.29551943773336264, + "grad_norm": 2.208136796951294, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6882641315460205, + "num_tokens": 69257640.0, + "step": 2691 + }, + { + "epoch": 0.2956292554359763, + "grad_norm": 2.160332441329956, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7061116099357605, + "num_tokens": 69283423.0, + "step": 2692 + }, + { + "epoch": 0.29573907313858994, + "grad_norm": 2.1534652709960938, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6830586194992065, + "num_tokens": 69312582.0, + "step": 2693 + }, + { + "epoch": 0.2958488908412036, + "grad_norm": 2.203207015991211, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7008521556854248, + "num_tokens": 69339366.0, + "step": 2694 + }, + { + "epoch": 0.2959587085438173, + "grad_norm": 2.3543479442596436, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7018016576766968, + "num_tokens": 69362217.0, + "step": 2695 + }, + { + "epoch": 0.29606852624643093, + "grad_norm": 2.2285711765289307, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7306168079376221, + "num_tokens": 69387270.0, + "step": 2696 + }, + { + "epoch": 0.2961783439490446, + "grad_norm": 2.253194808959961, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.698720395565033, + "num_tokens": 69412542.0, + "step": 2697 + }, + { + "epoch": 0.2962881616516582, + "grad_norm": 2.0486998558044434, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6949167251586914, + "num_tokens": 69441563.0, + "step": 2698 + }, + { + "epoch": 0.2963979793542719, + "grad_norm": 2.5860087871551514, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.682675302028656, + "num_tokens": 69463032.0, + "step": 2699 + }, + { + "epoch": 0.29650779705688557, + "grad_norm": 2.216852903366089, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6965668201446533, + "num_tokens": 69488169.0, + "step": 2700 + }, + { + "epoch": 0.2966176147594992, + "grad_norm": 2.139080047607422, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6866708993911743, + "num_tokens": 69515432.0, + "step": 2701 + }, + { + "epoch": 0.2967274324621129, + "grad_norm": 2.1795494556427, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7219293117523193, + "num_tokens": 69540555.0, + "step": 2702 + }, + { + "epoch": 0.29683725016472656, + "grad_norm": 2.8490402698516846, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7122446298599243, + "num_tokens": 69557411.0, + "step": 2703 + }, + { + "epoch": 0.2969470678673402, + "grad_norm": 1.9928737878799438, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7230172157287598, + "num_tokens": 69586560.0, + "step": 2704 + }, + { + "epoch": 0.29705688556995385, + "grad_norm": 2.325613021850586, + "learning_rate": 1e-06, + "loss": 1.0614, + "mean_token_accuracy": 0.6794419288635254, + "num_tokens": 69613491.0, + "step": 2705 + }, + { + "epoch": 0.29716670327256756, + "grad_norm": 2.599228620529175, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6857853531837463, + "num_tokens": 69633829.0, + "step": 2706 + }, + { + "epoch": 0.2972765209751812, + "grad_norm": 2.3332927227020264, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6849572658538818, + "num_tokens": 69659146.0, + "step": 2707 + }, + { + "epoch": 0.29738633867779485, + "grad_norm": 2.3150126934051514, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6996015310287476, + "num_tokens": 69684605.0, + "step": 2708 + }, + { + "epoch": 0.29749615638040855, + "grad_norm": 2.442253351211548, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7125980257987976, + "num_tokens": 69705519.0, + "step": 2709 + }, + { + "epoch": 0.2976059740830222, + "grad_norm": 2.21533203125, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6881934404373169, + "num_tokens": 69732875.0, + "step": 2710 + }, + { + "epoch": 0.29771579178563584, + "grad_norm": 2.164785861968994, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6980804204940796, + "num_tokens": 69763334.0, + "step": 2711 + }, + { + "epoch": 0.2978256094882495, + "grad_norm": 2.1047191619873047, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6951529383659363, + "num_tokens": 69791030.0, + "step": 2712 + }, + { + "epoch": 0.2979354271908632, + "grad_norm": 2.624173164367676, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7185349464416504, + "num_tokens": 69808944.0, + "step": 2713 + }, + { + "epoch": 0.29804524489347684, + "grad_norm": 2.449350118637085, + "learning_rate": 1e-06, + "loss": 1.0708, + "mean_token_accuracy": 0.6831986308097839, + "num_tokens": 69831207.0, + "step": 2714 + }, + { + "epoch": 0.2981550625960905, + "grad_norm": 2.138153314590454, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.704888641834259, + "num_tokens": 69858181.0, + "step": 2715 + }, + { + "epoch": 0.2982648802987041, + "grad_norm": 2.522944211959839, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6800106167793274, + "num_tokens": 69882180.0, + "step": 2716 + }, + { + "epoch": 0.29837469800131783, + "grad_norm": 2.0906589031219482, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6866218447685242, + "num_tokens": 69913378.0, + "step": 2717 + }, + { + "epoch": 0.2984845157039315, + "grad_norm": 2.3915483951568604, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.712939441204071, + "num_tokens": 69934732.0, + "step": 2718 + }, + { + "epoch": 0.2985943334065451, + "grad_norm": 2.5410783290863037, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7123539447784424, + "num_tokens": 69955836.0, + "step": 2719 + }, + { + "epoch": 0.2987041511091588, + "grad_norm": 2.452791929244995, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.6821520328521729, + "num_tokens": 69980010.0, + "step": 2720 + }, + { + "epoch": 0.29881396881177247, + "grad_norm": 2.2524795532226562, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6959238052368164, + "num_tokens": 70007716.0, + "step": 2721 + }, + { + "epoch": 0.2989237865143861, + "grad_norm": 2.3899364471435547, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6919972896575928, + "num_tokens": 70032150.0, + "step": 2722 + }, + { + "epoch": 0.29903360421699976, + "grad_norm": 2.2478199005126953, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7186046838760376, + "num_tokens": 70055241.0, + "step": 2723 + }, + { + "epoch": 0.29914342191961346, + "grad_norm": 2.1704373359680176, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7146927118301392, + "num_tokens": 70081564.0, + "step": 2724 + }, + { + "epoch": 0.2992532396222271, + "grad_norm": 2.2232649326324463, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.699743390083313, + "num_tokens": 70109088.0, + "step": 2725 + }, + { + "epoch": 0.29936305732484075, + "grad_norm": 2.4733057022094727, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7008715271949768, + "num_tokens": 70131258.0, + "step": 2726 + }, + { + "epoch": 0.2994728750274544, + "grad_norm": 2.841681718826294, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7144041061401367, + "num_tokens": 70148340.0, + "step": 2727 + }, + { + "epoch": 0.2995826927300681, + "grad_norm": 2.415461778640747, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6909099817276001, + "num_tokens": 70172990.0, + "step": 2728 + }, + { + "epoch": 0.29969251043268175, + "grad_norm": 2.485761880874634, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6944709420204163, + "num_tokens": 70195233.0, + "step": 2729 + }, + { + "epoch": 0.2998023281352954, + "grad_norm": 2.2315900325775146, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7028810381889343, + "num_tokens": 70220615.0, + "step": 2730 + }, + { + "epoch": 0.2999121458379091, + "grad_norm": 2.283829689025879, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.7042057514190674, + "num_tokens": 70247186.0, + "step": 2731 + }, + { + "epoch": 0.30002196354052274, + "grad_norm": 2.2650444507598877, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.704105794429779, + "num_tokens": 70273363.0, + "step": 2732 + }, + { + "epoch": 0.3001317812431364, + "grad_norm": 2.432698965072632, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6967143416404724, + "num_tokens": 70296770.0, + "step": 2733 + }, + { + "epoch": 0.30024159894575003, + "grad_norm": 2.514810562133789, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6952950954437256, + "num_tokens": 70321381.0, + "step": 2734 + }, + { + "epoch": 0.30035141664836373, + "grad_norm": 1.8626471757888794, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7038630843162537, + "num_tokens": 70358511.0, + "step": 2735 + }, + { + "epoch": 0.3004612343509774, + "grad_norm": 2.1388723850250244, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6869392395019531, + "num_tokens": 70387734.0, + "step": 2736 + }, + { + "epoch": 0.300571052053591, + "grad_norm": 2.0473880767822266, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7293427586555481, + "num_tokens": 70416279.0, + "step": 2737 + }, + { + "epoch": 0.3006808697562047, + "grad_norm": 2.349370002746582, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7285376787185669, + "num_tokens": 70438408.0, + "step": 2738 + }, + { + "epoch": 0.3007906874588184, + "grad_norm": 2.2466959953308105, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6984921097755432, + "num_tokens": 70462028.0, + "step": 2739 + }, + { + "epoch": 0.300900505161432, + "grad_norm": 2.0457630157470703, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.688042402267456, + "num_tokens": 70493228.0, + "step": 2740 + }, + { + "epoch": 0.30101032286404567, + "grad_norm": 2.2147247791290283, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6810132265090942, + "num_tokens": 70521961.0, + "step": 2741 + }, + { + "epoch": 0.30112014056665937, + "grad_norm": 2.280127763748169, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6726110577583313, + "num_tokens": 70547650.0, + "step": 2742 + }, + { + "epoch": 0.301229958269273, + "grad_norm": 2.084264039993286, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.690803050994873, + "num_tokens": 70576087.0, + "step": 2743 + }, + { + "epoch": 0.30133977597188666, + "grad_norm": 2.3699522018432617, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7048529386520386, + "num_tokens": 70599199.0, + "step": 2744 + }, + { + "epoch": 0.3014495936745003, + "grad_norm": 1.997434139251709, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6873899698257446, + "num_tokens": 70628659.0, + "step": 2745 + }, + { + "epoch": 0.301559411377114, + "grad_norm": 2.5231494903564453, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.7096904516220093, + "num_tokens": 70649974.0, + "step": 2746 + }, + { + "epoch": 0.30166922907972765, + "grad_norm": 2.629997491836548, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.706795334815979, + "num_tokens": 70670084.0, + "step": 2747 + }, + { + "epoch": 0.3017790467823413, + "grad_norm": 2.5463664531707764, + "learning_rate": 1e-06, + "loss": 1.1048, + "mean_token_accuracy": 0.6808131337165833, + "num_tokens": 70693967.0, + "step": 2748 + }, + { + "epoch": 0.301888864484955, + "grad_norm": 2.408644199371338, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6864484548568726, + "num_tokens": 70717415.0, + "step": 2749 + }, + { + "epoch": 0.30199868218756865, + "grad_norm": 2.1810600757598877, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6970754861831665, + "num_tokens": 70746988.0, + "step": 2750 + }, + { + "epoch": 0.3021084998901823, + "grad_norm": 2.3636205196380615, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6827938556671143, + "num_tokens": 70770290.0, + "step": 2751 + }, + { + "epoch": 0.30221831759279594, + "grad_norm": 2.3381521701812744, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6857632994651794, + "num_tokens": 70794124.0, + "step": 2752 + }, + { + "epoch": 0.30232813529540964, + "grad_norm": 2.089172840118408, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.693057119846344, + "num_tokens": 70825746.0, + "step": 2753 + }, + { + "epoch": 0.3024379529980233, + "grad_norm": 2.165886878967285, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.7089194655418396, + "num_tokens": 70852316.0, + "step": 2754 + }, + { + "epoch": 0.30254777070063693, + "grad_norm": 2.039290428161621, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7176468372344971, + "num_tokens": 70881820.0, + "step": 2755 + }, + { + "epoch": 0.3026575884032506, + "grad_norm": 2.209298849105835, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6924934387207031, + "num_tokens": 70908142.0, + "step": 2756 + }, + { + "epoch": 0.3027674061058643, + "grad_norm": 2.2827134132385254, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.696639895439148, + "num_tokens": 70931882.0, + "step": 2757 + }, + { + "epoch": 0.3028772238084779, + "grad_norm": 2.0058491230010986, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7035377025604248, + "num_tokens": 70961862.0, + "step": 2758 + }, + { + "epoch": 0.30298704151109157, + "grad_norm": 2.1541829109191895, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7177761197090149, + "num_tokens": 70990436.0, + "step": 2759 + }, + { + "epoch": 0.30309685921370527, + "grad_norm": 2.669048547744751, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7176465392112732, + "num_tokens": 71008405.0, + "step": 2760 + }, + { + "epoch": 0.3032066769163189, + "grad_norm": 2.236835479736328, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7111196517944336, + "num_tokens": 71034208.0, + "step": 2761 + }, + { + "epoch": 0.30331649461893256, + "grad_norm": 2.2628724575042725, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6855969429016113, + "num_tokens": 71060791.0, + "step": 2762 + }, + { + "epoch": 0.3034263123215462, + "grad_norm": 2.287848711013794, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7132052183151245, + "num_tokens": 71082906.0, + "step": 2763 + }, + { + "epoch": 0.3035361300241599, + "grad_norm": 2.3759653568267822, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6838962435722351, + "num_tokens": 71105562.0, + "step": 2764 + }, + { + "epoch": 0.30364594772677356, + "grad_norm": 2.187779426574707, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7101823687553406, + "num_tokens": 71131316.0, + "step": 2765 + }, + { + "epoch": 0.3037557654293872, + "grad_norm": 2.3924670219421387, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6912840008735657, + "num_tokens": 71156095.0, + "step": 2766 + }, + { + "epoch": 0.3038655831320009, + "grad_norm": 2.224382162094116, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6953943371772766, + "num_tokens": 71185429.0, + "step": 2767 + }, + { + "epoch": 0.30397540083461455, + "grad_norm": 2.3308722972869873, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.712565541267395, + "num_tokens": 71210553.0, + "step": 2768 + }, + { + "epoch": 0.3040852185372282, + "grad_norm": 2.3054990768432617, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6950357556343079, + "num_tokens": 71236549.0, + "step": 2769 + }, + { + "epoch": 0.30419503623984184, + "grad_norm": 2.4736952781677246, + "learning_rate": 1e-06, + "loss": 1.0765, + "mean_token_accuracy": 0.6834908127784729, + "num_tokens": 71259137.0, + "step": 2770 + }, + { + "epoch": 0.30430485394245554, + "grad_norm": 2.4812240600585938, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.701590895652771, + "num_tokens": 71281391.0, + "step": 2771 + }, + { + "epoch": 0.3044146716450692, + "grad_norm": 2.5280075073242188, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6885474324226379, + "num_tokens": 71303902.0, + "step": 2772 + }, + { + "epoch": 0.30452448934768284, + "grad_norm": 2.1000354290008545, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.703947901725769, + "num_tokens": 71332687.0, + "step": 2773 + }, + { + "epoch": 0.3046343070502965, + "grad_norm": 2.6090900897979736, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7073816061019897, + "num_tokens": 71354801.0, + "step": 2774 + }, + { + "epoch": 0.3047441247529102, + "grad_norm": 2.0389137268066406, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6834238767623901, + "num_tokens": 71385578.0, + "step": 2775 + }, + { + "epoch": 0.30485394245552383, + "grad_norm": 2.035421848297119, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.685124397277832, + "num_tokens": 71418529.0, + "step": 2776 + }, + { + "epoch": 0.3049637601581375, + "grad_norm": 2.633626699447632, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7250381708145142, + "num_tokens": 71438699.0, + "step": 2777 + }, + { + "epoch": 0.3050735778607512, + "grad_norm": 2.253998041152954, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.702370285987854, + "num_tokens": 71462947.0, + "step": 2778 + }, + { + "epoch": 0.3051833955633648, + "grad_norm": 1.9677679538726807, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7028121948242188, + "num_tokens": 71492465.0, + "step": 2779 + }, + { + "epoch": 0.30529321326597847, + "grad_norm": 1.9171088933944702, + "learning_rate": 1e-06, + "loss": 1.0875, + "mean_token_accuracy": 0.6811911463737488, + "num_tokens": 71526431.0, + "step": 2780 + }, + { + "epoch": 0.3054030309685921, + "grad_norm": 2.1087138652801514, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6810591220855713, + "num_tokens": 71554201.0, + "step": 2781 + }, + { + "epoch": 0.3055128486712058, + "grad_norm": 2.1952784061431885, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6918252110481262, + "num_tokens": 71582176.0, + "step": 2782 + }, + { + "epoch": 0.30562266637381946, + "grad_norm": 2.352665901184082, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6864363551139832, + "num_tokens": 71605876.0, + "step": 2783 + }, + { + "epoch": 0.3057324840764331, + "grad_norm": 2.0347237586975098, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7154976725578308, + "num_tokens": 71632329.0, + "step": 2784 + }, + { + "epoch": 0.30584230177904675, + "grad_norm": 2.042001485824585, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7106290459632874, + "num_tokens": 71661277.0, + "step": 2785 + }, + { + "epoch": 0.30595211948166046, + "grad_norm": 1.9049816131591797, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7044034004211426, + "num_tokens": 71696439.0, + "step": 2786 + }, + { + "epoch": 0.3060619371842741, + "grad_norm": 2.272122383117676, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.6872023344039917, + "num_tokens": 71721429.0, + "step": 2787 + }, + { + "epoch": 0.30617175488688775, + "grad_norm": 2.1722493171691895, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7267861366271973, + "num_tokens": 71747485.0, + "step": 2788 + }, + { + "epoch": 0.30628157258950145, + "grad_norm": 2.3353164196014404, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6961899995803833, + "num_tokens": 71772918.0, + "step": 2789 + }, + { + "epoch": 0.3063913902921151, + "grad_norm": 2.263887405395508, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7281045913696289, + "num_tokens": 71796523.0, + "step": 2790 + }, + { + "epoch": 0.30650120799472874, + "grad_norm": 2.6220483779907227, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7281476259231567, + "num_tokens": 71817444.0, + "step": 2791 + }, + { + "epoch": 0.3066110256973424, + "grad_norm": 2.0831384658813477, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7013427019119263, + "num_tokens": 71844469.0, + "step": 2792 + }, + { + "epoch": 0.3067208433999561, + "grad_norm": 2.1993327140808105, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6896693110466003, + "num_tokens": 71871338.0, + "step": 2793 + }, + { + "epoch": 0.30683066110256974, + "grad_norm": 2.422990560531616, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7117453813552856, + "num_tokens": 71892339.0, + "step": 2794 + }, + { + "epoch": 0.3069404788051834, + "grad_norm": 2.3769776821136475, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7114864587783813, + "num_tokens": 71915498.0, + "step": 2795 + }, + { + "epoch": 0.3070502965077971, + "grad_norm": 2.378648519515991, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.702572762966156, + "num_tokens": 71940841.0, + "step": 2796 + }, + { + "epoch": 0.30716011421041073, + "grad_norm": 2.2364864349365234, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.682563304901123, + "num_tokens": 71967744.0, + "step": 2797 + }, + { + "epoch": 0.3072699319130244, + "grad_norm": 2.166201114654541, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.710240364074707, + "num_tokens": 71995579.0, + "step": 2798 + }, + { + "epoch": 0.307379749615638, + "grad_norm": 2.355069637298584, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7042628526687622, + "num_tokens": 72019681.0, + "step": 2799 + }, + { + "epoch": 0.3074895673182517, + "grad_norm": 2.1856021881103516, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6910150647163391, + "num_tokens": 72048337.0, + "step": 2800 + }, + { + "epoch": 0.30759938502086537, + "grad_norm": 1.9892258644104004, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6962810158729553, + "num_tokens": 72079622.0, + "step": 2801 + }, + { + "epoch": 0.307709202723479, + "grad_norm": 2.149019956588745, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6780229806900024, + "num_tokens": 72107797.0, + "step": 2802 + }, + { + "epoch": 0.30781902042609266, + "grad_norm": 2.2916605472564697, + "learning_rate": 1e-06, + "loss": 1.1417, + "mean_token_accuracy": 0.6651760339736938, + "num_tokens": 72134298.0, + "step": 2803 + }, + { + "epoch": 0.30792883812870636, + "grad_norm": 1.9812947511672974, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6850376129150391, + "num_tokens": 72165932.0, + "step": 2804 + }, + { + "epoch": 0.30803865583132, + "grad_norm": 2.4793176651000977, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7037940621376038, + "num_tokens": 72186010.0, + "step": 2805 + }, + { + "epoch": 0.30814847353393365, + "grad_norm": 2.0387301445007324, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.701526403427124, + "num_tokens": 72215904.0, + "step": 2806 + }, + { + "epoch": 0.30825829123654735, + "grad_norm": 2.340195655822754, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7237647771835327, + "num_tokens": 72238011.0, + "step": 2807 + }, + { + "epoch": 0.308368108939161, + "grad_norm": 2.0611631870269775, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7278286814689636, + "num_tokens": 72264900.0, + "step": 2808 + }, + { + "epoch": 0.30847792664177465, + "grad_norm": 2.368380546569824, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6946018934249878, + "num_tokens": 72287157.0, + "step": 2809 + }, + { + "epoch": 0.3085877443443883, + "grad_norm": 2.21972393989563, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6926852464675903, + "num_tokens": 72312575.0, + "step": 2810 + }, + { + "epoch": 0.308697562047002, + "grad_norm": 2.1300766468048096, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.688962459564209, + "num_tokens": 72340143.0, + "step": 2811 + }, + { + "epoch": 0.30880737974961564, + "grad_norm": 2.0545096397399902, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7096787691116333, + "num_tokens": 72369629.0, + "step": 2812 + }, + { + "epoch": 0.3089171974522293, + "grad_norm": 2.110063076019287, + "learning_rate": 1e-06, + "loss": 1.0709, + "mean_token_accuracy": 0.6774806976318359, + "num_tokens": 72401078.0, + "step": 2813 + }, + { + "epoch": 0.309027015154843, + "grad_norm": 1.931673526763916, + "learning_rate": 1e-06, + "loss": 1.092, + "mean_token_accuracy": 0.6747004985809326, + "num_tokens": 72436205.0, + "step": 2814 + }, + { + "epoch": 0.30913683285745663, + "grad_norm": 2.1943609714508057, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6801490783691406, + "num_tokens": 72461817.0, + "step": 2815 + }, + { + "epoch": 0.3092466505600703, + "grad_norm": 2.3931286334991455, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.7053360342979431, + "num_tokens": 72487620.0, + "step": 2816 + }, + { + "epoch": 0.3093564682626839, + "grad_norm": 2.2109811305999756, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7012679576873779, + "num_tokens": 72515292.0, + "step": 2817 + }, + { + "epoch": 0.3094662859652976, + "grad_norm": 2.3968799114227295, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6986435651779175, + "num_tokens": 72537155.0, + "step": 2818 + }, + { + "epoch": 0.3095761036679113, + "grad_norm": 2.422778367996216, + "learning_rate": 1e-06, + "loss": 1.0815, + "mean_token_accuracy": 0.6802285313606262, + "num_tokens": 72561608.0, + "step": 2819 + }, + { + "epoch": 0.3096859213705249, + "grad_norm": 2.319674491882324, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6926935911178589, + "num_tokens": 72585586.0, + "step": 2820 + }, + { + "epoch": 0.30979573907313857, + "grad_norm": 1.9890834093093872, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6880656480789185, + "num_tokens": 72619250.0, + "step": 2821 + }, + { + "epoch": 0.30990555677575227, + "grad_norm": 2.2941360473632812, + "learning_rate": 1e-06, + "loss": 1.0633, + "mean_token_accuracy": 0.6780910491943359, + "num_tokens": 72645798.0, + "step": 2822 + }, + { + "epoch": 0.3100153744783659, + "grad_norm": 2.3500075340270996, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7137858271598816, + "num_tokens": 72668395.0, + "step": 2823 + }, + { + "epoch": 0.31012519218097956, + "grad_norm": 2.1495823860168457, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6976787447929382, + "num_tokens": 72694071.0, + "step": 2824 + }, + { + "epoch": 0.31023500988359326, + "grad_norm": 2.15645432472229, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7065965533256531, + "num_tokens": 72722133.0, + "step": 2825 + }, + { + "epoch": 0.3103448275862069, + "grad_norm": 2.0086114406585693, + "learning_rate": 1e-06, + "loss": 1.0776, + "mean_token_accuracy": 0.6743863821029663, + "num_tokens": 72753971.0, + "step": 2826 + }, + { + "epoch": 0.31045464528882055, + "grad_norm": 2.6067211627960205, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7204997539520264, + "num_tokens": 72770710.0, + "step": 2827 + }, + { + "epoch": 0.3105644629914342, + "grad_norm": 2.1648142337799072, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6847658753395081, + "num_tokens": 72798938.0, + "step": 2828 + }, + { + "epoch": 0.3106742806940479, + "grad_norm": 2.238115072250366, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6883219480514526, + "num_tokens": 72823460.0, + "step": 2829 + }, + { + "epoch": 0.31078409839666155, + "grad_norm": 2.0969762802124023, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6817617416381836, + "num_tokens": 72852317.0, + "step": 2830 + }, + { + "epoch": 0.3108939160992752, + "grad_norm": 2.1990206241607666, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7085494995117188, + "num_tokens": 72878548.0, + "step": 2831 + }, + { + "epoch": 0.31100373380188884, + "grad_norm": 2.1918203830718994, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.7045072317123413, + "num_tokens": 72904542.0, + "step": 2832 + }, + { + "epoch": 0.31111355150450254, + "grad_norm": 2.3336517810821533, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6984425187110901, + "num_tokens": 72930604.0, + "step": 2833 + }, + { + "epoch": 0.3112233692071162, + "grad_norm": 2.3735547065734863, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6865417957305908, + "num_tokens": 72955416.0, + "step": 2834 + }, + { + "epoch": 0.31133318690972983, + "grad_norm": 2.265075922012329, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6780059337615967, + "num_tokens": 72980245.0, + "step": 2835 + }, + { + "epoch": 0.31144300461234353, + "grad_norm": 2.074355125427246, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7275422215461731, + "num_tokens": 73009422.0, + "step": 2836 + }, + { + "epoch": 0.3115528223149572, + "grad_norm": 2.252847194671631, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.697742223739624, + "num_tokens": 73035377.0, + "step": 2837 + }, + { + "epoch": 0.3116626400175708, + "grad_norm": 2.300642490386963, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6925241947174072, + "num_tokens": 73060371.0, + "step": 2838 + }, + { + "epoch": 0.31177245772018447, + "grad_norm": 2.218022584915161, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6952817440032959, + "num_tokens": 73086060.0, + "step": 2839 + }, + { + "epoch": 0.31188227542279817, + "grad_norm": 2.3248398303985596, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6964163780212402, + "num_tokens": 73108297.0, + "step": 2840 + }, + { + "epoch": 0.3119920931254118, + "grad_norm": 2.1011180877685547, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7045633792877197, + "num_tokens": 73135379.0, + "step": 2841 + }, + { + "epoch": 0.31210191082802546, + "grad_norm": 2.221778631210327, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6986809968948364, + "num_tokens": 73162023.0, + "step": 2842 + }, + { + "epoch": 0.31221172853063917, + "grad_norm": 1.8561800718307495, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6935341358184814, + "num_tokens": 73196159.0, + "step": 2843 + }, + { + "epoch": 0.3123215462332528, + "grad_norm": 2.3932178020477295, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6999046802520752, + "num_tokens": 73219080.0, + "step": 2844 + }, + { + "epoch": 0.31243136393586646, + "grad_norm": 2.206738233566284, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6931849718093872, + "num_tokens": 73247380.0, + "step": 2845 + }, + { + "epoch": 0.3125411816384801, + "grad_norm": 2.2073397636413574, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.711559534072876, + "num_tokens": 73271690.0, + "step": 2846 + }, + { + "epoch": 0.3126509993410938, + "grad_norm": 2.425485610961914, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6998664140701294, + "num_tokens": 73292421.0, + "step": 2847 + }, + { + "epoch": 0.31276081704370745, + "grad_norm": 2.1402690410614014, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7034746408462524, + "num_tokens": 73319691.0, + "step": 2848 + }, + { + "epoch": 0.3128706347463211, + "grad_norm": 2.239915609359741, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6857868432998657, + "num_tokens": 73346090.0, + "step": 2849 + }, + { + "epoch": 0.31298045244893474, + "grad_norm": 2.2789146900177, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6988750100135803, + "num_tokens": 73370375.0, + "step": 2850 + }, + { + "epoch": 0.31309027015154844, + "grad_norm": 2.256042242050171, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.704208254814148, + "num_tokens": 73395846.0, + "step": 2851 + }, + { + "epoch": 0.3132000878541621, + "grad_norm": 1.9766459465026855, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.7016265988349915, + "num_tokens": 73424054.0, + "step": 2852 + }, + { + "epoch": 0.31330990555677574, + "grad_norm": 2.343698024749756, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6858257055282593, + "num_tokens": 73447756.0, + "step": 2853 + }, + { + "epoch": 0.31341972325938944, + "grad_norm": 2.151154041290283, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.691300630569458, + "num_tokens": 73476588.0, + "step": 2854 + }, + { + "epoch": 0.3135295409620031, + "grad_norm": 2.0481486320495605, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7169091701507568, + "num_tokens": 73506510.0, + "step": 2855 + }, + { + "epoch": 0.31363935866461673, + "grad_norm": 2.095120668411255, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6990631818771362, + "num_tokens": 73537145.0, + "step": 2856 + }, + { + "epoch": 0.3137491763672304, + "grad_norm": 2.2793993949890137, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6937181949615479, + "num_tokens": 73561824.0, + "step": 2857 + }, + { + "epoch": 0.3138589940698441, + "grad_norm": 1.8971741199493408, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7173258662223816, + "num_tokens": 73593356.0, + "step": 2858 + }, + { + "epoch": 0.3139688117724577, + "grad_norm": 2.308894157409668, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7257474660873413, + "num_tokens": 73618252.0, + "step": 2859 + }, + { + "epoch": 0.31407862947507137, + "grad_norm": 2.3472487926483154, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7016580104827881, + "num_tokens": 73641866.0, + "step": 2860 + }, + { + "epoch": 0.314188447177685, + "grad_norm": 2.099149703979492, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6946338415145874, + "num_tokens": 73672357.0, + "step": 2861 + }, + { + "epoch": 0.3142982648802987, + "grad_norm": 2.3190698623657227, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7060605883598328, + "num_tokens": 73695297.0, + "step": 2862 + }, + { + "epoch": 0.31440808258291236, + "grad_norm": 2.264211654663086, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6946359872817993, + "num_tokens": 73720415.0, + "step": 2863 + }, + { + "epoch": 0.314517900285526, + "grad_norm": 2.2950124740600586, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6880230903625488, + "num_tokens": 73745175.0, + "step": 2864 + }, + { + "epoch": 0.3146277179881397, + "grad_norm": 2.2583556175231934, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7138899564743042, + "num_tokens": 73770217.0, + "step": 2865 + }, + { + "epoch": 0.31473753569075336, + "grad_norm": 2.232649564743042, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.68175208568573, + "num_tokens": 73796266.0, + "step": 2866 + }, + { + "epoch": 0.314847353393367, + "grad_norm": 2.3437271118164062, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7154674530029297, + "num_tokens": 73819594.0, + "step": 2867 + }, + { + "epoch": 0.31495717109598065, + "grad_norm": 2.2208619117736816, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6960643529891968, + "num_tokens": 73842898.0, + "step": 2868 + }, + { + "epoch": 0.31506698879859435, + "grad_norm": 2.4833531379699707, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6971721649169922, + "num_tokens": 73865223.0, + "step": 2869 + }, + { + "epoch": 0.315176806501208, + "grad_norm": 2.1913793087005615, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6919801235198975, + "num_tokens": 73889161.0, + "step": 2870 + }, + { + "epoch": 0.31528662420382164, + "grad_norm": 1.906585693359375, + "learning_rate": 1e-06, + "loss": 1.0837, + "mean_token_accuracy": 0.677539587020874, + "num_tokens": 73923807.0, + "step": 2871 + }, + { + "epoch": 0.31539644190643534, + "grad_norm": 2.2152607440948486, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.690436065196991, + "num_tokens": 73949920.0, + "step": 2872 + }, + { + "epoch": 0.315506259609049, + "grad_norm": 2.1352641582489014, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7082886695861816, + "num_tokens": 73975950.0, + "step": 2873 + }, + { + "epoch": 0.31561607731166264, + "grad_norm": 2.0596089363098145, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7178176641464233, + "num_tokens": 74003808.0, + "step": 2874 + }, + { + "epoch": 0.3157258950142763, + "grad_norm": 2.318549871444702, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7066439986228943, + "num_tokens": 74027669.0, + "step": 2875 + }, + { + "epoch": 0.31583571271689, + "grad_norm": 2.1040854454040527, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6922823190689087, + "num_tokens": 74054647.0, + "step": 2876 + }, + { + "epoch": 0.31594553041950363, + "grad_norm": 2.3520755767822266, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6956688165664673, + "num_tokens": 74078506.0, + "step": 2877 + }, + { + "epoch": 0.3160553481221173, + "grad_norm": 2.1914970874786377, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6901872158050537, + "num_tokens": 74106507.0, + "step": 2878 + }, + { + "epoch": 0.3161651658247309, + "grad_norm": 2.3659415245056152, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7215461134910583, + "num_tokens": 74129737.0, + "step": 2879 + }, + { + "epoch": 0.3162749835273446, + "grad_norm": 2.243729591369629, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7302069067955017, + "num_tokens": 74151641.0, + "step": 2880 + }, + { + "epoch": 0.31638480122995827, + "grad_norm": 2.198357343673706, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7168039083480835, + "num_tokens": 74176773.0, + "step": 2881 + }, + { + "epoch": 0.3164946189325719, + "grad_norm": 2.128812074661255, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6764543056488037, + "num_tokens": 74206514.0, + "step": 2882 + }, + { + "epoch": 0.3166044366351856, + "grad_norm": 2.339515209197998, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6740713119506836, + "num_tokens": 74231038.0, + "step": 2883 + }, + { + "epoch": 0.31671425433779926, + "grad_norm": 2.2937209606170654, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7178943753242493, + "num_tokens": 74254654.0, + "step": 2884 + }, + { + "epoch": 0.3168240720404129, + "grad_norm": 2.3767311573028564, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7205144762992859, + "num_tokens": 74276704.0, + "step": 2885 + }, + { + "epoch": 0.31693388974302655, + "grad_norm": 2.039421796798706, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6933689713478088, + "num_tokens": 74308059.0, + "step": 2886 + }, + { + "epoch": 0.31704370744564025, + "grad_norm": 2.2587127685546875, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7010335922241211, + "num_tokens": 74333528.0, + "step": 2887 + }, + { + "epoch": 0.3171535251482539, + "grad_norm": 2.2880139350891113, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7101103067398071, + "num_tokens": 74358891.0, + "step": 2888 + }, + { + "epoch": 0.31726334285086755, + "grad_norm": 2.3160688877105713, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6901295781135559, + "num_tokens": 74382578.0, + "step": 2889 + }, + { + "epoch": 0.31737316055348125, + "grad_norm": 2.016439437866211, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6940237283706665, + "num_tokens": 74413316.0, + "step": 2890 + }, + { + "epoch": 0.3174829782560949, + "grad_norm": 2.172549247741699, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6994611620903015, + "num_tokens": 74439603.0, + "step": 2891 + }, + { + "epoch": 0.31759279595870854, + "grad_norm": 2.64285945892334, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7058792114257812, + "num_tokens": 74459904.0, + "step": 2892 + }, + { + "epoch": 0.3177026136613222, + "grad_norm": 2.290428876876831, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6857333779335022, + "num_tokens": 74488314.0, + "step": 2893 + }, + { + "epoch": 0.3178124313639359, + "grad_norm": 2.1378445625305176, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7063778042793274, + "num_tokens": 74514166.0, + "step": 2894 + }, + { + "epoch": 0.31792224906654953, + "grad_norm": 2.3852524757385254, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.6866568326950073, + "num_tokens": 74540406.0, + "step": 2895 + }, + { + "epoch": 0.3180320667691632, + "grad_norm": 2.1440441608428955, + "learning_rate": 1e-06, + "loss": 1.1147, + "mean_token_accuracy": 0.6775168180465698, + "num_tokens": 74570266.0, + "step": 2896 + }, + { + "epoch": 0.3181418844717768, + "grad_norm": 2.0923993587493896, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6845740079879761, + "num_tokens": 74598211.0, + "step": 2897 + }, + { + "epoch": 0.3182517021743905, + "grad_norm": 2.2619733810424805, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.694995641708374, + "num_tokens": 74623156.0, + "step": 2898 + }, + { + "epoch": 0.3183615198770042, + "grad_norm": 2.20802903175354, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7173553705215454, + "num_tokens": 74648853.0, + "step": 2899 + }, + { + "epoch": 0.3184713375796178, + "grad_norm": 2.3991267681121826, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7035810947418213, + "num_tokens": 74671878.0, + "step": 2900 + }, + { + "epoch": 0.3185811552822315, + "grad_norm": 2.2251718044281006, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7077882885932922, + "num_tokens": 74698962.0, + "step": 2901 + }, + { + "epoch": 0.31869097298484517, + "grad_norm": 2.2111098766326904, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6873540878295898, + "num_tokens": 74726056.0, + "step": 2902 + }, + { + "epoch": 0.3188007906874588, + "grad_norm": 2.2153847217559814, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6951978206634521, + "num_tokens": 74752635.0, + "step": 2903 + }, + { + "epoch": 0.31891060839007246, + "grad_norm": 2.951892137527466, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6880286335945129, + "num_tokens": 74770510.0, + "step": 2904 + }, + { + "epoch": 0.31902042609268616, + "grad_norm": 2.441988706588745, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7011441588401794, + "num_tokens": 74793356.0, + "step": 2905 + }, + { + "epoch": 0.3191302437952998, + "grad_norm": 2.03029203414917, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6969797611236572, + "num_tokens": 74821631.0, + "step": 2906 + }, + { + "epoch": 0.31924006149791345, + "grad_norm": 2.218076467514038, + "learning_rate": 1e-06, + "loss": 1.1088, + "mean_token_accuracy": 0.6645779013633728, + "num_tokens": 74850395.0, + "step": 2907 + }, + { + "epoch": 0.3193498792005271, + "grad_norm": 2.3211092948913574, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7224814891815186, + "num_tokens": 74874437.0, + "step": 2908 + }, + { + "epoch": 0.3194596969031408, + "grad_norm": 2.4591691493988037, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7039717435836792, + "num_tokens": 74896235.0, + "step": 2909 + }, + { + "epoch": 0.31956951460575445, + "grad_norm": 2.3959648609161377, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6992701292037964, + "num_tokens": 74919717.0, + "step": 2910 + }, + { + "epoch": 0.3196793323083681, + "grad_norm": 2.223128318786621, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7141338586807251, + "num_tokens": 74942848.0, + "step": 2911 + }, + { + "epoch": 0.3197891500109818, + "grad_norm": 2.388551950454712, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7076728343963623, + "num_tokens": 74965921.0, + "step": 2912 + }, + { + "epoch": 0.31989896771359544, + "grad_norm": 2.2515265941619873, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7074803113937378, + "num_tokens": 74990866.0, + "step": 2913 + }, + { + "epoch": 0.3200087854162091, + "grad_norm": 2.2331953048706055, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6992011070251465, + "num_tokens": 75019039.0, + "step": 2914 + }, + { + "epoch": 0.32011860311882273, + "grad_norm": 2.0272467136383057, + "learning_rate": 1e-06, + "loss": 1.0925, + "mean_token_accuracy": 0.6733068227767944, + "num_tokens": 75048836.0, + "step": 2915 + }, + { + "epoch": 0.32022842082143643, + "grad_norm": 2.5064797401428223, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7175956964492798, + "num_tokens": 75069403.0, + "step": 2916 + }, + { + "epoch": 0.3203382385240501, + "grad_norm": 2.337606906890869, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7072436809539795, + "num_tokens": 75093109.0, + "step": 2917 + }, + { + "epoch": 0.3204480562266637, + "grad_norm": 2.1389408111572266, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7042241096496582, + "num_tokens": 75120665.0, + "step": 2918 + }, + { + "epoch": 0.3205578739292774, + "grad_norm": 2.02546763420105, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7212317585945129, + "num_tokens": 75147280.0, + "step": 2919 + }, + { + "epoch": 0.32066769163189107, + "grad_norm": 2.1814322471618652, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6901750564575195, + "num_tokens": 75173023.0, + "step": 2920 + }, + { + "epoch": 0.3207775093345047, + "grad_norm": 2.467989444732666, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6993663311004639, + "num_tokens": 75194084.0, + "step": 2921 + }, + { + "epoch": 0.32088732703711836, + "grad_norm": 2.0721960067749023, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6979600191116333, + "num_tokens": 75223613.0, + "step": 2922 + }, + { + "epoch": 0.32099714473973207, + "grad_norm": 1.9954339265823364, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6953255534172058, + "num_tokens": 75255293.0, + "step": 2923 + }, + { + "epoch": 0.3211069624423457, + "grad_norm": 2.190035581588745, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6862605810165405, + "num_tokens": 75282183.0, + "step": 2924 + }, + { + "epoch": 0.32121678014495936, + "grad_norm": 2.0601894855499268, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7049479484558105, + "num_tokens": 75311428.0, + "step": 2925 + }, + { + "epoch": 0.321326597847573, + "grad_norm": 2.3100757598876953, + "learning_rate": 1e-06, + "loss": 1.1245, + "mean_token_accuracy": 0.6681163311004639, + "num_tokens": 75339102.0, + "step": 2926 + }, + { + "epoch": 0.3214364155501867, + "grad_norm": 2.4964795112609863, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.709030032157898, + "num_tokens": 75359380.0, + "step": 2927 + }, + { + "epoch": 0.32154623325280035, + "grad_norm": 1.9921460151672363, + "learning_rate": 1e-06, + "loss": 1.1143, + "mean_token_accuracy": 0.6703874468803406, + "num_tokens": 75394187.0, + "step": 2928 + }, + { + "epoch": 0.321656050955414, + "grad_norm": 2.4081180095672607, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7267824411392212, + "num_tokens": 75417534.0, + "step": 2929 + }, + { + "epoch": 0.3217658686580277, + "grad_norm": 2.2950925827026367, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7051141858100891, + "num_tokens": 75442157.0, + "step": 2930 + }, + { + "epoch": 0.32187568636064134, + "grad_norm": 2.2954468727111816, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.688071608543396, + "num_tokens": 75466642.0, + "step": 2931 + }, + { + "epoch": 0.321985504063255, + "grad_norm": 2.435482978820801, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6989454030990601, + "num_tokens": 75487229.0, + "step": 2932 + }, + { + "epoch": 0.32209532176586864, + "grad_norm": 2.576237678527832, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.705889105796814, + "num_tokens": 75506440.0, + "step": 2933 + }, + { + "epoch": 0.32220513946848234, + "grad_norm": 2.2598397731781006, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7089253067970276, + "num_tokens": 75530019.0, + "step": 2934 + }, + { + "epoch": 0.322314957171096, + "grad_norm": 2.458796262741089, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7113604545593262, + "num_tokens": 75551030.0, + "step": 2935 + }, + { + "epoch": 0.32242477487370963, + "grad_norm": 2.2139201164245605, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7007842063903809, + "num_tokens": 75578901.0, + "step": 2936 + }, + { + "epoch": 0.3225345925763233, + "grad_norm": 1.9977660179138184, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6899734139442444, + "num_tokens": 75612917.0, + "step": 2937 + }, + { + "epoch": 0.322644410278937, + "grad_norm": 2.128020763397217, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6942884922027588, + "num_tokens": 75640136.0, + "step": 2938 + }, + { + "epoch": 0.3227542279815506, + "grad_norm": 2.175605297088623, + "learning_rate": 1e-06, + "loss": 1.0612, + "mean_token_accuracy": 0.6827856302261353, + "num_tokens": 75669857.0, + "step": 2939 + }, + { + "epoch": 0.32286404568416427, + "grad_norm": 2.208677291870117, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7190201282501221, + "num_tokens": 75694102.0, + "step": 2940 + }, + { + "epoch": 0.32297386338677797, + "grad_norm": 2.383643627166748, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7133774757385254, + "num_tokens": 75716926.0, + "step": 2941 + }, + { + "epoch": 0.3230836810893916, + "grad_norm": 2.1070284843444824, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6859890818595886, + "num_tokens": 75745203.0, + "step": 2942 + }, + { + "epoch": 0.32319349879200526, + "grad_norm": 2.1358561515808105, + "learning_rate": 1e-06, + "loss": 1.1228, + "mean_token_accuracy": 0.6643242835998535, + "num_tokens": 75776336.0, + "step": 2943 + }, + { + "epoch": 0.3233033164946189, + "grad_norm": 2.517322301864624, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7072723507881165, + "num_tokens": 75797655.0, + "step": 2944 + }, + { + "epoch": 0.3234131341972326, + "grad_norm": 2.494489908218384, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6960948705673218, + "num_tokens": 75819285.0, + "step": 2945 + }, + { + "epoch": 0.32352295189984626, + "grad_norm": 2.117398262023926, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6914926767349243, + "num_tokens": 75844960.0, + "step": 2946 + }, + { + "epoch": 0.3236327696024599, + "grad_norm": 2.1357009410858154, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6841225624084473, + "num_tokens": 75875466.0, + "step": 2947 + }, + { + "epoch": 0.3237425873050736, + "grad_norm": 2.596888780593872, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7033576965332031, + "num_tokens": 75895315.0, + "step": 2948 + }, + { + "epoch": 0.32385240500768725, + "grad_norm": 2.4657468795776367, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7024712562561035, + "num_tokens": 75917494.0, + "step": 2949 + }, + { + "epoch": 0.3239622227103009, + "grad_norm": 2.2836334705352783, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6907482743263245, + "num_tokens": 75942413.0, + "step": 2950 + }, + { + "epoch": 0.32407204041291454, + "grad_norm": 2.5268914699554443, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6854791641235352, + "num_tokens": 75963228.0, + "step": 2951 + }, + { + "epoch": 0.32418185811552824, + "grad_norm": 2.579662561416626, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6962916254997253, + "num_tokens": 75983712.0, + "step": 2952 + }, + { + "epoch": 0.3242916758181419, + "grad_norm": 2.1074600219726562, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7074177265167236, + "num_tokens": 76013460.0, + "step": 2953 + }, + { + "epoch": 0.32440149352075554, + "grad_norm": 2.202770233154297, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6808485984802246, + "num_tokens": 76040865.0, + "step": 2954 + }, + { + "epoch": 0.3245113112233692, + "grad_norm": 2.098881959915161, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7063722014427185, + "num_tokens": 76070562.0, + "step": 2955 + }, + { + "epoch": 0.3246211289259829, + "grad_norm": 2.11757230758667, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7091212272644043, + "num_tokens": 76097769.0, + "step": 2956 + }, + { + "epoch": 0.32473094662859653, + "grad_norm": 1.9442086219787598, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6891811490058899, + "num_tokens": 76127432.0, + "step": 2957 + }, + { + "epoch": 0.3248407643312102, + "grad_norm": 2.0961499214172363, + "learning_rate": 1e-06, + "loss": 1.0643, + "mean_token_accuracy": 0.6844760775566101, + "num_tokens": 76157377.0, + "step": 2958 + }, + { + "epoch": 0.3249505820338239, + "grad_norm": 2.511002779006958, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7045687437057495, + "num_tokens": 76177219.0, + "step": 2959 + }, + { + "epoch": 0.3250603997364375, + "grad_norm": 2.3031413555145264, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7064358592033386, + "num_tokens": 76201051.0, + "step": 2960 + }, + { + "epoch": 0.32517021743905117, + "grad_norm": 2.583979845046997, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7031442523002625, + "num_tokens": 76221000.0, + "step": 2961 + }, + { + "epoch": 0.3252800351416648, + "grad_norm": 1.9631270170211792, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6969531774520874, + "num_tokens": 76250675.0, + "step": 2962 + }, + { + "epoch": 0.3253898528442785, + "grad_norm": 2.12489914894104, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6989345550537109, + "num_tokens": 76277738.0, + "step": 2963 + }, + { + "epoch": 0.32549967054689216, + "grad_norm": 2.071577310562134, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6887325644493103, + "num_tokens": 76305999.0, + "step": 2964 + }, + { + "epoch": 0.3256094882495058, + "grad_norm": 2.130553722381592, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6805358529090881, + "num_tokens": 76334571.0, + "step": 2965 + }, + { + "epoch": 0.3257193059521195, + "grad_norm": 2.0347750186920166, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6870698928833008, + "num_tokens": 76365725.0, + "step": 2966 + }, + { + "epoch": 0.32582912365473315, + "grad_norm": 2.101024866104126, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7264111042022705, + "num_tokens": 76392046.0, + "step": 2967 + }, + { + "epoch": 0.3259389413573468, + "grad_norm": 2.201247215270996, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7398857474327087, + "num_tokens": 76417286.0, + "step": 2968 + }, + { + "epoch": 0.32604875905996045, + "grad_norm": 2.2473182678222656, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7159643769264221, + "num_tokens": 76441833.0, + "step": 2969 + }, + { + "epoch": 0.32615857676257415, + "grad_norm": 2.170057773590088, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.705541729927063, + "num_tokens": 76469005.0, + "step": 2970 + }, + { + "epoch": 0.3262683944651878, + "grad_norm": 2.2128472328186035, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.724761962890625, + "num_tokens": 76492367.0, + "step": 2971 + }, + { + "epoch": 0.32637821216780144, + "grad_norm": 2.4860026836395264, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7001826763153076, + "num_tokens": 76513918.0, + "step": 2972 + }, + { + "epoch": 0.3264880298704151, + "grad_norm": 2.1218197345733643, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7089976668357849, + "num_tokens": 76540289.0, + "step": 2973 + }, + { + "epoch": 0.3265978475730288, + "grad_norm": 2.324983596801758, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6957789063453674, + "num_tokens": 76562101.0, + "step": 2974 + }, + { + "epoch": 0.32670766527564243, + "grad_norm": 2.109701633453369, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6969258189201355, + "num_tokens": 76589119.0, + "step": 2975 + }, + { + "epoch": 0.3268174829782561, + "grad_norm": 2.7499051094055176, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6915702819824219, + "num_tokens": 76609351.0, + "step": 2976 + }, + { + "epoch": 0.3269273006808698, + "grad_norm": 2.4277801513671875, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6927927136421204, + "num_tokens": 76632241.0, + "step": 2977 + }, + { + "epoch": 0.3270371183834834, + "grad_norm": 2.2650110721588135, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6981070637702942, + "num_tokens": 76658944.0, + "step": 2978 + }, + { + "epoch": 0.3271469360860971, + "grad_norm": 2.2473933696746826, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7016856074333191, + "num_tokens": 76681538.0, + "step": 2979 + }, + { + "epoch": 0.3272567537887107, + "grad_norm": 2.4088916778564453, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6891260147094727, + "num_tokens": 76704334.0, + "step": 2980 + }, + { + "epoch": 0.3273665714913244, + "grad_norm": 2.1716501712799072, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6966930627822876, + "num_tokens": 76729240.0, + "step": 2981 + }, + { + "epoch": 0.32747638919393807, + "grad_norm": 2.1804091930389404, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.7018895149230957, + "num_tokens": 76755681.0, + "step": 2982 + }, + { + "epoch": 0.3275862068965517, + "grad_norm": 1.9902430772781372, + "learning_rate": 1e-06, + "loss": 1.0602, + "mean_token_accuracy": 0.6805064678192139, + "num_tokens": 76788488.0, + "step": 2983 + }, + { + "epoch": 0.32769602459916536, + "grad_norm": 2.310570478439331, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6839884519577026, + "num_tokens": 76812408.0, + "step": 2984 + }, + { + "epoch": 0.32780584230177906, + "grad_norm": 2.2466304302215576, + "learning_rate": 1e-06, + "loss": 1.0614, + "mean_token_accuracy": 0.6857950687408447, + "num_tokens": 76837160.0, + "step": 2985 + }, + { + "epoch": 0.3279156600043927, + "grad_norm": 2.2077150344848633, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7059410810470581, + "num_tokens": 76861166.0, + "step": 2986 + }, + { + "epoch": 0.32802547770700635, + "grad_norm": 1.9704855680465698, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6848052740097046, + "num_tokens": 76894243.0, + "step": 2987 + }, + { + "epoch": 0.32813529540962005, + "grad_norm": 2.085286855697632, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7073773145675659, + "num_tokens": 76920494.0, + "step": 2988 + }, + { + "epoch": 0.3282451131122337, + "grad_norm": 2.4418654441833496, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6867192983627319, + "num_tokens": 76943201.0, + "step": 2989 + }, + { + "epoch": 0.32835493081484735, + "grad_norm": 2.1933133602142334, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6864393353462219, + "num_tokens": 76967613.0, + "step": 2990 + }, + { + "epoch": 0.328464748517461, + "grad_norm": 2.3047025203704834, + "learning_rate": 1e-06, + "loss": 1.085, + "mean_token_accuracy": 0.6779390573501587, + "num_tokens": 76991577.0, + "step": 2991 + }, + { + "epoch": 0.3285745662200747, + "grad_norm": 2.2208125591278076, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.70683753490448, + "num_tokens": 77016074.0, + "step": 2992 + }, + { + "epoch": 0.32868438392268834, + "grad_norm": 2.330582618713379, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7259640097618103, + "num_tokens": 77036673.0, + "step": 2993 + }, + { + "epoch": 0.328794201625302, + "grad_norm": 2.232569456100464, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6794829368591309, + "num_tokens": 77064384.0, + "step": 2994 + }, + { + "epoch": 0.3289040193279157, + "grad_norm": 2.22591495513916, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7074872255325317, + "num_tokens": 77089798.0, + "step": 2995 + }, + { + "epoch": 0.32901383703052933, + "grad_norm": 2.12077260017395, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.691868007183075, + "num_tokens": 77117133.0, + "step": 2996 + }, + { + "epoch": 0.329123654733143, + "grad_norm": 2.1586310863494873, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7025091648101807, + "num_tokens": 77143950.0, + "step": 2997 + }, + { + "epoch": 0.3292334724357566, + "grad_norm": 2.3248660564422607, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.689061164855957, + "num_tokens": 77167224.0, + "step": 2998 + }, + { + "epoch": 0.3293432901383703, + "grad_norm": 2.516223907470703, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6805444359779358, + "num_tokens": 77192887.0, + "step": 2999 + }, + { + "epoch": 0.32945310784098397, + "grad_norm": 2.7695348262786865, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7096745371818542, + "num_tokens": 77211633.0, + "step": 3000 + }, + { + "epoch": 0.3295629255435976, + "grad_norm": 2.297379970550537, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7069534659385681, + "num_tokens": 77235881.0, + "step": 3001 + }, + { + "epoch": 0.32967274324621126, + "grad_norm": 2.0291528701782227, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6876394748687744, + "num_tokens": 77265711.0, + "step": 3002 + }, + { + "epoch": 0.32978256094882497, + "grad_norm": 2.2911367416381836, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7129002213478088, + "num_tokens": 77291682.0, + "step": 3003 + }, + { + "epoch": 0.3298923786514386, + "grad_norm": 2.283848524093628, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6882578134536743, + "num_tokens": 77316198.0, + "step": 3004 + }, + { + "epoch": 0.33000219635405226, + "grad_norm": 2.0901966094970703, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6924880146980286, + "num_tokens": 77343786.0, + "step": 3005 + }, + { + "epoch": 0.33011201405666596, + "grad_norm": 2.1354942321777344, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6941403150558472, + "num_tokens": 77372548.0, + "step": 3006 + }, + { + "epoch": 0.3302218317592796, + "grad_norm": 2.13795804977417, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6910995244979858, + "num_tokens": 77398843.0, + "step": 3007 + }, + { + "epoch": 0.33033164946189325, + "grad_norm": 2.190375328063965, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7111220955848694, + "num_tokens": 77426016.0, + "step": 3008 + }, + { + "epoch": 0.3304414671645069, + "grad_norm": 2.4270071983337402, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7045155167579651, + "num_tokens": 77448123.0, + "step": 3009 + }, + { + "epoch": 0.3305512848671206, + "grad_norm": 2.065638542175293, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6958519220352173, + "num_tokens": 77476744.0, + "step": 3010 + }, + { + "epoch": 0.33066110256973424, + "grad_norm": 2.135472297668457, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6929070949554443, + "num_tokens": 77503776.0, + "step": 3011 + }, + { + "epoch": 0.3307709202723479, + "grad_norm": 2.1997087001800537, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.695019006729126, + "num_tokens": 77531288.0, + "step": 3012 + }, + { + "epoch": 0.33088073797496154, + "grad_norm": 2.477799415588379, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6931942701339722, + "num_tokens": 77551281.0, + "step": 3013 + }, + { + "epoch": 0.33099055567757524, + "grad_norm": 2.449658155441284, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7040502429008484, + "num_tokens": 77572930.0, + "step": 3014 + }, + { + "epoch": 0.3311003733801889, + "grad_norm": 2.077326774597168, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.722075343132019, + "num_tokens": 77600115.0, + "step": 3015 + }, + { + "epoch": 0.33121019108280253, + "grad_norm": 2.248241424560547, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6958094835281372, + "num_tokens": 77626404.0, + "step": 3016 + }, + { + "epoch": 0.33132000878541623, + "grad_norm": 2.4725537300109863, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7232872843742371, + "num_tokens": 77649021.0, + "step": 3017 + }, + { + "epoch": 0.3314298264880299, + "grad_norm": 2.228346109390259, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7019941806793213, + "num_tokens": 77675175.0, + "step": 3018 + }, + { + "epoch": 0.3315396441906435, + "grad_norm": 2.1499834060668945, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.710719883441925, + "num_tokens": 77702860.0, + "step": 3019 + }, + { + "epoch": 0.33164946189325717, + "grad_norm": 2.2102904319763184, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6940038204193115, + "num_tokens": 77730701.0, + "step": 3020 + }, + { + "epoch": 0.33175927959587087, + "grad_norm": 2.126880645751953, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6898680925369263, + "num_tokens": 77758125.0, + "step": 3021 + }, + { + "epoch": 0.3318690972984845, + "grad_norm": 2.2530550956726074, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7127259969711304, + "num_tokens": 77783290.0, + "step": 3022 + }, + { + "epoch": 0.33197891500109816, + "grad_norm": 2.2547130584716797, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6893855333328247, + "num_tokens": 77808719.0, + "step": 3023 + }, + { + "epoch": 0.33208873270371186, + "grad_norm": 2.3562707901000977, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.703576922416687, + "num_tokens": 77834244.0, + "step": 3024 + }, + { + "epoch": 0.3321985504063255, + "grad_norm": 2.1678271293640137, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.686364471912384, + "num_tokens": 77862587.0, + "step": 3025 + }, + { + "epoch": 0.33230836810893916, + "grad_norm": 2.0455610752105713, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.70162433385849, + "num_tokens": 77889374.0, + "step": 3026 + }, + { + "epoch": 0.3324181858115528, + "grad_norm": 2.246509552001953, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.699714720249176, + "num_tokens": 77914861.0, + "step": 3027 + }, + { + "epoch": 0.3325280035141665, + "grad_norm": 2.1558120250701904, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7090031504631042, + "num_tokens": 77941592.0, + "step": 3028 + }, + { + "epoch": 0.33263782121678015, + "grad_norm": 1.8837971687316895, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7024425268173218, + "num_tokens": 77975526.0, + "step": 3029 + }, + { + "epoch": 0.3327476389193938, + "grad_norm": 2.192232370376587, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.6856247782707214, + "num_tokens": 78002590.0, + "step": 3030 + }, + { + "epoch": 0.33285745662200744, + "grad_norm": 2.3091375827789307, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6997714638710022, + "num_tokens": 78026072.0, + "step": 3031 + }, + { + "epoch": 0.33296727432462114, + "grad_norm": 2.151724338531494, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6828649044036865, + "num_tokens": 78053535.0, + "step": 3032 + }, + { + "epoch": 0.3330770920272348, + "grad_norm": 2.0458028316497803, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.7016386389732361, + "num_tokens": 78083784.0, + "step": 3033 + }, + { + "epoch": 0.33318690972984844, + "grad_norm": 2.2961392402648926, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6990057229995728, + "num_tokens": 78108987.0, + "step": 3034 + }, + { + "epoch": 0.33329672743246214, + "grad_norm": 2.097405433654785, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6949112415313721, + "num_tokens": 78137163.0, + "step": 3035 + }, + { + "epoch": 0.3334065451350758, + "grad_norm": 2.1370718479156494, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7109229564666748, + "num_tokens": 78165473.0, + "step": 3036 + }, + { + "epoch": 0.33351636283768943, + "grad_norm": 2.261960983276367, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7052184343338013, + "num_tokens": 78190180.0, + "step": 3037 + }, + { + "epoch": 0.3336261805403031, + "grad_norm": 2.4332809448242188, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6994186043739319, + "num_tokens": 78213206.0, + "step": 3038 + }, + { + "epoch": 0.3337359982429168, + "grad_norm": 2.447753429412842, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6844006776809692, + "num_tokens": 78236026.0, + "step": 3039 + }, + { + "epoch": 0.3338458159455304, + "grad_norm": 2.201127767562866, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7012381553649902, + "num_tokens": 78266220.0, + "step": 3040 + }, + { + "epoch": 0.33395563364814407, + "grad_norm": 1.9977585077285767, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6994187831878662, + "num_tokens": 78298087.0, + "step": 3041 + }, + { + "epoch": 0.33406545135075777, + "grad_norm": 2.095155715942383, + "learning_rate": 1e-06, + "loss": 1.0997, + "mean_token_accuracy": 0.6704343557357788, + "num_tokens": 78326217.0, + "step": 3042 + }, + { + "epoch": 0.3341752690533714, + "grad_norm": 2.2732138633728027, + "learning_rate": 1e-06, + "loss": 1.0618, + "mean_token_accuracy": 0.685581386089325, + "num_tokens": 78354016.0, + "step": 3043 + }, + { + "epoch": 0.33428508675598506, + "grad_norm": 2.024226188659668, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6873401403427124, + "num_tokens": 78385942.0, + "step": 3044 + }, + { + "epoch": 0.3343949044585987, + "grad_norm": 2.3531715869903564, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6974771022796631, + "num_tokens": 78409283.0, + "step": 3045 + }, + { + "epoch": 0.3345047221612124, + "grad_norm": 2.655189275741577, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6990671157836914, + "num_tokens": 78428557.0, + "step": 3046 + }, + { + "epoch": 0.33461453986382605, + "grad_norm": 2.2512624263763428, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6918838024139404, + "num_tokens": 78457686.0, + "step": 3047 + }, + { + "epoch": 0.3347243575664397, + "grad_norm": 2.418978691101074, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7256900072097778, + "num_tokens": 78480015.0, + "step": 3048 + }, + { + "epoch": 0.33483417526905335, + "grad_norm": 2.201702356338501, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6761542558670044, + "num_tokens": 78504999.0, + "step": 3049 + }, + { + "epoch": 0.33494399297166705, + "grad_norm": 1.9909443855285645, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6895462274551392, + "num_tokens": 78536499.0, + "step": 3050 + }, + { + "epoch": 0.3350538106742807, + "grad_norm": 2.396090269088745, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7257421612739563, + "num_tokens": 78558025.0, + "step": 3051 + }, + { + "epoch": 0.33516362837689434, + "grad_norm": 2.318650722503662, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6899185180664062, + "num_tokens": 78581767.0, + "step": 3052 + }, + { + "epoch": 0.33527344607950804, + "grad_norm": 2.1461615562438965, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6845420598983765, + "num_tokens": 78609670.0, + "step": 3053 + }, + { + "epoch": 0.3353832637821217, + "grad_norm": 2.2853171825408936, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6979374885559082, + "num_tokens": 78632841.0, + "step": 3054 + }, + { + "epoch": 0.33549308148473533, + "grad_norm": 2.1166234016418457, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7123514413833618, + "num_tokens": 78659157.0, + "step": 3055 + }, + { + "epoch": 0.335602899187349, + "grad_norm": 2.4342339038848877, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7199156284332275, + "num_tokens": 78684257.0, + "step": 3056 + }, + { + "epoch": 0.3357127168899627, + "grad_norm": 2.14695405960083, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7145137786865234, + "num_tokens": 78711359.0, + "step": 3057 + }, + { + "epoch": 0.3358225345925763, + "grad_norm": 2.0428528785705566, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6912757158279419, + "num_tokens": 78741527.0, + "step": 3058 + }, + { + "epoch": 0.33593235229519, + "grad_norm": 2.332512140274048, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7093174457550049, + "num_tokens": 78764751.0, + "step": 3059 + }, + { + "epoch": 0.3360421699978036, + "grad_norm": 2.2303240299224854, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6925684809684753, + "num_tokens": 78789438.0, + "step": 3060 + }, + { + "epoch": 0.3361519877004173, + "grad_norm": 2.564432382583618, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7170583009719849, + "num_tokens": 78811033.0, + "step": 3061 + }, + { + "epoch": 0.33626180540303097, + "grad_norm": 1.9847348928451538, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7065289616584778, + "num_tokens": 78840440.0, + "step": 3062 + }, + { + "epoch": 0.3363716231056446, + "grad_norm": 2.333299398422241, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6855299472808838, + "num_tokens": 78864884.0, + "step": 3063 + }, + { + "epoch": 0.3364814408082583, + "grad_norm": 2.3034555912017822, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6859515905380249, + "num_tokens": 78889274.0, + "step": 3064 + }, + { + "epoch": 0.33659125851087196, + "grad_norm": 1.947504997253418, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6838032603263855, + "num_tokens": 78920069.0, + "step": 3065 + }, + { + "epoch": 0.3367010762134856, + "grad_norm": 2.0140140056610107, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.696887731552124, + "num_tokens": 78949912.0, + "step": 3066 + }, + { + "epoch": 0.33681089391609925, + "grad_norm": 2.2093727588653564, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.687288224697113, + "num_tokens": 78974824.0, + "step": 3067 + }, + { + "epoch": 0.33692071161871295, + "grad_norm": 2.391880512237549, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7098121643066406, + "num_tokens": 78996146.0, + "step": 3068 + }, + { + "epoch": 0.3370305293213266, + "grad_norm": 2.1849629878997803, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7190295457839966, + "num_tokens": 79022799.0, + "step": 3069 + }, + { + "epoch": 0.33714034702394025, + "grad_norm": 2.300549268722534, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7252835631370544, + "num_tokens": 79045477.0, + "step": 3070 + }, + { + "epoch": 0.33725016472655395, + "grad_norm": 2.2512853145599365, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.698135256767273, + "num_tokens": 79071179.0, + "step": 3071 + }, + { + "epoch": 0.3373599824291676, + "grad_norm": 2.1030561923980713, + "learning_rate": 1e-06, + "loss": 1.1026, + "mean_token_accuracy": 0.6686739921569824, + "num_tokens": 79101471.0, + "step": 3072 + }, + { + "epoch": 0.33746980013178124, + "grad_norm": 2.1769142150878906, + "learning_rate": 1e-06, + "loss": 1.1007, + "mean_token_accuracy": 0.6781481504440308, + "num_tokens": 79128743.0, + "step": 3073 + }, + { + "epoch": 0.3375796178343949, + "grad_norm": 2.7552661895751953, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7183712720870972, + "num_tokens": 79149034.0, + "step": 3074 + }, + { + "epoch": 0.3376894355370086, + "grad_norm": 1.9778707027435303, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.718320906162262, + "num_tokens": 79178367.0, + "step": 3075 + }, + { + "epoch": 0.33779925323962223, + "grad_norm": 2.08917498588562, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6943902969360352, + "num_tokens": 79205846.0, + "step": 3076 + }, + { + "epoch": 0.3379090709422359, + "grad_norm": 2.297759771347046, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6931777596473694, + "num_tokens": 79234183.0, + "step": 3077 + }, + { + "epoch": 0.3380188886448495, + "grad_norm": 2.216153860092163, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.697537362575531, + "num_tokens": 79260907.0, + "step": 3078 + }, + { + "epoch": 0.3381287063474632, + "grad_norm": 2.2523128986358643, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7119475603103638, + "num_tokens": 79284039.0, + "step": 3079 + }, + { + "epoch": 0.33823852405007687, + "grad_norm": 2.154473066329956, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6988047361373901, + "num_tokens": 79311337.0, + "step": 3080 + }, + { + "epoch": 0.3383483417526905, + "grad_norm": 2.233701705932617, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6893155574798584, + "num_tokens": 79336979.0, + "step": 3081 + }, + { + "epoch": 0.3384581594553042, + "grad_norm": 2.5622127056121826, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6946466565132141, + "num_tokens": 79356955.0, + "step": 3082 + }, + { + "epoch": 0.33856797715791787, + "grad_norm": 2.083977222442627, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7137223482131958, + "num_tokens": 79387203.0, + "step": 3083 + }, + { + "epoch": 0.3386777948605315, + "grad_norm": 1.9922211170196533, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7022688388824463, + "num_tokens": 79418835.0, + "step": 3084 + }, + { + "epoch": 0.33878761256314516, + "grad_norm": 2.27351975440979, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7058593034744263, + "num_tokens": 79441851.0, + "step": 3085 + }, + { + "epoch": 0.33889743026575886, + "grad_norm": 2.1495907306671143, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7008265256881714, + "num_tokens": 79468930.0, + "step": 3086 + }, + { + "epoch": 0.3390072479683725, + "grad_norm": 2.437429666519165, + "learning_rate": 1e-06, + "loss": 1.0778, + "mean_token_accuracy": 0.6825525760650635, + "num_tokens": 79491569.0, + "step": 3087 + }, + { + "epoch": 0.33911706567098615, + "grad_norm": 2.2507200241088867, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7190104722976685, + "num_tokens": 79517222.0, + "step": 3088 + }, + { + "epoch": 0.3392268833735998, + "grad_norm": 2.222494602203369, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7047463655471802, + "num_tokens": 79543358.0, + "step": 3089 + }, + { + "epoch": 0.3393367010762135, + "grad_norm": 2.3979201316833496, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6918530464172363, + "num_tokens": 79564736.0, + "step": 3090 + }, + { + "epoch": 0.33944651877882714, + "grad_norm": 2.2290143966674805, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7155309915542603, + "num_tokens": 79590722.0, + "step": 3091 + }, + { + "epoch": 0.3395563364814408, + "grad_norm": 2.256732940673828, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6934224367141724, + "num_tokens": 79613994.0, + "step": 3092 + }, + { + "epoch": 0.3396661541840545, + "grad_norm": 2.2177155017852783, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6936739683151245, + "num_tokens": 79639623.0, + "step": 3093 + }, + { + "epoch": 0.33977597188666814, + "grad_norm": 2.076289415359497, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7091618776321411, + "num_tokens": 79666572.0, + "step": 3094 + }, + { + "epoch": 0.3398857895892818, + "grad_norm": 2.452760934829712, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7007230520248413, + "num_tokens": 79686947.0, + "step": 3095 + }, + { + "epoch": 0.33999560729189543, + "grad_norm": 2.255317449569702, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6972714066505432, + "num_tokens": 79711620.0, + "step": 3096 + }, + { + "epoch": 0.34010542499450913, + "grad_norm": 2.4665639400482178, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6964624524116516, + "num_tokens": 79735277.0, + "step": 3097 + }, + { + "epoch": 0.3402152426971228, + "grad_norm": 2.238001585006714, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6923283338546753, + "num_tokens": 79759972.0, + "step": 3098 + }, + { + "epoch": 0.3403250603997364, + "grad_norm": 1.968420386314392, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.70671546459198, + "num_tokens": 79793374.0, + "step": 3099 + }, + { + "epoch": 0.3404348781023501, + "grad_norm": 1.9875577688217163, + "learning_rate": 1e-06, + "loss": 1.1488, + "mean_token_accuracy": 0.6568185687065125, + "num_tokens": 79826182.0, + "step": 3100 + }, + { + "epoch": 0.34054469580496377, + "grad_norm": 2.2399957180023193, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7105449438095093, + "num_tokens": 79851758.0, + "step": 3101 + }, + { + "epoch": 0.3406545135075774, + "grad_norm": 2.3414130210876465, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6833580136299133, + "num_tokens": 79875070.0, + "step": 3102 + }, + { + "epoch": 0.34076433121019106, + "grad_norm": 2.278259754180908, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6928426027297974, + "num_tokens": 79900081.0, + "step": 3103 + }, + { + "epoch": 0.34087414891280476, + "grad_norm": 2.242384910583496, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6975804567337036, + "num_tokens": 79924198.0, + "step": 3104 + }, + { + "epoch": 0.3409839666154184, + "grad_norm": 2.239391326904297, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6952199339866638, + "num_tokens": 79949924.0, + "step": 3105 + }, + { + "epoch": 0.34109378431803206, + "grad_norm": 2.3812406063079834, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7065470218658447, + "num_tokens": 79972709.0, + "step": 3106 + }, + { + "epoch": 0.3412036020206457, + "grad_norm": 2.307608127593994, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6980304718017578, + "num_tokens": 79997325.0, + "step": 3107 + }, + { + "epoch": 0.3413134197232594, + "grad_norm": 2.3722610473632812, + "learning_rate": 1e-06, + "loss": 1.0775, + "mean_token_accuracy": 0.6771116852760315, + "num_tokens": 80024041.0, + "step": 3108 + }, + { + "epoch": 0.34142323742587305, + "grad_norm": 2.454359769821167, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6901567578315735, + "num_tokens": 80049836.0, + "step": 3109 + }, + { + "epoch": 0.3415330551284867, + "grad_norm": 2.0214791297912598, + "learning_rate": 1e-06, + "loss": 1.119, + "mean_token_accuracy": 0.666051983833313, + "num_tokens": 80081375.0, + "step": 3110 + }, + { + "epoch": 0.3416428728311004, + "grad_norm": 2.330911159515381, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.6699076294898987, + "num_tokens": 80105221.0, + "step": 3111 + }, + { + "epoch": 0.34175269053371404, + "grad_norm": 2.366802930831909, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7201747894287109, + "num_tokens": 80126631.0, + "step": 3112 + }, + { + "epoch": 0.3418625082363277, + "grad_norm": 2.151170253753662, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6881888508796692, + "num_tokens": 80155495.0, + "step": 3113 + }, + { + "epoch": 0.34197232593894134, + "grad_norm": 2.5617265701293945, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7245103120803833, + "num_tokens": 80174029.0, + "step": 3114 + }, + { + "epoch": 0.34208214364155504, + "grad_norm": 2.3213157653808594, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7101010084152222, + "num_tokens": 80195108.0, + "step": 3115 + }, + { + "epoch": 0.3421919613441687, + "grad_norm": 2.111637592315674, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6897284984588623, + "num_tokens": 80222251.0, + "step": 3116 + }, + { + "epoch": 0.34230177904678233, + "grad_norm": 2.6217195987701416, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6879346966743469, + "num_tokens": 80250177.0, + "step": 3117 + }, + { + "epoch": 0.34241159674939603, + "grad_norm": 2.1596226692199707, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7092511653900146, + "num_tokens": 80275516.0, + "step": 3118 + }, + { + "epoch": 0.3425214144520097, + "grad_norm": 2.1462671756744385, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.700485348701477, + "num_tokens": 80300831.0, + "step": 3119 + }, + { + "epoch": 0.3426312321546233, + "grad_norm": 2.177046298980713, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6808053851127625, + "num_tokens": 80327784.0, + "step": 3120 + }, + { + "epoch": 0.34274104985723697, + "grad_norm": 1.9653428792953491, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7077049016952515, + "num_tokens": 80358734.0, + "step": 3121 + }, + { + "epoch": 0.34285086755985067, + "grad_norm": 2.293901205062866, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6965917348861694, + "num_tokens": 80383908.0, + "step": 3122 + }, + { + "epoch": 0.3429606852624643, + "grad_norm": 1.9579664468765259, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6773665547370911, + "num_tokens": 80415417.0, + "step": 3123 + }, + { + "epoch": 0.34307050296507796, + "grad_norm": 2.3741812705993652, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.709218442440033, + "num_tokens": 80439163.0, + "step": 3124 + }, + { + "epoch": 0.3431803206676916, + "grad_norm": 2.009469985961914, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.691351056098938, + "num_tokens": 80470606.0, + "step": 3125 + }, + { + "epoch": 0.3432901383703053, + "grad_norm": 2.373591661453247, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6978676915168762, + "num_tokens": 80495076.0, + "step": 3126 + }, + { + "epoch": 0.34339995607291895, + "grad_norm": 2.574538469314575, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7099311351776123, + "num_tokens": 80515251.0, + "step": 3127 + }, + { + "epoch": 0.3435097737755326, + "grad_norm": 2.0818581581115723, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6954066753387451, + "num_tokens": 80544255.0, + "step": 3128 + }, + { + "epoch": 0.3436195914781463, + "grad_norm": 2.356931686401367, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.7017600536346436, + "num_tokens": 80567356.0, + "step": 3129 + }, + { + "epoch": 0.34372940918075995, + "grad_norm": 2.073762893676758, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6967902183532715, + "num_tokens": 80594850.0, + "step": 3130 + }, + { + "epoch": 0.3438392268833736, + "grad_norm": 2.247408628463745, + "learning_rate": 1e-06, + "loss": 1.0854, + "mean_token_accuracy": 0.676413893699646, + "num_tokens": 80622138.0, + "step": 3131 + }, + { + "epoch": 0.34394904458598724, + "grad_norm": 2.2422916889190674, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6826854944229126, + "num_tokens": 80648646.0, + "step": 3132 + }, + { + "epoch": 0.34405886228860094, + "grad_norm": 2.2766218185424805, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7026121616363525, + "num_tokens": 80672050.0, + "step": 3133 + }, + { + "epoch": 0.3441686799912146, + "grad_norm": 2.134610176086426, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7012326717376709, + "num_tokens": 80698680.0, + "step": 3134 + }, + { + "epoch": 0.34427849769382823, + "grad_norm": 2.4871857166290283, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7311094999313354, + "num_tokens": 80718122.0, + "step": 3135 + }, + { + "epoch": 0.3443883153964419, + "grad_norm": 2.3884687423706055, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6959091424942017, + "num_tokens": 80740989.0, + "step": 3136 + }, + { + "epoch": 0.3444981330990556, + "grad_norm": 2.579362154006958, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7020587921142578, + "num_tokens": 80762153.0, + "step": 3137 + }, + { + "epoch": 0.3446079508016692, + "grad_norm": 2.2671265602111816, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7045503854751587, + "num_tokens": 80786598.0, + "step": 3138 + }, + { + "epoch": 0.3447177685042829, + "grad_norm": 2.053805112838745, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6911923885345459, + "num_tokens": 80814919.0, + "step": 3139 + }, + { + "epoch": 0.3448275862068966, + "grad_norm": 2.0973598957061768, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7052668333053589, + "num_tokens": 80842954.0, + "step": 3140 + }, + { + "epoch": 0.3449374039095102, + "grad_norm": 2.3280065059661865, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6898603439331055, + "num_tokens": 80867586.0, + "step": 3141 + }, + { + "epoch": 0.34504722161212387, + "grad_norm": 2.166943311691284, + "learning_rate": 1e-06, + "loss": 1.107, + "mean_token_accuracy": 0.6645373106002808, + "num_tokens": 80895232.0, + "step": 3142 + }, + { + "epoch": 0.3451570393147375, + "grad_norm": 2.1173899173736572, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6865835785865784, + "num_tokens": 80922030.0, + "step": 3143 + }, + { + "epoch": 0.3452668570173512, + "grad_norm": 2.344503402709961, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7328653335571289, + "num_tokens": 80943702.0, + "step": 3144 + }, + { + "epoch": 0.34537667471996486, + "grad_norm": 2.4311344623565674, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7165915966033936, + "num_tokens": 80964851.0, + "step": 3145 + }, + { + "epoch": 0.3454864924225785, + "grad_norm": 2.2144322395324707, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7236400842666626, + "num_tokens": 80989635.0, + "step": 3146 + }, + { + "epoch": 0.3455963101251922, + "grad_norm": 2.1233909130096436, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7184339761734009, + "num_tokens": 81015697.0, + "step": 3147 + }, + { + "epoch": 0.34570612782780585, + "grad_norm": 2.5420472621917725, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7335953712463379, + "num_tokens": 81034915.0, + "step": 3148 + }, + { + "epoch": 0.3458159455304195, + "grad_norm": 2.1308627128601074, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6967530250549316, + "num_tokens": 81062459.0, + "step": 3149 + }, + { + "epoch": 0.34592576323303315, + "grad_norm": 2.2593882083892822, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7041338682174683, + "num_tokens": 81087618.0, + "step": 3150 + }, + { + "epoch": 0.34603558093564685, + "grad_norm": 2.374323606491089, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6982853412628174, + "num_tokens": 81111327.0, + "step": 3151 + }, + { + "epoch": 0.3461453986382605, + "grad_norm": 2.2655675411224365, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7016894817352295, + "num_tokens": 81137359.0, + "step": 3152 + }, + { + "epoch": 0.34625521634087414, + "grad_norm": 1.9893816709518433, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6910677552223206, + "num_tokens": 81167867.0, + "step": 3153 + }, + { + "epoch": 0.3463650340434878, + "grad_norm": 2.9412758350372314, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7185684442520142, + "num_tokens": 81190684.0, + "step": 3154 + }, + { + "epoch": 0.3464748517461015, + "grad_norm": 2.27022385597229, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7221885323524475, + "num_tokens": 81215645.0, + "step": 3155 + }, + { + "epoch": 0.34658466944871513, + "grad_norm": 2.2402360439300537, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6842011213302612, + "num_tokens": 81241147.0, + "step": 3156 + }, + { + "epoch": 0.3466944871513288, + "grad_norm": 1.9593448638916016, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7064827084541321, + "num_tokens": 81272273.0, + "step": 3157 + }, + { + "epoch": 0.3468043048539425, + "grad_norm": 2.5051045417785645, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6981481313705444, + "num_tokens": 81293183.0, + "step": 3158 + }, + { + "epoch": 0.3469141225565561, + "grad_norm": 2.0334532260894775, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6934791803359985, + "num_tokens": 81323872.0, + "step": 3159 + }, + { + "epoch": 0.34702394025916977, + "grad_norm": 2.020637035369873, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.6828970313072205, + "num_tokens": 81355008.0, + "step": 3160 + }, + { + "epoch": 0.3471337579617834, + "grad_norm": 2.10528302192688, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7038091421127319, + "num_tokens": 81383919.0, + "step": 3161 + }, + { + "epoch": 0.3472435756643971, + "grad_norm": 2.1495351791381836, + "learning_rate": 1e-06, + "loss": 1.0963, + "mean_token_accuracy": 0.6743375062942505, + "num_tokens": 81414778.0, + "step": 3162 + }, + { + "epoch": 0.34735339336701077, + "grad_norm": 1.9787671566009521, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6937227249145508, + "num_tokens": 81446889.0, + "step": 3163 + }, + { + "epoch": 0.3474632110696244, + "grad_norm": 2.32537841796875, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7124468684196472, + "num_tokens": 81470500.0, + "step": 3164 + }, + { + "epoch": 0.34757302877223806, + "grad_norm": 1.833228349685669, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6994247436523438, + "num_tokens": 81505791.0, + "step": 3165 + }, + { + "epoch": 0.34768284647485176, + "grad_norm": 2.11507248878479, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7078849077224731, + "num_tokens": 81531951.0, + "step": 3166 + }, + { + "epoch": 0.3477926641774654, + "grad_norm": 2.2225093841552734, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6860915422439575, + "num_tokens": 81559661.0, + "step": 3167 + }, + { + "epoch": 0.34790248188007905, + "grad_norm": 2.500239610671997, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7145677804946899, + "num_tokens": 81580467.0, + "step": 3168 + }, + { + "epoch": 0.34801229958269275, + "grad_norm": 2.015195369720459, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7153170108795166, + "num_tokens": 81607198.0, + "step": 3169 + }, + { + "epoch": 0.3481221172853064, + "grad_norm": 2.6185295581817627, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7116514444351196, + "num_tokens": 81626123.0, + "step": 3170 + }, + { + "epoch": 0.34823193498792004, + "grad_norm": 2.304588794708252, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7038203477859497, + "num_tokens": 81648338.0, + "step": 3171 + }, + { + "epoch": 0.3483417526905337, + "grad_norm": 2.3191065788269043, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7138952016830444, + "num_tokens": 81671922.0, + "step": 3172 + }, + { + "epoch": 0.3484515703931474, + "grad_norm": 2.123715400695801, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7240467071533203, + "num_tokens": 81698909.0, + "step": 3173 + }, + { + "epoch": 0.34856138809576104, + "grad_norm": 2.4000840187072754, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7019503116607666, + "num_tokens": 81720468.0, + "step": 3174 + }, + { + "epoch": 0.3486712057983747, + "grad_norm": 2.2875547409057617, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7010643482208252, + "num_tokens": 81744958.0, + "step": 3175 + }, + { + "epoch": 0.3487810235009884, + "grad_norm": 2.174987316131592, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7064222693443298, + "num_tokens": 81769156.0, + "step": 3176 + }, + { + "epoch": 0.34889084120360203, + "grad_norm": 2.325688123703003, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.696198582649231, + "num_tokens": 81792590.0, + "step": 3177 + }, + { + "epoch": 0.3490006589062157, + "grad_norm": 2.149242639541626, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6992962956428528, + "num_tokens": 81818137.0, + "step": 3178 + }, + { + "epoch": 0.3491104766088293, + "grad_norm": 2.47377872467041, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.682096540927887, + "num_tokens": 81840563.0, + "step": 3179 + }, + { + "epoch": 0.349220294311443, + "grad_norm": 2.3899145126342773, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6879854202270508, + "num_tokens": 81863846.0, + "step": 3180 + }, + { + "epoch": 0.34933011201405667, + "grad_norm": 2.532545804977417, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7048922181129456, + "num_tokens": 81885352.0, + "step": 3181 + }, + { + "epoch": 0.3494399297166703, + "grad_norm": 2.2772552967071533, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6817634105682373, + "num_tokens": 81911362.0, + "step": 3182 + }, + { + "epoch": 0.34954974741928396, + "grad_norm": 2.2501754760742188, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.69267737865448, + "num_tokens": 81937768.0, + "step": 3183 + }, + { + "epoch": 0.34965956512189766, + "grad_norm": 1.9158982038497925, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6956788897514343, + "num_tokens": 81970408.0, + "step": 3184 + }, + { + "epoch": 0.3497693828245113, + "grad_norm": 2.371887445449829, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.6986557841300964, + "num_tokens": 81994076.0, + "step": 3185 + }, + { + "epoch": 0.34987920052712496, + "grad_norm": 2.5486178398132324, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7076664566993713, + "num_tokens": 82014366.0, + "step": 3186 + }, + { + "epoch": 0.34998901822973866, + "grad_norm": 2.323833465576172, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.7023022174835205, + "num_tokens": 82039087.0, + "step": 3187 + }, + { + "epoch": 0.3500988359323523, + "grad_norm": 2.492800235748291, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7020778656005859, + "num_tokens": 82060799.0, + "step": 3188 + }, + { + "epoch": 0.35020865363496595, + "grad_norm": 2.2965755462646484, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7164103984832764, + "num_tokens": 82083256.0, + "step": 3189 + }, + { + "epoch": 0.3503184713375796, + "grad_norm": 2.5826027393341064, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6999422907829285, + "num_tokens": 82103072.0, + "step": 3190 + }, + { + "epoch": 0.3504282890401933, + "grad_norm": 2.080810546875, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.703933596611023, + "num_tokens": 82130659.0, + "step": 3191 + }, + { + "epoch": 0.35053810674280694, + "grad_norm": 2.28826642036438, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6945261359214783, + "num_tokens": 82154636.0, + "step": 3192 + }, + { + "epoch": 0.3506479244454206, + "grad_norm": 2.543421745300293, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.704444944858551, + "num_tokens": 82179662.0, + "step": 3193 + }, + { + "epoch": 0.3507577421480343, + "grad_norm": 2.3106091022491455, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6945549249649048, + "num_tokens": 82206016.0, + "step": 3194 + }, + { + "epoch": 0.35086755985064794, + "grad_norm": 2.204556941986084, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.693203866481781, + "num_tokens": 82234355.0, + "step": 3195 + }, + { + "epoch": 0.3509773775532616, + "grad_norm": 2.6806960105895996, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7251685857772827, + "num_tokens": 82251334.0, + "step": 3196 + }, + { + "epoch": 0.35108719525587523, + "grad_norm": 2.343623161315918, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6986615657806396, + "num_tokens": 82276189.0, + "step": 3197 + }, + { + "epoch": 0.35119701295848893, + "grad_norm": 2.237205743789673, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.7000663876533508, + "num_tokens": 82301782.0, + "step": 3198 + }, + { + "epoch": 0.3513068306611026, + "grad_norm": 2.22222638130188, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6895107626914978, + "num_tokens": 82326928.0, + "step": 3199 + }, + { + "epoch": 0.3514166483637162, + "grad_norm": 2.1296162605285645, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6894792914390564, + "num_tokens": 82354459.0, + "step": 3200 + }, + { + "epoch": 0.35152646606632987, + "grad_norm": 2.5332577228546143, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7167336344718933, + "num_tokens": 82375248.0, + "step": 3201 + }, + { + "epoch": 0.35163628376894357, + "grad_norm": 2.12878155708313, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7100229263305664, + "num_tokens": 82402800.0, + "step": 3202 + }, + { + "epoch": 0.3517461014715572, + "grad_norm": 2.1242494583129883, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6852482557296753, + "num_tokens": 82431023.0, + "step": 3203 + }, + { + "epoch": 0.35185591917417086, + "grad_norm": 2.211918354034424, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.692192792892456, + "num_tokens": 82457093.0, + "step": 3204 + }, + { + "epoch": 0.35196573687678456, + "grad_norm": 2.3194117546081543, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7100960612297058, + "num_tokens": 82481572.0, + "step": 3205 + }, + { + "epoch": 0.3520755545793982, + "grad_norm": 1.8807631731033325, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6791729927062988, + "num_tokens": 82517530.0, + "step": 3206 + }, + { + "epoch": 0.35218537228201185, + "grad_norm": 2.2293624877929688, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7055355906486511, + "num_tokens": 82541762.0, + "step": 3207 + }, + { + "epoch": 0.3522951899846255, + "grad_norm": 2.201456308364868, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6853920817375183, + "num_tokens": 82568459.0, + "step": 3208 + }, + { + "epoch": 0.3524050076872392, + "grad_norm": 1.9862467050552368, + "learning_rate": 1e-06, + "loss": 1.0772, + "mean_token_accuracy": 0.6831905841827393, + "num_tokens": 82599189.0, + "step": 3209 + }, + { + "epoch": 0.35251482538985285, + "grad_norm": 2.245471477508545, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7153164148330688, + "num_tokens": 82621647.0, + "step": 3210 + }, + { + "epoch": 0.3526246430924665, + "grad_norm": 2.2371044158935547, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6947296857833862, + "num_tokens": 82646047.0, + "step": 3211 + }, + { + "epoch": 0.35273446079508014, + "grad_norm": 2.182469606399536, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7095434665679932, + "num_tokens": 82671524.0, + "step": 3212 + }, + { + "epoch": 0.35284427849769384, + "grad_norm": 1.9362761974334717, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7026645541191101, + "num_tokens": 82701876.0, + "step": 3213 + }, + { + "epoch": 0.3529540962003075, + "grad_norm": 2.304544687271118, + "learning_rate": 1e-06, + "loss": 1.0804, + "mean_token_accuracy": 0.6832870244979858, + "num_tokens": 82727334.0, + "step": 3214 + }, + { + "epoch": 0.35306391390292113, + "grad_norm": 2.320084810256958, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6942352056503296, + "num_tokens": 82750680.0, + "step": 3215 + }, + { + "epoch": 0.35317373160553484, + "grad_norm": 1.8426353931427002, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6849216222763062, + "num_tokens": 82787304.0, + "step": 3216 + }, + { + "epoch": 0.3532835493081485, + "grad_norm": 2.279306411743164, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.702971339225769, + "num_tokens": 82813159.0, + "step": 3217 + }, + { + "epoch": 0.3533933670107621, + "grad_norm": 2.2471730709075928, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.698236346244812, + "num_tokens": 82838287.0, + "step": 3218 + }, + { + "epoch": 0.3535031847133758, + "grad_norm": 2.3075459003448486, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6966552734375, + "num_tokens": 82861250.0, + "step": 3219 + }, + { + "epoch": 0.3536130024159895, + "grad_norm": 2.055280923843384, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.685696542263031, + "num_tokens": 82891384.0, + "step": 3220 + }, + { + "epoch": 0.3537228201186031, + "grad_norm": 2.2894303798675537, + "learning_rate": 1e-06, + "loss": 1.0715, + "mean_token_accuracy": 0.6854063868522644, + "num_tokens": 82916520.0, + "step": 3221 + }, + { + "epoch": 0.35383263782121677, + "grad_norm": 2.1138181686401367, + "learning_rate": 1e-06, + "loss": 1.1069, + "mean_token_accuracy": 0.6693688631057739, + "num_tokens": 82946528.0, + "step": 3222 + }, + { + "epoch": 0.35394245552383047, + "grad_norm": 2.2068846225738525, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6918327808380127, + "num_tokens": 82971047.0, + "step": 3223 + }, + { + "epoch": 0.3540522732264441, + "grad_norm": 2.376007080078125, + "learning_rate": 1e-06, + "loss": 1.0723, + "mean_token_accuracy": 0.6793021559715271, + "num_tokens": 82995870.0, + "step": 3224 + }, + { + "epoch": 0.35416209092905776, + "grad_norm": 2.4337728023529053, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7237977981567383, + "num_tokens": 83018385.0, + "step": 3225 + }, + { + "epoch": 0.3542719086316714, + "grad_norm": 2.404113531112671, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.722808837890625, + "num_tokens": 83039593.0, + "step": 3226 + }, + { + "epoch": 0.3543817263342851, + "grad_norm": 1.9949661493301392, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6871947050094604, + "num_tokens": 83071138.0, + "step": 3227 + }, + { + "epoch": 0.35449154403689875, + "grad_norm": 2.3109548091888428, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7154434323310852, + "num_tokens": 83094136.0, + "step": 3228 + }, + { + "epoch": 0.3546013617395124, + "grad_norm": 2.3105788230895996, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7373562455177307, + "num_tokens": 83115036.0, + "step": 3229 + }, + { + "epoch": 0.35471117944212605, + "grad_norm": 2.2376906871795654, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.707964301109314, + "num_tokens": 83138785.0, + "step": 3230 + }, + { + "epoch": 0.35482099714473975, + "grad_norm": 2.7977800369262695, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7006852626800537, + "num_tokens": 83160271.0, + "step": 3231 + }, + { + "epoch": 0.3549308148473534, + "grad_norm": 2.1296615600585938, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6849411129951477, + "num_tokens": 83189694.0, + "step": 3232 + }, + { + "epoch": 0.35504063254996704, + "grad_norm": 2.172078847885132, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7189232707023621, + "num_tokens": 83215371.0, + "step": 3233 + }, + { + "epoch": 0.35515045025258074, + "grad_norm": 2.444420099258423, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6978839635848999, + "num_tokens": 83239169.0, + "step": 3234 + }, + { + "epoch": 0.3552602679551944, + "grad_norm": 2.237765073776245, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7157891988754272, + "num_tokens": 83263629.0, + "step": 3235 + }, + { + "epoch": 0.35537008565780803, + "grad_norm": 2.071868896484375, + "learning_rate": 1e-06, + "loss": 1.0841, + "mean_token_accuracy": 0.6767175793647766, + "num_tokens": 83293856.0, + "step": 3236 + }, + { + "epoch": 0.3554799033604217, + "grad_norm": 2.2055788040161133, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6990436315536499, + "num_tokens": 83321670.0, + "step": 3237 + }, + { + "epoch": 0.3555897210630354, + "grad_norm": 2.099060535430908, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.721189022064209, + "num_tokens": 83348258.0, + "step": 3238 + }, + { + "epoch": 0.355699538765649, + "grad_norm": 2.319856643676758, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7060866355895996, + "num_tokens": 83371606.0, + "step": 3239 + }, + { + "epoch": 0.35580935646826267, + "grad_norm": 1.9807775020599365, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.708317756652832, + "num_tokens": 83404630.0, + "step": 3240 + }, + { + "epoch": 0.3559191741708763, + "grad_norm": 1.8669487237930298, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7160992622375488, + "num_tokens": 83436140.0, + "step": 3241 + }, + { + "epoch": 0.35602899187349, + "grad_norm": 2.209287166595459, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7090582847595215, + "num_tokens": 83460095.0, + "step": 3242 + }, + { + "epoch": 0.35613880957610367, + "grad_norm": 2.402657985687256, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7051967978477478, + "num_tokens": 83483407.0, + "step": 3243 + }, + { + "epoch": 0.3562486272787173, + "grad_norm": 2.3911592960357666, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6914101243019104, + "num_tokens": 83506900.0, + "step": 3244 + }, + { + "epoch": 0.356358444981331, + "grad_norm": 2.01875376701355, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6972357630729675, + "num_tokens": 83535773.0, + "step": 3245 + }, + { + "epoch": 0.35646826268394466, + "grad_norm": 2.66094970703125, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.697917103767395, + "num_tokens": 83556485.0, + "step": 3246 + }, + { + "epoch": 0.3565780803865583, + "grad_norm": 2.0520894527435303, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7130099534988403, + "num_tokens": 83586232.0, + "step": 3247 + }, + { + "epoch": 0.35668789808917195, + "grad_norm": 2.2380406856536865, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7148122191429138, + "num_tokens": 83608921.0, + "step": 3248 + }, + { + "epoch": 0.35679771579178565, + "grad_norm": 2.016953468322754, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7014586925506592, + "num_tokens": 83639135.0, + "step": 3249 + }, + { + "epoch": 0.3569075334943993, + "grad_norm": 2.264899253845215, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.696308970451355, + "num_tokens": 83665170.0, + "step": 3250 + }, + { + "epoch": 0.35701735119701294, + "grad_norm": 2.4359614849090576, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.692283034324646, + "num_tokens": 83687010.0, + "step": 3251 + }, + { + "epoch": 0.35712716889962665, + "grad_norm": 1.9735709428787231, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7139694094657898, + "num_tokens": 83716844.0, + "step": 3252 + }, + { + "epoch": 0.3572369866022403, + "grad_norm": 2.133906364440918, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6904788017272949, + "num_tokens": 83743173.0, + "step": 3253 + }, + { + "epoch": 0.35734680430485394, + "grad_norm": 2.3520195484161377, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6894459128379822, + "num_tokens": 83765916.0, + "step": 3254 + }, + { + "epoch": 0.3574566220074676, + "grad_norm": 2.387592077255249, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7015930414199829, + "num_tokens": 83792371.0, + "step": 3255 + }, + { + "epoch": 0.3575664397100813, + "grad_norm": 2.2139573097229004, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7054288387298584, + "num_tokens": 83818202.0, + "step": 3256 + }, + { + "epoch": 0.35767625741269493, + "grad_norm": 2.099506378173828, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.6916425228118896, + "num_tokens": 83847795.0, + "step": 3257 + }, + { + "epoch": 0.3577860751153086, + "grad_norm": 2.287435531616211, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6971375942230225, + "num_tokens": 83871988.0, + "step": 3258 + }, + { + "epoch": 0.3578958928179222, + "grad_norm": 2.367103099822998, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7018624544143677, + "num_tokens": 83894161.0, + "step": 3259 + }, + { + "epoch": 0.3580057105205359, + "grad_norm": 1.9077262878417969, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.716987669467926, + "num_tokens": 83925253.0, + "step": 3260 + }, + { + "epoch": 0.35811552822314957, + "grad_norm": 2.405625820159912, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7041209936141968, + "num_tokens": 83946960.0, + "step": 3261 + }, + { + "epoch": 0.3582253459257632, + "grad_norm": 2.2677693367004395, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.698297917842865, + "num_tokens": 83971775.0, + "step": 3262 + }, + { + "epoch": 0.3583351636283769, + "grad_norm": 2.257448673248291, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6982477903366089, + "num_tokens": 83997963.0, + "step": 3263 + }, + { + "epoch": 0.35844498133099056, + "grad_norm": 2.5180704593658447, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7170848846435547, + "num_tokens": 84019311.0, + "step": 3264 + }, + { + "epoch": 0.3585547990336042, + "grad_norm": 2.1435272693634033, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6801724433898926, + "num_tokens": 84046659.0, + "step": 3265 + }, + { + "epoch": 0.35866461673621786, + "grad_norm": 2.01332426071167, + "learning_rate": 1e-06, + "loss": 1.1079, + "mean_token_accuracy": 0.6743039488792419, + "num_tokens": 84079132.0, + "step": 3266 + }, + { + "epoch": 0.35877443443883156, + "grad_norm": 2.476447105407715, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6814683675765991, + "num_tokens": 84101465.0, + "step": 3267 + }, + { + "epoch": 0.3588842521414452, + "grad_norm": 2.191352605819702, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.6779831051826477, + "num_tokens": 84130091.0, + "step": 3268 + }, + { + "epoch": 0.35899406984405885, + "grad_norm": 2.2915377616882324, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.7033172249794006, + "num_tokens": 84154276.0, + "step": 3269 + }, + { + "epoch": 0.35910388754667255, + "grad_norm": 2.0308122634887695, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6990571022033691, + "num_tokens": 84183850.0, + "step": 3270 + }, + { + "epoch": 0.3592137052492862, + "grad_norm": 2.4267261028289795, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6946111917495728, + "num_tokens": 84206798.0, + "step": 3271 + }, + { + "epoch": 0.35932352295189984, + "grad_norm": 2.1807267665863037, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6928471326828003, + "num_tokens": 84231414.0, + "step": 3272 + }, + { + "epoch": 0.3594333406545135, + "grad_norm": 2.4823384284973145, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7065132856369019, + "num_tokens": 84252294.0, + "step": 3273 + }, + { + "epoch": 0.3595431583571272, + "grad_norm": 2.172245502471924, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.68443363904953, + "num_tokens": 84279314.0, + "step": 3274 + }, + { + "epoch": 0.35965297605974084, + "grad_norm": 2.4175069332122803, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6954388618469238, + "num_tokens": 84302044.0, + "step": 3275 + }, + { + "epoch": 0.3597627937623545, + "grad_norm": 2.21755313873291, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.6791912913322449, + "num_tokens": 84330548.0, + "step": 3276 + }, + { + "epoch": 0.35987261146496813, + "grad_norm": 2.4420766830444336, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6936908960342407, + "num_tokens": 84352146.0, + "step": 3277 + }, + { + "epoch": 0.35998242916758183, + "grad_norm": 2.0489721298217773, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.689117431640625, + "num_tokens": 84381148.0, + "step": 3278 + }, + { + "epoch": 0.3600922468701955, + "grad_norm": 2.185246706008911, + "learning_rate": 1e-06, + "loss": 1.0962, + "mean_token_accuracy": 0.6786352396011353, + "num_tokens": 84409414.0, + "step": 3279 + }, + { + "epoch": 0.3602020645728091, + "grad_norm": 2.5504982471466064, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7076331377029419, + "num_tokens": 84429864.0, + "step": 3280 + }, + { + "epoch": 0.3603118822754228, + "grad_norm": 2.3605079650878906, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7082182168960571, + "num_tokens": 84452094.0, + "step": 3281 + }, + { + "epoch": 0.36042169997803647, + "grad_norm": 2.6016788482666016, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7102360129356384, + "num_tokens": 84474298.0, + "step": 3282 + }, + { + "epoch": 0.3605315176806501, + "grad_norm": 2.1719768047332764, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7169312238693237, + "num_tokens": 84499236.0, + "step": 3283 + }, + { + "epoch": 0.36064133538326376, + "grad_norm": 2.0703978538513184, + "learning_rate": 1e-06, + "loss": 1.1105, + "mean_token_accuracy": 0.6668366193771362, + "num_tokens": 84528957.0, + "step": 3284 + }, + { + "epoch": 0.36075115308587746, + "grad_norm": 2.4057910442352295, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7109960317611694, + "num_tokens": 84550869.0, + "step": 3285 + }, + { + "epoch": 0.3608609707884911, + "grad_norm": 2.614569664001465, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7078245282173157, + "num_tokens": 84570252.0, + "step": 3286 + }, + { + "epoch": 0.36097078849110475, + "grad_norm": 2.7277512550354004, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7094248533248901, + "num_tokens": 84587543.0, + "step": 3287 + }, + { + "epoch": 0.3610806061937184, + "grad_norm": 1.916764736175537, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6818413734436035, + "num_tokens": 84618426.0, + "step": 3288 + }, + { + "epoch": 0.3611904238963321, + "grad_norm": 2.1588239669799805, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7105762958526611, + "num_tokens": 84643323.0, + "step": 3289 + }, + { + "epoch": 0.36130024159894575, + "grad_norm": 2.2799267768859863, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7283012866973877, + "num_tokens": 84665676.0, + "step": 3290 + }, + { + "epoch": 0.3614100593015594, + "grad_norm": 2.1345980167388916, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7065683007240295, + "num_tokens": 84693127.0, + "step": 3291 + }, + { + "epoch": 0.3615198770041731, + "grad_norm": 2.273325204849243, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6776326298713684, + "num_tokens": 84719314.0, + "step": 3292 + }, + { + "epoch": 0.36162969470678674, + "grad_norm": 2.227771520614624, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7098280191421509, + "num_tokens": 84747492.0, + "step": 3293 + }, + { + "epoch": 0.3617395124094004, + "grad_norm": 2.2110342979431152, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7012860774993896, + "num_tokens": 84775382.0, + "step": 3294 + }, + { + "epoch": 0.36184933011201403, + "grad_norm": 2.2648746967315674, + "learning_rate": 1e-06, + "loss": 1.1096, + "mean_token_accuracy": 0.6701865196228027, + "num_tokens": 84804471.0, + "step": 3295 + }, + { + "epoch": 0.36195914781462774, + "grad_norm": 2.039790391921997, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7034925818443298, + "num_tokens": 84834746.0, + "step": 3296 + }, + { + "epoch": 0.3620689655172414, + "grad_norm": 1.9211621284484863, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6814887523651123, + "num_tokens": 84871187.0, + "step": 3297 + }, + { + "epoch": 0.362178783219855, + "grad_norm": 2.1442482471466064, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7144349813461304, + "num_tokens": 84897226.0, + "step": 3298 + }, + { + "epoch": 0.36228860092246873, + "grad_norm": 1.8526092767715454, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6900449991226196, + "num_tokens": 84930847.0, + "step": 3299 + }, + { + "epoch": 0.3623984186250824, + "grad_norm": 2.163695812225342, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7013518810272217, + "num_tokens": 84956870.0, + "step": 3300 + }, + { + "epoch": 0.362508236327696, + "grad_norm": 2.199496269226074, + "learning_rate": 1e-06, + "loss": 1.0931, + "mean_token_accuracy": 0.6764432787895203, + "num_tokens": 84987710.0, + "step": 3301 + }, + { + "epoch": 0.36261805403030967, + "grad_norm": 2.552669048309326, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7216964960098267, + "num_tokens": 85008793.0, + "step": 3302 + }, + { + "epoch": 0.36272787173292337, + "grad_norm": 2.268357515335083, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7170262336730957, + "num_tokens": 85033503.0, + "step": 3303 + }, + { + "epoch": 0.362837689435537, + "grad_norm": 2.26641845703125, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.7004212141036987, + "num_tokens": 85060630.0, + "step": 3304 + }, + { + "epoch": 0.36294750713815066, + "grad_norm": 2.2212462425231934, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7168420553207397, + "num_tokens": 85083385.0, + "step": 3305 + }, + { + "epoch": 0.3630573248407643, + "grad_norm": 2.248565912246704, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6964904069900513, + "num_tokens": 85108756.0, + "step": 3306 + }, + { + "epoch": 0.363167142543378, + "grad_norm": 2.553051710128784, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6966152191162109, + "num_tokens": 85132365.0, + "step": 3307 + }, + { + "epoch": 0.36327696024599165, + "grad_norm": 2.4579827785491943, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7014811038970947, + "num_tokens": 85155928.0, + "step": 3308 + }, + { + "epoch": 0.3633867779486053, + "grad_norm": 2.55129075050354, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6875780820846558, + "num_tokens": 85176246.0, + "step": 3309 + }, + { + "epoch": 0.363496595651219, + "grad_norm": 2.2876267433166504, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6989660263061523, + "num_tokens": 85199877.0, + "step": 3310 + }, + { + "epoch": 0.36360641335383265, + "grad_norm": 2.205228567123413, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6965482234954834, + "num_tokens": 85224782.0, + "step": 3311 + }, + { + "epoch": 0.3637162310564463, + "grad_norm": 2.2895665168762207, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7081859111785889, + "num_tokens": 85251965.0, + "step": 3312 + }, + { + "epoch": 0.36382604875905994, + "grad_norm": 2.1675870418548584, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6995847225189209, + "num_tokens": 85280031.0, + "step": 3313 + }, + { + "epoch": 0.36393586646167364, + "grad_norm": 2.0137782096862793, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6895745396614075, + "num_tokens": 85312191.0, + "step": 3314 + }, + { + "epoch": 0.3640456841642873, + "grad_norm": 1.9452623128890991, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6883050799369812, + "num_tokens": 85343691.0, + "step": 3315 + }, + { + "epoch": 0.36415550186690093, + "grad_norm": 2.381138563156128, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7037268877029419, + "num_tokens": 85366033.0, + "step": 3316 + }, + { + "epoch": 0.3642653195695146, + "grad_norm": 2.3391685485839844, + "learning_rate": 1e-06, + "loss": 1.0999, + "mean_token_accuracy": 0.670464038848877, + "num_tokens": 85391112.0, + "step": 3317 + }, + { + "epoch": 0.3643751372721283, + "grad_norm": 2.05073881149292, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6832464933395386, + "num_tokens": 85420383.0, + "step": 3318 + }, + { + "epoch": 0.3644849549747419, + "grad_norm": 1.9445631504058838, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6813708543777466, + "num_tokens": 85451369.0, + "step": 3319 + }, + { + "epoch": 0.36459477267735557, + "grad_norm": 2.393681049346924, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7258595824241638, + "num_tokens": 85475111.0, + "step": 3320 + }, + { + "epoch": 0.3647045903799693, + "grad_norm": 2.36811900138855, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.702691912651062, + "num_tokens": 85499584.0, + "step": 3321 + }, + { + "epoch": 0.3648144080825829, + "grad_norm": 1.9547150135040283, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6964297294616699, + "num_tokens": 85532267.0, + "step": 3322 + }, + { + "epoch": 0.36492422578519657, + "grad_norm": 2.5004446506500244, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7025843858718872, + "num_tokens": 85553078.0, + "step": 3323 + }, + { + "epoch": 0.3650340434878102, + "grad_norm": 2.222382068634033, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7081292867660522, + "num_tokens": 85578997.0, + "step": 3324 + }, + { + "epoch": 0.3651438611904239, + "grad_norm": 2.0969951152801514, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7096468806266785, + "num_tokens": 85605897.0, + "step": 3325 + }, + { + "epoch": 0.36525367889303756, + "grad_norm": 2.55399227142334, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.695442259311676, + "num_tokens": 85627033.0, + "step": 3326 + }, + { + "epoch": 0.3653634965956512, + "grad_norm": 2.487433671951294, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7065762877464294, + "num_tokens": 85648049.0, + "step": 3327 + }, + { + "epoch": 0.3654733142982649, + "grad_norm": 2.2344419956207275, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7101536989212036, + "num_tokens": 85674686.0, + "step": 3328 + }, + { + "epoch": 0.36558313200087855, + "grad_norm": 2.3771440982818604, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.707852840423584, + "num_tokens": 85697899.0, + "step": 3329 + }, + { + "epoch": 0.3656929497034922, + "grad_norm": 1.939287781715393, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7069714069366455, + "num_tokens": 85729301.0, + "step": 3330 + }, + { + "epoch": 0.36580276740610584, + "grad_norm": 2.0669913291931152, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6953600645065308, + "num_tokens": 85757916.0, + "step": 3331 + }, + { + "epoch": 0.36591258510871955, + "grad_norm": 2.5696523189544678, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7108504772186279, + "num_tokens": 85777311.0, + "step": 3332 + }, + { + "epoch": 0.3660224028113332, + "grad_norm": 2.6929590702056885, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.69400554895401, + "num_tokens": 85797132.0, + "step": 3333 + }, + { + "epoch": 0.36613222051394684, + "grad_norm": 1.9661369323730469, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7143906354904175, + "num_tokens": 85826855.0, + "step": 3334 + }, + { + "epoch": 0.3662420382165605, + "grad_norm": 2.0798609256744385, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7178186178207397, + "num_tokens": 85854544.0, + "step": 3335 + }, + { + "epoch": 0.3663518559191742, + "grad_norm": 2.3749802112579346, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6993718147277832, + "num_tokens": 85878818.0, + "step": 3336 + }, + { + "epoch": 0.36646167362178783, + "grad_norm": 2.096820116043091, + "learning_rate": 1e-06, + "loss": 1.0694, + "mean_token_accuracy": 0.6806011199951172, + "num_tokens": 85908819.0, + "step": 3337 + }, + { + "epoch": 0.3665714913244015, + "grad_norm": 2.0789601802825928, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7044926881790161, + "num_tokens": 85936180.0, + "step": 3338 + }, + { + "epoch": 0.3666813090270152, + "grad_norm": 2.2123379707336426, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6942225694656372, + "num_tokens": 85960455.0, + "step": 3339 + }, + { + "epoch": 0.3667911267296288, + "grad_norm": 2.372393846511841, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.703567385673523, + "num_tokens": 85983157.0, + "step": 3340 + }, + { + "epoch": 0.36690094443224247, + "grad_norm": 2.29508113861084, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6881679892539978, + "num_tokens": 86008399.0, + "step": 3341 + }, + { + "epoch": 0.3670107621348561, + "grad_norm": 2.04884672164917, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6893994808197021, + "num_tokens": 86037878.0, + "step": 3342 + }, + { + "epoch": 0.3671205798374698, + "grad_norm": 2.8029050827026367, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7412343621253967, + "num_tokens": 86054233.0, + "step": 3343 + }, + { + "epoch": 0.36723039754008346, + "grad_norm": 2.358849287033081, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6925968527793884, + "num_tokens": 86077885.0, + "step": 3344 + }, + { + "epoch": 0.3673402152426971, + "grad_norm": 2.2068631649017334, + "learning_rate": 1e-06, + "loss": 1.0957, + "mean_token_accuracy": 0.6708750128746033, + "num_tokens": 86105970.0, + "step": 3345 + }, + { + "epoch": 0.3674500329453108, + "grad_norm": 1.9754306077957153, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6986703276634216, + "num_tokens": 86136079.0, + "step": 3346 + }, + { + "epoch": 0.36755985064792446, + "grad_norm": 2.2280070781707764, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7029637098312378, + "num_tokens": 86162889.0, + "step": 3347 + }, + { + "epoch": 0.3676696683505381, + "grad_norm": 2.44482421875, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7066766619682312, + "num_tokens": 86185927.0, + "step": 3348 + }, + { + "epoch": 0.36777948605315175, + "grad_norm": 2.4094393253326416, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6913796663284302, + "num_tokens": 86209511.0, + "step": 3349 + }, + { + "epoch": 0.36788930375576545, + "grad_norm": 2.456592559814453, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7048631310462952, + "num_tokens": 86231273.0, + "step": 3350 + }, + { + "epoch": 0.3679991214583791, + "grad_norm": 2.298215627670288, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6999337077140808, + "num_tokens": 86255341.0, + "step": 3351 + }, + { + "epoch": 0.36810893916099274, + "grad_norm": 2.263127088546753, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7071164846420288, + "num_tokens": 86280727.0, + "step": 3352 + }, + { + "epoch": 0.3682187568636064, + "grad_norm": 2.3223729133605957, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.710940957069397, + "num_tokens": 86304477.0, + "step": 3353 + }, + { + "epoch": 0.3683285745662201, + "grad_norm": 2.0822463035583496, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6906753182411194, + "num_tokens": 86333880.0, + "step": 3354 + }, + { + "epoch": 0.36843839226883374, + "grad_norm": 2.3075363636016846, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6996219158172607, + "num_tokens": 86356578.0, + "step": 3355 + }, + { + "epoch": 0.3685482099714474, + "grad_norm": 2.1291306018829346, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7052932977676392, + "num_tokens": 86385670.0, + "step": 3356 + }, + { + "epoch": 0.3686580276740611, + "grad_norm": 2.1802213191986084, + "learning_rate": 1e-06, + "loss": 1.1136, + "mean_token_accuracy": 0.6706119775772095, + "num_tokens": 86415031.0, + "step": 3357 + }, + { + "epoch": 0.36876784537667473, + "grad_norm": 2.106224536895752, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.71099853515625, + "num_tokens": 86442014.0, + "step": 3358 + }, + { + "epoch": 0.3688776630792884, + "grad_norm": 2.4339170455932617, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6867730021476746, + "num_tokens": 86462958.0, + "step": 3359 + }, + { + "epoch": 0.368987480781902, + "grad_norm": 2.297337532043457, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6863182187080383, + "num_tokens": 86487505.0, + "step": 3360 + }, + { + "epoch": 0.3690972984845157, + "grad_norm": 2.4898860454559326, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.697730541229248, + "num_tokens": 86508234.0, + "step": 3361 + }, + { + "epoch": 0.36920711618712937, + "grad_norm": 2.016256093978882, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6992557048797607, + "num_tokens": 86537497.0, + "step": 3362 + }, + { + "epoch": 0.369316933889743, + "grad_norm": 2.237921714782715, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6865031123161316, + "num_tokens": 86565017.0, + "step": 3363 + }, + { + "epoch": 0.36942675159235666, + "grad_norm": 2.494443893432617, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6917506456375122, + "num_tokens": 86585970.0, + "step": 3364 + }, + { + "epoch": 0.36953656929497036, + "grad_norm": 2.4421281814575195, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6984906196594238, + "num_tokens": 86609460.0, + "step": 3365 + }, + { + "epoch": 0.369646386997584, + "grad_norm": 2.3222007751464844, + "learning_rate": 1e-06, + "loss": 1.087, + "mean_token_accuracy": 0.6774585247039795, + "num_tokens": 86634500.0, + "step": 3366 + }, + { + "epoch": 0.36975620470019765, + "grad_norm": 2.1638588905334473, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7201194763183594, + "num_tokens": 86661550.0, + "step": 3367 + }, + { + "epoch": 0.36986602240281136, + "grad_norm": 2.1025314331054688, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6947133541107178, + "num_tokens": 86692574.0, + "step": 3368 + }, + { + "epoch": 0.369975840105425, + "grad_norm": 2.3417210578918457, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.699893593788147, + "num_tokens": 86715979.0, + "step": 3369 + }, + { + "epoch": 0.37008565780803865, + "grad_norm": 2.1616687774658203, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7016170024871826, + "num_tokens": 86742155.0, + "step": 3370 + }, + { + "epoch": 0.3701954755106523, + "grad_norm": 2.184248447418213, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7124893665313721, + "num_tokens": 86770638.0, + "step": 3371 + }, + { + "epoch": 0.370305293213266, + "grad_norm": 2.216360092163086, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7000737190246582, + "num_tokens": 86797083.0, + "step": 3372 + }, + { + "epoch": 0.37041511091587964, + "grad_norm": 2.1162664890289307, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6899108290672302, + "num_tokens": 86826733.0, + "step": 3373 + }, + { + "epoch": 0.3705249286184933, + "grad_norm": 2.1535041332244873, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6986551284790039, + "num_tokens": 86854377.0, + "step": 3374 + }, + { + "epoch": 0.370634746321107, + "grad_norm": 2.2218852043151855, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.7006858587265015, + "num_tokens": 86879789.0, + "step": 3375 + }, + { + "epoch": 0.37074456402372064, + "grad_norm": 2.346973419189453, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.690276026725769, + "num_tokens": 86902825.0, + "step": 3376 + }, + { + "epoch": 0.3708543817263343, + "grad_norm": 2.122526168823242, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.718742847442627, + "num_tokens": 86930689.0, + "step": 3377 + }, + { + "epoch": 0.3709641994289479, + "grad_norm": 2.3662002086639404, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7247442603111267, + "num_tokens": 86951312.0, + "step": 3378 + }, + { + "epoch": 0.37107401713156163, + "grad_norm": 2.4668803215026855, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.716454029083252, + "num_tokens": 86972493.0, + "step": 3379 + }, + { + "epoch": 0.3711838348341753, + "grad_norm": 1.9890625476837158, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6939471960067749, + "num_tokens": 87002964.0, + "step": 3380 + }, + { + "epoch": 0.3712936525367889, + "grad_norm": 2.2263453006744385, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7014517784118652, + "num_tokens": 87028347.0, + "step": 3381 + }, + { + "epoch": 0.37140347023940257, + "grad_norm": 2.331761121749878, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7079188823699951, + "num_tokens": 87051764.0, + "step": 3382 + }, + { + "epoch": 0.37151328794201627, + "grad_norm": 2.064584255218506, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7603808045387268, + "num_tokens": 87075658.0, + "step": 3383 + }, + { + "epoch": 0.3716231056446299, + "grad_norm": 2.298037052154541, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6992055773735046, + "num_tokens": 87101324.0, + "step": 3384 + }, + { + "epoch": 0.37173292334724356, + "grad_norm": 2.444823741912842, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6953476667404175, + "num_tokens": 87125682.0, + "step": 3385 + }, + { + "epoch": 0.37184274104985726, + "grad_norm": 1.958916187286377, + "learning_rate": 1e-06, + "loss": 1.0973, + "mean_token_accuracy": 0.67930006980896, + "num_tokens": 87161821.0, + "step": 3386 + }, + { + "epoch": 0.3719525587524709, + "grad_norm": 2.478502035140991, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7081679105758667, + "num_tokens": 87185915.0, + "step": 3387 + }, + { + "epoch": 0.37206237645508455, + "grad_norm": 2.440303087234497, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7048294544219971, + "num_tokens": 87207806.0, + "step": 3388 + }, + { + "epoch": 0.3721721941576982, + "grad_norm": 2.5220794677734375, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7040097713470459, + "num_tokens": 87228365.0, + "step": 3389 + }, + { + "epoch": 0.3722820118603119, + "grad_norm": 2.1785850524902344, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7072017788887024, + "num_tokens": 87253360.0, + "step": 3390 + }, + { + "epoch": 0.37239182956292555, + "grad_norm": 2.3963472843170166, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.709902286529541, + "num_tokens": 87275787.0, + "step": 3391 + }, + { + "epoch": 0.3725016472655392, + "grad_norm": 2.1444449424743652, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6920309662818909, + "num_tokens": 87305237.0, + "step": 3392 + }, + { + "epoch": 0.37261146496815284, + "grad_norm": 2.3154938220977783, + "learning_rate": 1e-06, + "loss": 1.0938, + "mean_token_accuracy": 0.6789851784706116, + "num_tokens": 87328739.0, + "step": 3393 + }, + { + "epoch": 0.37272128267076654, + "grad_norm": 2.296114683151245, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7026913166046143, + "num_tokens": 87351031.0, + "step": 3394 + }, + { + "epoch": 0.3728311003733802, + "grad_norm": 2.049268960952759, + "learning_rate": 1e-06, + "loss": 1.0922, + "mean_token_accuracy": 0.6759483814239502, + "num_tokens": 87381289.0, + "step": 3395 + }, + { + "epoch": 0.37294091807599383, + "grad_norm": 2.2160964012145996, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6986459493637085, + "num_tokens": 87406857.0, + "step": 3396 + }, + { + "epoch": 0.37305073577860753, + "grad_norm": 2.0136516094207764, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6898350715637207, + "num_tokens": 87436361.0, + "step": 3397 + }, + { + "epoch": 0.3731605534812212, + "grad_norm": 2.1006762981414795, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7105982303619385, + "num_tokens": 87462277.0, + "step": 3398 + }, + { + "epoch": 0.3732703711838348, + "grad_norm": 2.25728702545166, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7076324224472046, + "num_tokens": 87487667.0, + "step": 3399 + }, + { + "epoch": 0.37338018888644847, + "grad_norm": 2.140613079071045, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6883420348167419, + "num_tokens": 87517270.0, + "step": 3400 + }, + { + "epoch": 0.3734900065890622, + "grad_norm": 2.0213732719421387, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.7015783190727234, + "num_tokens": 87548935.0, + "step": 3401 + }, + { + "epoch": 0.3735998242916758, + "grad_norm": 2.262425661087036, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7178117036819458, + "num_tokens": 87574283.0, + "step": 3402 + }, + { + "epoch": 0.37370964199428947, + "grad_norm": 2.1345529556274414, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7209247350692749, + "num_tokens": 87598648.0, + "step": 3403 + }, + { + "epoch": 0.37381945969690317, + "grad_norm": 2.044282913208008, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6965795159339905, + "num_tokens": 87628709.0, + "step": 3404 + }, + { + "epoch": 0.3739292773995168, + "grad_norm": 2.174400568008423, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7195744514465332, + "num_tokens": 87656685.0, + "step": 3405 + }, + { + "epoch": 0.37403909510213046, + "grad_norm": 2.4511170387268066, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7025817036628723, + "num_tokens": 87678140.0, + "step": 3406 + }, + { + "epoch": 0.3741489128047441, + "grad_norm": 2.1821930408477783, + "learning_rate": 1e-06, + "loss": 1.093, + "mean_token_accuracy": 0.6754451990127563, + "num_tokens": 87708743.0, + "step": 3407 + }, + { + "epoch": 0.3742587305073578, + "grad_norm": 2.019137382507324, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7002400755882263, + "num_tokens": 87739110.0, + "step": 3408 + }, + { + "epoch": 0.37436854820997145, + "grad_norm": 2.1054816246032715, + "learning_rate": 1e-06, + "loss": 1.0941, + "mean_token_accuracy": 0.6726928949356079, + "num_tokens": 87771342.0, + "step": 3409 + }, + { + "epoch": 0.3744783659125851, + "grad_norm": 1.9628517627716064, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6917974352836609, + "num_tokens": 87805432.0, + "step": 3410 + }, + { + "epoch": 0.37458818361519874, + "grad_norm": 2.526888370513916, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.7006000280380249, + "num_tokens": 87826316.0, + "step": 3411 + }, + { + "epoch": 0.37469800131781245, + "grad_norm": 2.3010709285736084, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6995376944541931, + "num_tokens": 87851300.0, + "step": 3412 + }, + { + "epoch": 0.3748078190204261, + "grad_norm": 2.0245025157928467, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7200747728347778, + "num_tokens": 87880454.0, + "step": 3413 + }, + { + "epoch": 0.37491763672303974, + "grad_norm": 1.992966890335083, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.7009683847427368, + "num_tokens": 87911462.0, + "step": 3414 + }, + { + "epoch": 0.37502745442565344, + "grad_norm": 2.3316380977630615, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7115968465805054, + "num_tokens": 87934910.0, + "step": 3415 + }, + { + "epoch": 0.3751372721282671, + "grad_norm": 2.7540910243988037, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7011005878448486, + "num_tokens": 87956076.0, + "step": 3416 + }, + { + "epoch": 0.37524708983088073, + "grad_norm": 2.2635843753814697, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7151228785514832, + "num_tokens": 87980300.0, + "step": 3417 + }, + { + "epoch": 0.3753569075334944, + "grad_norm": 2.4237101078033447, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.695148229598999, + "num_tokens": 88005048.0, + "step": 3418 + }, + { + "epoch": 0.3754667252361081, + "grad_norm": 2.471609354019165, + "learning_rate": 1e-06, + "loss": 1.0762, + "mean_token_accuracy": 0.6848117113113403, + "num_tokens": 88028811.0, + "step": 3419 + }, + { + "epoch": 0.3755765429387217, + "grad_norm": 2.160362958908081, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6980136036872864, + "num_tokens": 88054373.0, + "step": 3420 + }, + { + "epoch": 0.37568636064133537, + "grad_norm": 2.179811716079712, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.688300609588623, + "num_tokens": 88080066.0, + "step": 3421 + }, + { + "epoch": 0.37579617834394907, + "grad_norm": 2.413646697998047, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7140165567398071, + "num_tokens": 88104128.0, + "step": 3422 + }, + { + "epoch": 0.3759059960465627, + "grad_norm": 2.090952157974243, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7082340121269226, + "num_tokens": 88132344.0, + "step": 3423 + }, + { + "epoch": 0.37601581374917636, + "grad_norm": 1.89049232006073, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6985434293746948, + "num_tokens": 88163841.0, + "step": 3424 + }, + { + "epoch": 0.37612563145179, + "grad_norm": 2.0167617797851562, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6933450698852539, + "num_tokens": 88194303.0, + "step": 3425 + }, + { + "epoch": 0.3762354491544037, + "grad_norm": 2.2150211334228516, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6863878965377808, + "num_tokens": 88220998.0, + "step": 3426 + }, + { + "epoch": 0.37634526685701736, + "grad_norm": 2.2377769947052, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6963374614715576, + "num_tokens": 88247053.0, + "step": 3427 + }, + { + "epoch": 0.376455084559631, + "grad_norm": 2.1730618476867676, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.7014725804328918, + "num_tokens": 88274293.0, + "step": 3428 + }, + { + "epoch": 0.37656490226224465, + "grad_norm": 2.3971872329711914, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7224165797233582, + "num_tokens": 88295856.0, + "step": 3429 + }, + { + "epoch": 0.37667471996485835, + "grad_norm": 2.0889058113098145, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6833592057228088, + "num_tokens": 88325898.0, + "step": 3430 + }, + { + "epoch": 0.376784537667472, + "grad_norm": 2.0083749294281006, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7130009531974792, + "num_tokens": 88355755.0, + "step": 3431 + }, + { + "epoch": 0.37689435537008564, + "grad_norm": 2.6014716625213623, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7079187631607056, + "num_tokens": 88375654.0, + "step": 3432 + }, + { + "epoch": 0.37700417307269934, + "grad_norm": 2.073579788208008, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6902903914451599, + "num_tokens": 88404196.0, + "step": 3433 + }, + { + "epoch": 0.377113990775313, + "grad_norm": 2.278059720993042, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6940693855285645, + "num_tokens": 88427832.0, + "step": 3434 + }, + { + "epoch": 0.37722380847792664, + "grad_norm": 2.431736469268799, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7090588808059692, + "num_tokens": 88449302.0, + "step": 3435 + }, + { + "epoch": 0.3773336261805403, + "grad_norm": 2.3138861656188965, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6936659216880798, + "num_tokens": 88472008.0, + "step": 3436 + }, + { + "epoch": 0.377443443883154, + "grad_norm": 2.197842597961426, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7028018236160278, + "num_tokens": 88498164.0, + "step": 3437 + }, + { + "epoch": 0.37755326158576763, + "grad_norm": 2.0090384483337402, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7095431089401245, + "num_tokens": 88528626.0, + "step": 3438 + }, + { + "epoch": 0.3776630792883813, + "grad_norm": 2.3374202251434326, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7108914852142334, + "num_tokens": 88552963.0, + "step": 3439 + }, + { + "epoch": 0.3777728969909949, + "grad_norm": 1.8857221603393555, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6878234148025513, + "num_tokens": 88585426.0, + "step": 3440 + }, + { + "epoch": 0.3778827146936086, + "grad_norm": 2.162278413772583, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6945904493331909, + "num_tokens": 88613691.0, + "step": 3441 + }, + { + "epoch": 0.37799253239622227, + "grad_norm": 2.3984897136688232, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6931809782981873, + "num_tokens": 88637369.0, + "step": 3442 + }, + { + "epoch": 0.3781023500988359, + "grad_norm": 2.2868924140930176, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6891874074935913, + "num_tokens": 88663252.0, + "step": 3443 + }, + { + "epoch": 0.3782121678014496, + "grad_norm": 2.3542919158935547, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6948990821838379, + "num_tokens": 88686349.0, + "step": 3444 + }, + { + "epoch": 0.37832198550406326, + "grad_norm": 2.101238250732422, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6915336847305298, + "num_tokens": 88712504.0, + "step": 3445 + }, + { + "epoch": 0.3784318032066769, + "grad_norm": 2.2027370929718018, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6789867877960205, + "num_tokens": 88740358.0, + "step": 3446 + }, + { + "epoch": 0.37854162090929055, + "grad_norm": 2.294611692428589, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6955562233924866, + "num_tokens": 88763491.0, + "step": 3447 + }, + { + "epoch": 0.37865143861190426, + "grad_norm": 2.242476224899292, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6952078938484192, + "num_tokens": 88790437.0, + "step": 3448 + }, + { + "epoch": 0.3787612563145179, + "grad_norm": 2.460667371749878, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7407021522521973, + "num_tokens": 88813464.0, + "step": 3449 + }, + { + "epoch": 0.37887107401713155, + "grad_norm": 2.3663339614868164, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.7021174430847168, + "num_tokens": 88836524.0, + "step": 3450 + }, + { + "epoch": 0.37898089171974525, + "grad_norm": 2.2547531127929688, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6792424917221069, + "num_tokens": 88861276.0, + "step": 3451 + }, + { + "epoch": 0.3790907094223589, + "grad_norm": 2.0466229915618896, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7038887739181519, + "num_tokens": 88891443.0, + "step": 3452 + }, + { + "epoch": 0.37920052712497254, + "grad_norm": 1.9725397825241089, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6773363947868347, + "num_tokens": 88923587.0, + "step": 3453 + }, + { + "epoch": 0.3793103448275862, + "grad_norm": 2.6843645572662354, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7044295072555542, + "num_tokens": 88944282.0, + "step": 3454 + }, + { + "epoch": 0.3794201625301999, + "grad_norm": 2.416383981704712, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7133075594902039, + "num_tokens": 88966147.0, + "step": 3455 + }, + { + "epoch": 0.37952998023281354, + "grad_norm": 2.2548646926879883, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7234417200088501, + "num_tokens": 88990089.0, + "step": 3456 + }, + { + "epoch": 0.3796397979354272, + "grad_norm": 2.5342116355895996, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7111706137657166, + "num_tokens": 89011012.0, + "step": 3457 + }, + { + "epoch": 0.3797496156380408, + "grad_norm": 2.3364439010620117, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7018685340881348, + "num_tokens": 89034778.0, + "step": 3458 + }, + { + "epoch": 0.37985943334065453, + "grad_norm": 2.5921967029571533, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.7058873772621155, + "num_tokens": 89056876.0, + "step": 3459 + }, + { + "epoch": 0.3799692510432682, + "grad_norm": 2.5040876865386963, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7135415077209473, + "num_tokens": 89079355.0, + "step": 3460 + }, + { + "epoch": 0.3800790687458818, + "grad_norm": 2.3038039207458496, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6893093585968018, + "num_tokens": 89106374.0, + "step": 3461 + }, + { + "epoch": 0.3801888864484955, + "grad_norm": 2.0094008445739746, + "learning_rate": 1e-06, + "loss": 1.0646, + "mean_token_accuracy": 0.6801810264587402, + "num_tokens": 89138547.0, + "step": 3462 + }, + { + "epoch": 0.38029870415110917, + "grad_norm": 2.3627402782440186, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7230643630027771, + "num_tokens": 89162699.0, + "step": 3463 + }, + { + "epoch": 0.3804085218537228, + "grad_norm": 2.6443021297454834, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7042158842086792, + "num_tokens": 89183103.0, + "step": 3464 + }, + { + "epoch": 0.38051833955633646, + "grad_norm": 2.2489397525787354, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7261700630187988, + "num_tokens": 89209728.0, + "step": 3465 + }, + { + "epoch": 0.38062815725895016, + "grad_norm": 2.183021306991577, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7191489934921265, + "num_tokens": 89234306.0, + "step": 3466 + }, + { + "epoch": 0.3807379749615638, + "grad_norm": 2.073395013809204, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6835938692092896, + "num_tokens": 89265234.0, + "step": 3467 + }, + { + "epoch": 0.38084779266417745, + "grad_norm": 2.4575459957122803, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7058525681495667, + "num_tokens": 89288639.0, + "step": 3468 + }, + { + "epoch": 0.3809576103667911, + "grad_norm": 2.0601840019226074, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7140152454376221, + "num_tokens": 89315730.0, + "step": 3469 + }, + { + "epoch": 0.3810674280694048, + "grad_norm": 1.9337178468704224, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7389520406723022, + "num_tokens": 89342233.0, + "step": 3470 + }, + { + "epoch": 0.38117724577201845, + "grad_norm": 2.256049871444702, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7142939567565918, + "num_tokens": 89365830.0, + "step": 3471 + }, + { + "epoch": 0.3812870634746321, + "grad_norm": 2.32898211479187, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6945508122444153, + "num_tokens": 89390299.0, + "step": 3472 + }, + { + "epoch": 0.3813968811772458, + "grad_norm": 2.2146782875061035, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.718235969543457, + "num_tokens": 89413288.0, + "step": 3473 + }, + { + "epoch": 0.38150669887985944, + "grad_norm": 2.3852691650390625, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6978156566619873, + "num_tokens": 89437704.0, + "step": 3474 + }, + { + "epoch": 0.3816165165824731, + "grad_norm": 2.0993504524230957, + "learning_rate": 1e-06, + "loss": 1.0657, + "mean_token_accuracy": 0.6858022212982178, + "num_tokens": 89466775.0, + "step": 3475 + }, + { + "epoch": 0.38172633428508673, + "grad_norm": 2.0479109287261963, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6892609596252441, + "num_tokens": 89495955.0, + "step": 3476 + }, + { + "epoch": 0.38183615198770043, + "grad_norm": 2.364490509033203, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6951836943626404, + "num_tokens": 89518808.0, + "step": 3477 + }, + { + "epoch": 0.3819459696903141, + "grad_norm": 2.13693904876709, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7214158773422241, + "num_tokens": 89545487.0, + "step": 3478 + }, + { + "epoch": 0.3820557873929277, + "grad_norm": 2.346406936645508, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7227944135665894, + "num_tokens": 89566936.0, + "step": 3479 + }, + { + "epoch": 0.3821656050955414, + "grad_norm": 2.1830594539642334, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6872179508209229, + "num_tokens": 89592919.0, + "step": 3480 + }, + { + "epoch": 0.3822754227981551, + "grad_norm": 2.107757329940796, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7043573260307312, + "num_tokens": 89620914.0, + "step": 3481 + }, + { + "epoch": 0.3823852405007687, + "grad_norm": 2.3681678771972656, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7050265669822693, + "num_tokens": 89642765.0, + "step": 3482 + }, + { + "epoch": 0.38249505820338237, + "grad_norm": 2.380357027053833, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6800386905670166, + "num_tokens": 89667486.0, + "step": 3483 + }, + { + "epoch": 0.38260487590599607, + "grad_norm": 2.0776379108428955, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7056882977485657, + "num_tokens": 89695062.0, + "step": 3484 + }, + { + "epoch": 0.3827146936086097, + "grad_norm": 2.275710344314575, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7240554690361023, + "num_tokens": 89717941.0, + "step": 3485 + }, + { + "epoch": 0.38282451131122336, + "grad_norm": 2.101125955581665, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6988851428031921, + "num_tokens": 89746136.0, + "step": 3486 + }, + { + "epoch": 0.382934329013837, + "grad_norm": 2.0551605224609375, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7146134972572327, + "num_tokens": 89774276.0, + "step": 3487 + }, + { + "epoch": 0.3830441467164507, + "grad_norm": 1.9761743545532227, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6962085962295532, + "num_tokens": 89806387.0, + "step": 3488 + }, + { + "epoch": 0.38315396441906435, + "grad_norm": 2.221921443939209, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6993411779403687, + "num_tokens": 89832941.0, + "step": 3489 + }, + { + "epoch": 0.383263782121678, + "grad_norm": 2.153264045715332, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7041105628013611, + "num_tokens": 89858691.0, + "step": 3490 + }, + { + "epoch": 0.3833735998242917, + "grad_norm": 2.315650224685669, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.72429358959198, + "num_tokens": 89880836.0, + "step": 3491 + }, + { + "epoch": 0.38348341752690535, + "grad_norm": 2.078195571899414, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7281646132469177, + "num_tokens": 89909825.0, + "step": 3492 + }, + { + "epoch": 0.383593235229519, + "grad_norm": 2.2355847358703613, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7189642190933228, + "num_tokens": 89934297.0, + "step": 3493 + }, + { + "epoch": 0.38370305293213264, + "grad_norm": 2.07309627532959, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7069437503814697, + "num_tokens": 89963996.0, + "step": 3494 + }, + { + "epoch": 0.38381287063474634, + "grad_norm": 2.09485125541687, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6782433986663818, + "num_tokens": 89994692.0, + "step": 3495 + }, + { + "epoch": 0.38392268833736, + "grad_norm": 2.2040297985076904, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6929159760475159, + "num_tokens": 90020217.0, + "step": 3496 + }, + { + "epoch": 0.38403250603997363, + "grad_norm": 2.339945077896118, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7019661664962769, + "num_tokens": 90043014.0, + "step": 3497 + }, + { + "epoch": 0.38414232374258733, + "grad_norm": 2.322660207748413, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7287642955780029, + "num_tokens": 90066133.0, + "step": 3498 + }, + { + "epoch": 0.384252141445201, + "grad_norm": 2.5241425037384033, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7120345830917358, + "num_tokens": 90086012.0, + "step": 3499 + }, + { + "epoch": 0.3843619591478146, + "grad_norm": 2.241565704345703, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.701253354549408, + "num_tokens": 90111927.0, + "step": 3500 + }, + { + "epoch": 0.38447177685042827, + "grad_norm": 2.0630764961242676, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7133824229240417, + "num_tokens": 90140260.0, + "step": 3501 + }, + { + "epoch": 0.38458159455304197, + "grad_norm": 2.445333242416382, + "learning_rate": 1e-06, + "loss": 1.076, + "mean_token_accuracy": 0.6808123588562012, + "num_tokens": 90163406.0, + "step": 3502 + }, + { + "epoch": 0.3846914122556556, + "grad_norm": 2.123701572418213, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7190933227539062, + "num_tokens": 90190270.0, + "step": 3503 + }, + { + "epoch": 0.38480122995826926, + "grad_norm": 1.9229850769042969, + "learning_rate": 1e-06, + "loss": 1.094, + "mean_token_accuracy": 0.6704475283622742, + "num_tokens": 90226670.0, + "step": 3504 + }, + { + "epoch": 0.3849110476608829, + "grad_norm": 2.298455238342285, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7214661836624146, + "num_tokens": 90249852.0, + "step": 3505 + }, + { + "epoch": 0.3850208653634966, + "grad_norm": 2.251013994216919, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7088277339935303, + "num_tokens": 90275019.0, + "step": 3506 + }, + { + "epoch": 0.38513068306611026, + "grad_norm": 2.5887441635131836, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7171315550804138, + "num_tokens": 90295975.0, + "step": 3507 + }, + { + "epoch": 0.3852405007687239, + "grad_norm": 2.063845157623291, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7136513590812683, + "num_tokens": 90324585.0, + "step": 3508 + }, + { + "epoch": 0.3853503184713376, + "grad_norm": 2.5762546062469482, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7442856431007385, + "num_tokens": 90345099.0, + "step": 3509 + }, + { + "epoch": 0.38546013617395125, + "grad_norm": 2.50175404548645, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.713223397731781, + "num_tokens": 90364521.0, + "step": 3510 + }, + { + "epoch": 0.3855699538765649, + "grad_norm": 2.339778423309326, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7044122219085693, + "num_tokens": 90387084.0, + "step": 3511 + }, + { + "epoch": 0.38567977157917854, + "grad_norm": 2.4183523654937744, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7007992267608643, + "num_tokens": 90409535.0, + "step": 3512 + }, + { + "epoch": 0.38578958928179224, + "grad_norm": 2.3030664920806885, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7106497287750244, + "num_tokens": 90431634.0, + "step": 3513 + }, + { + "epoch": 0.3858994069844059, + "grad_norm": 2.1457743644714355, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6947294473648071, + "num_tokens": 90458045.0, + "step": 3514 + }, + { + "epoch": 0.38600922468701954, + "grad_norm": 2.2520503997802734, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7077006697654724, + "num_tokens": 90486156.0, + "step": 3515 + }, + { + "epoch": 0.3861190423896332, + "grad_norm": 2.284411907196045, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.71038419008255, + "num_tokens": 90509677.0, + "step": 3516 + }, + { + "epoch": 0.3862288600922469, + "grad_norm": 2.6912102699279785, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.711735725402832, + "num_tokens": 90528092.0, + "step": 3517 + }, + { + "epoch": 0.38633867779486053, + "grad_norm": 2.069687604904175, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6947100162506104, + "num_tokens": 90557805.0, + "step": 3518 + }, + { + "epoch": 0.3864484954974742, + "grad_norm": 2.373473882675171, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7157780528068542, + "num_tokens": 90578252.0, + "step": 3519 + }, + { + "epoch": 0.3865583132000879, + "grad_norm": 2.4411680698394775, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7336275577545166, + "num_tokens": 90599054.0, + "step": 3520 + }, + { + "epoch": 0.3866681309027015, + "grad_norm": 1.909715175628662, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7206678986549377, + "num_tokens": 90633147.0, + "step": 3521 + }, + { + "epoch": 0.38677794860531517, + "grad_norm": 2.4380385875701904, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6980378031730652, + "num_tokens": 90656101.0, + "step": 3522 + }, + { + "epoch": 0.3868877663079288, + "grad_norm": 2.251840591430664, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6788962483406067, + "num_tokens": 90682255.0, + "step": 3523 + }, + { + "epoch": 0.3869975840105425, + "grad_norm": 2.269970417022705, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7081302404403687, + "num_tokens": 90712929.0, + "step": 3524 + }, + { + "epoch": 0.38710740171315616, + "grad_norm": 2.216632127761841, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7027339935302734, + "num_tokens": 90736727.0, + "step": 3525 + }, + { + "epoch": 0.3872172194157698, + "grad_norm": 1.9330627918243408, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.700454831123352, + "num_tokens": 90767977.0, + "step": 3526 + }, + { + "epoch": 0.3873270371183835, + "grad_norm": 2.0106492042541504, + "learning_rate": 1e-06, + "loss": 1.1385, + "mean_token_accuracy": 0.6613849401473999, + "num_tokens": 90798988.0, + "step": 3527 + }, + { + "epoch": 0.38743685482099716, + "grad_norm": 2.537795305252075, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6894819140434265, + "num_tokens": 90820119.0, + "step": 3528 + }, + { + "epoch": 0.3875466725236108, + "grad_norm": 2.4770498275756836, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6862285137176514, + "num_tokens": 90842651.0, + "step": 3529 + }, + { + "epoch": 0.38765649022622445, + "grad_norm": 2.3607656955718994, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7097314596176147, + "num_tokens": 90867605.0, + "step": 3530 + }, + { + "epoch": 0.38776630792883815, + "grad_norm": 2.220898151397705, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7225477695465088, + "num_tokens": 90893773.0, + "step": 3531 + }, + { + "epoch": 0.3878761256314518, + "grad_norm": 2.4786458015441895, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6883469820022583, + "num_tokens": 90917063.0, + "step": 3532 + }, + { + "epoch": 0.38798594333406544, + "grad_norm": 2.0173065662384033, + "learning_rate": 1e-06, + "loss": 1.0656, + "mean_token_accuracy": 0.6850155591964722, + "num_tokens": 90948129.0, + "step": 3533 + }, + { + "epoch": 0.3880957610366791, + "grad_norm": 2.4001643657684326, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7255740165710449, + "num_tokens": 90970257.0, + "step": 3534 + }, + { + "epoch": 0.3882055787392928, + "grad_norm": 2.507239580154419, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.724988579750061, + "num_tokens": 90990312.0, + "step": 3535 + }, + { + "epoch": 0.38831539644190644, + "grad_norm": 2.2227730751037598, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7018966674804688, + "num_tokens": 91017122.0, + "step": 3536 + }, + { + "epoch": 0.3884252141445201, + "grad_norm": 2.147735595703125, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7005479335784912, + "num_tokens": 91043299.0, + "step": 3537 + }, + { + "epoch": 0.3885350318471338, + "grad_norm": 2.4347617626190186, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7003486156463623, + "num_tokens": 91065288.0, + "step": 3538 + }, + { + "epoch": 0.38864484954974743, + "grad_norm": 2.237884044647217, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6950889825820923, + "num_tokens": 91091077.0, + "step": 3539 + }, + { + "epoch": 0.3887546672523611, + "grad_norm": 1.7998605966567993, + "learning_rate": 1e-06, + "loss": 1.1037, + "mean_token_accuracy": 0.6726162433624268, + "num_tokens": 91130820.0, + "step": 3540 + }, + { + "epoch": 0.3888644849549747, + "grad_norm": 2.1426026821136475, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7098388671875, + "num_tokens": 91157675.0, + "step": 3541 + }, + { + "epoch": 0.3889743026575884, + "grad_norm": 1.8291634321212769, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6908218264579773, + "num_tokens": 91190512.0, + "step": 3542 + }, + { + "epoch": 0.38908412036020207, + "grad_norm": 2.4491806030273438, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.72247314453125, + "num_tokens": 91210538.0, + "step": 3543 + }, + { + "epoch": 0.3891939380628157, + "grad_norm": 2.118685483932495, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.691694974899292, + "num_tokens": 91238867.0, + "step": 3544 + }, + { + "epoch": 0.38930375576542936, + "grad_norm": 2.0894250869750977, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6906780004501343, + "num_tokens": 91267379.0, + "step": 3545 + }, + { + "epoch": 0.38941357346804306, + "grad_norm": 2.2660317420959473, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7125920653343201, + "num_tokens": 91291941.0, + "step": 3546 + }, + { + "epoch": 0.3895233911706567, + "grad_norm": 2.0827088356018066, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6788144111633301, + "num_tokens": 91320825.0, + "step": 3547 + }, + { + "epoch": 0.38963320887327035, + "grad_norm": 2.2752490043640137, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7225009202957153, + "num_tokens": 91344177.0, + "step": 3548 + }, + { + "epoch": 0.38974302657588406, + "grad_norm": 2.0900590419769287, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.6877598762512207, + "num_tokens": 91373432.0, + "step": 3549 + }, + { + "epoch": 0.3898528442784977, + "grad_norm": 2.2247474193573, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.702979564666748, + "num_tokens": 91400063.0, + "step": 3550 + }, + { + "epoch": 0.38996266198111135, + "grad_norm": 1.9759286642074585, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6950226426124573, + "num_tokens": 91429706.0, + "step": 3551 + }, + { + "epoch": 0.390072479683725, + "grad_norm": 2.10957932472229, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6985490322113037, + "num_tokens": 91457159.0, + "step": 3552 + }, + { + "epoch": 0.3901822973863387, + "grad_norm": 2.5779950618743896, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.722496509552002, + "num_tokens": 91475911.0, + "step": 3553 + }, + { + "epoch": 0.39029211508895234, + "grad_norm": 2.6135921478271484, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7397744655609131, + "num_tokens": 91494545.0, + "step": 3554 + }, + { + "epoch": 0.390401932791566, + "grad_norm": 2.2044339179992676, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7172306776046753, + "num_tokens": 91519304.0, + "step": 3555 + }, + { + "epoch": 0.3905117504941797, + "grad_norm": 2.2610554695129395, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7146161794662476, + "num_tokens": 91543697.0, + "step": 3556 + }, + { + "epoch": 0.39062156819679333, + "grad_norm": 2.306256055831909, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6917064189910889, + "num_tokens": 91567315.0, + "step": 3557 + }, + { + "epoch": 0.390731385899407, + "grad_norm": 2.3380932807922363, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6963925361633301, + "num_tokens": 91590299.0, + "step": 3558 + }, + { + "epoch": 0.3908412036020206, + "grad_norm": 2.022899627685547, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6867475509643555, + "num_tokens": 91620412.0, + "step": 3559 + }, + { + "epoch": 0.3909510213046343, + "grad_norm": 2.13698148727417, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6796591281890869, + "num_tokens": 91649181.0, + "step": 3560 + }, + { + "epoch": 0.391060839007248, + "grad_norm": 2.6601974964141846, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7313377857208252, + "num_tokens": 91666729.0, + "step": 3561 + }, + { + "epoch": 0.3911706567098616, + "grad_norm": 2.292009115219116, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7043370008468628, + "num_tokens": 91689595.0, + "step": 3562 + }, + { + "epoch": 0.39128047441247527, + "grad_norm": 2.3053276538848877, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7114073038101196, + "num_tokens": 91713542.0, + "step": 3563 + }, + { + "epoch": 0.39139029211508897, + "grad_norm": 2.2978930473327637, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7233989238739014, + "num_tokens": 91737726.0, + "step": 3564 + }, + { + "epoch": 0.3915001098177026, + "grad_norm": 2.472689628601074, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7046963572502136, + "num_tokens": 91759239.0, + "step": 3565 + }, + { + "epoch": 0.39160992752031626, + "grad_norm": 2.113086223602295, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.693436861038208, + "num_tokens": 91787781.0, + "step": 3566 + }, + { + "epoch": 0.39171974522292996, + "grad_norm": 2.0696215629577637, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.714491605758667, + "num_tokens": 91817764.0, + "step": 3567 + }, + { + "epoch": 0.3918295629255436, + "grad_norm": 1.977993130683899, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6938016414642334, + "num_tokens": 91851139.0, + "step": 3568 + }, + { + "epoch": 0.39193938062815725, + "grad_norm": 2.2318053245544434, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7183026075363159, + "num_tokens": 91874900.0, + "step": 3569 + }, + { + "epoch": 0.3920491983307709, + "grad_norm": 2.3177671432495117, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6936722993850708, + "num_tokens": 91898046.0, + "step": 3570 + }, + { + "epoch": 0.3921590160333846, + "grad_norm": 2.4996204376220703, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7052870392799377, + "num_tokens": 91919080.0, + "step": 3571 + }, + { + "epoch": 0.39226883373599825, + "grad_norm": 2.238651752471924, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7032803893089294, + "num_tokens": 91943844.0, + "step": 3572 + }, + { + "epoch": 0.3923786514386119, + "grad_norm": 2.4902355670928955, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7062183618545532, + "num_tokens": 91965998.0, + "step": 3573 + }, + { + "epoch": 0.3924884691412256, + "grad_norm": 2.540235757827759, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7093517780303955, + "num_tokens": 91985474.0, + "step": 3574 + }, + { + "epoch": 0.39259828684383924, + "grad_norm": 2.358745574951172, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7027892470359802, + "num_tokens": 92009550.0, + "step": 3575 + }, + { + "epoch": 0.3927081045464529, + "grad_norm": 2.567380428314209, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7206371426582336, + "num_tokens": 92029557.0, + "step": 3576 + }, + { + "epoch": 0.39281792224906653, + "grad_norm": 2.483886480331421, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6983450651168823, + "num_tokens": 92051707.0, + "step": 3577 + }, + { + "epoch": 0.39292773995168023, + "grad_norm": 2.226285219192505, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6899118423461914, + "num_tokens": 92080383.0, + "step": 3578 + }, + { + "epoch": 0.3930375576542939, + "grad_norm": 1.9671876430511475, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6955300569534302, + "num_tokens": 92110793.0, + "step": 3579 + }, + { + "epoch": 0.3931473753569075, + "grad_norm": 2.0890376567840576, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6979076266288757, + "num_tokens": 92139882.0, + "step": 3580 + }, + { + "epoch": 0.39325719305952117, + "grad_norm": 2.2380495071411133, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7090719938278198, + "num_tokens": 92165638.0, + "step": 3581 + }, + { + "epoch": 0.39336701076213487, + "grad_norm": 2.3486154079437256, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6982163786888123, + "num_tokens": 92189447.0, + "step": 3582 + }, + { + "epoch": 0.3934768284647485, + "grad_norm": 2.1784756183624268, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7031977772712708, + "num_tokens": 92213698.0, + "step": 3583 + }, + { + "epoch": 0.39358664616736216, + "grad_norm": 2.0635986328125, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6918898820877075, + "num_tokens": 92241696.0, + "step": 3584 + }, + { + "epoch": 0.39369646386997587, + "grad_norm": 2.198195219039917, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6819026470184326, + "num_tokens": 92268772.0, + "step": 3585 + }, + { + "epoch": 0.3938062815725895, + "grad_norm": 2.296678066253662, + "learning_rate": 1e-06, + "loss": 1.0893, + "mean_token_accuracy": 0.672925591468811, + "num_tokens": 92294046.0, + "step": 3586 + }, + { + "epoch": 0.39391609927520316, + "grad_norm": 2.087521553039551, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.704232931137085, + "num_tokens": 92322413.0, + "step": 3587 + }, + { + "epoch": 0.3940259169778168, + "grad_norm": 2.021341323852539, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.693289041519165, + "num_tokens": 92354062.0, + "step": 3588 + }, + { + "epoch": 0.3941357346804305, + "grad_norm": 2.0638701915740967, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7082812190055847, + "num_tokens": 92381885.0, + "step": 3589 + }, + { + "epoch": 0.39424555238304415, + "grad_norm": 2.077690839767456, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6919697523117065, + "num_tokens": 92408723.0, + "step": 3590 + }, + { + "epoch": 0.3943553700856578, + "grad_norm": 2.2178468704223633, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7037203907966614, + "num_tokens": 92434105.0, + "step": 3591 + }, + { + "epoch": 0.39446518778827144, + "grad_norm": 2.031405448913574, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7020469903945923, + "num_tokens": 92464043.0, + "step": 3592 + }, + { + "epoch": 0.39457500549088514, + "grad_norm": 2.275876760482788, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6914067268371582, + "num_tokens": 92487477.0, + "step": 3593 + }, + { + "epoch": 0.3946848231934988, + "grad_norm": 2.263911008834839, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6982495784759521, + "num_tokens": 92514272.0, + "step": 3594 + }, + { + "epoch": 0.39479464089611244, + "grad_norm": 2.1094448566436768, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6963660717010498, + "num_tokens": 92542800.0, + "step": 3595 + }, + { + "epoch": 0.39490445859872614, + "grad_norm": 2.0281989574432373, + "learning_rate": 1e-06, + "loss": 1.1086, + "mean_token_accuracy": 0.6665133833885193, + "num_tokens": 92575750.0, + "step": 3596 + }, + { + "epoch": 0.3950142763013398, + "grad_norm": 2.0032386779785156, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6957155466079712, + "num_tokens": 92606136.0, + "step": 3597 + }, + { + "epoch": 0.39512409400395343, + "grad_norm": 1.9282978773117065, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6865698099136353, + "num_tokens": 92636642.0, + "step": 3598 + }, + { + "epoch": 0.3952339117065671, + "grad_norm": 2.2199342250823975, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6989527940750122, + "num_tokens": 92661124.0, + "step": 3599 + }, + { + "epoch": 0.3953437294091808, + "grad_norm": 2.1834752559661865, + "learning_rate": 1e-06, + "loss": 1.0692, + "mean_token_accuracy": 0.6809617280960083, + "num_tokens": 92689213.0, + "step": 3600 + }, + { + "epoch": 0.3954535471117944, + "grad_norm": 2.2152769565582275, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7109925150871277, + "num_tokens": 92716517.0, + "step": 3601 + }, + { + "epoch": 0.39556336481440807, + "grad_norm": 2.25822377204895, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6885975003242493, + "num_tokens": 92743999.0, + "step": 3602 + }, + { + "epoch": 0.39567318251702177, + "grad_norm": 2.119368076324463, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6958199739456177, + "num_tokens": 92771207.0, + "step": 3603 + }, + { + "epoch": 0.3957830002196354, + "grad_norm": 2.4001946449279785, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6888555288314819, + "num_tokens": 92794309.0, + "step": 3604 + }, + { + "epoch": 0.39589281792224906, + "grad_norm": 2.0630998611450195, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6856698989868164, + "num_tokens": 92824820.0, + "step": 3605 + }, + { + "epoch": 0.3960026356248627, + "grad_norm": 2.269782543182373, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6943691968917847, + "num_tokens": 92849013.0, + "step": 3606 + }, + { + "epoch": 0.3961124533274764, + "grad_norm": 2.445321798324585, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.67942214012146, + "num_tokens": 92870410.0, + "step": 3607 + }, + { + "epoch": 0.39622227103009006, + "grad_norm": 2.3522932529449463, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6957166790962219, + "num_tokens": 92892372.0, + "step": 3608 + }, + { + "epoch": 0.3963320887327037, + "grad_norm": 2.2632083892822266, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7143383026123047, + "num_tokens": 92917313.0, + "step": 3609 + }, + { + "epoch": 0.39644190643531735, + "grad_norm": 2.193026542663574, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6839479804039001, + "num_tokens": 92942876.0, + "step": 3610 + }, + { + "epoch": 0.39655172413793105, + "grad_norm": 2.454624891281128, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.701173722743988, + "num_tokens": 92963521.0, + "step": 3611 + }, + { + "epoch": 0.3966615418405447, + "grad_norm": 2.479565382003784, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7070918083190918, + "num_tokens": 92989271.0, + "step": 3612 + }, + { + "epoch": 0.39677135954315834, + "grad_norm": 2.085468053817749, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6928802132606506, + "num_tokens": 93017996.0, + "step": 3613 + }, + { + "epoch": 0.39688117724577204, + "grad_norm": 2.2177321910858154, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.6844146251678467, + "num_tokens": 93046707.0, + "step": 3614 + }, + { + "epoch": 0.3969909949483857, + "grad_norm": 1.9619351625442505, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6827747225761414, + "num_tokens": 93079516.0, + "step": 3615 + }, + { + "epoch": 0.39710081265099934, + "grad_norm": 2.28456711769104, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.685240626335144, + "num_tokens": 93105516.0, + "step": 3616 + }, + { + "epoch": 0.397210630353613, + "grad_norm": 2.0224828720092773, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.704889714717865, + "num_tokens": 93136530.0, + "step": 3617 + }, + { + "epoch": 0.3973204480562267, + "grad_norm": 2.2424230575561523, + "learning_rate": 1e-06, + "loss": 1.0783, + "mean_token_accuracy": 0.6829177141189575, + "num_tokens": 93163555.0, + "step": 3618 + }, + { + "epoch": 0.39743026575884033, + "grad_norm": 2.1267452239990234, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7153828740119934, + "num_tokens": 93189177.0, + "step": 3619 + }, + { + "epoch": 0.397540083461454, + "grad_norm": 2.0835185050964355, + "learning_rate": 1e-06, + "loss": 1.0977, + "mean_token_accuracy": 0.6776567697525024, + "num_tokens": 93219534.0, + "step": 3620 + }, + { + "epoch": 0.3976499011640676, + "grad_norm": 2.2803516387939453, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.723333477973938, + "num_tokens": 93239971.0, + "step": 3621 + }, + { + "epoch": 0.3977597188666813, + "grad_norm": 2.3940627574920654, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7100257873535156, + "num_tokens": 93261695.0, + "step": 3622 + }, + { + "epoch": 0.39786953656929497, + "grad_norm": 2.24839186668396, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.709475040435791, + "num_tokens": 93285569.0, + "step": 3623 + }, + { + "epoch": 0.3979793542719086, + "grad_norm": 2.331878423690796, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7070780396461487, + "num_tokens": 93309084.0, + "step": 3624 + }, + { + "epoch": 0.3980891719745223, + "grad_norm": 2.010627508163452, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6986174583435059, + "num_tokens": 93340728.0, + "step": 3625 + }, + { + "epoch": 0.39819898967713596, + "grad_norm": 1.9580495357513428, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6896547079086304, + "num_tokens": 93372869.0, + "step": 3626 + }, + { + "epoch": 0.3983088073797496, + "grad_norm": 2.0774145126342773, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7166056632995605, + "num_tokens": 93399782.0, + "step": 3627 + }, + { + "epoch": 0.39841862508236325, + "grad_norm": 2.238173246383667, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.737703800201416, + "num_tokens": 93423654.0, + "step": 3628 + }, + { + "epoch": 0.39852844278497696, + "grad_norm": 2.1314826011657715, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6990289092063904, + "num_tokens": 93449836.0, + "step": 3629 + }, + { + "epoch": 0.3986382604875906, + "grad_norm": 2.0792276859283447, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7126315236091614, + "num_tokens": 93479387.0, + "step": 3630 + }, + { + "epoch": 0.39874807819020425, + "grad_norm": 2.2106730937957764, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6999642252922058, + "num_tokens": 93505884.0, + "step": 3631 + }, + { + "epoch": 0.39885789589281795, + "grad_norm": 2.153179883956909, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.697621762752533, + "num_tokens": 93532442.0, + "step": 3632 + }, + { + "epoch": 0.3989677135954316, + "grad_norm": 2.5950193405151367, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6946507096290588, + "num_tokens": 93552598.0, + "step": 3633 + }, + { + "epoch": 0.39907753129804524, + "grad_norm": 2.1510813236236572, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7079944014549255, + "num_tokens": 93576947.0, + "step": 3634 + }, + { + "epoch": 0.3991873490006589, + "grad_norm": 1.965328335762024, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6994497179985046, + "num_tokens": 93606546.0, + "step": 3635 + }, + { + "epoch": 0.3992971667032726, + "grad_norm": 2.587324857711792, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.721015214920044, + "num_tokens": 93625832.0, + "step": 3636 + }, + { + "epoch": 0.39940698440588623, + "grad_norm": 1.9802981615066528, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.6914386749267578, + "num_tokens": 93659980.0, + "step": 3637 + }, + { + "epoch": 0.3995168021084999, + "grad_norm": 2.055715799331665, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6912530660629272, + "num_tokens": 93690296.0, + "step": 3638 + }, + { + "epoch": 0.3996266198111135, + "grad_norm": 2.3145687580108643, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7079617381095886, + "num_tokens": 93712743.0, + "step": 3639 + }, + { + "epoch": 0.3997364375137272, + "grad_norm": 2.0253255367279053, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6934235095977783, + "num_tokens": 93743396.0, + "step": 3640 + }, + { + "epoch": 0.3998462552163409, + "grad_norm": 2.0580925941467285, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7130406498908997, + "num_tokens": 93771263.0, + "step": 3641 + }, + { + "epoch": 0.3999560729189545, + "grad_norm": 2.2121849060058594, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7163969278335571, + "num_tokens": 93795946.0, + "step": 3642 + }, + { + "epoch": 0.4000658906215682, + "grad_norm": 1.9978669881820679, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7168579697608948, + "num_tokens": 93825862.0, + "step": 3643 + }, + { + "epoch": 0.40017570832418187, + "grad_norm": 2.0828700065612793, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6996592283248901, + "num_tokens": 93854483.0, + "step": 3644 + }, + { + "epoch": 0.4002855260267955, + "grad_norm": 1.9045850038528442, + "learning_rate": 1e-06, + "loss": 1.1285, + "mean_token_accuracy": 0.6707102060317993, + "num_tokens": 93889491.0, + "step": 3645 + }, + { + "epoch": 0.40039534372940916, + "grad_norm": 2.3553626537323, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7091149687767029, + "num_tokens": 93911468.0, + "step": 3646 + }, + { + "epoch": 0.40050516143202286, + "grad_norm": 2.2407126426696777, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7111337184906006, + "num_tokens": 93936579.0, + "step": 3647 + }, + { + "epoch": 0.4006149791346365, + "grad_norm": 2.0965728759765625, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.686086893081665, + "num_tokens": 93964534.0, + "step": 3648 + }, + { + "epoch": 0.40072479683725015, + "grad_norm": 2.2118940353393555, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6829646229743958, + "num_tokens": 93990242.0, + "step": 3649 + }, + { + "epoch": 0.4008346145398638, + "grad_norm": 2.4559359550476074, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6895607709884644, + "num_tokens": 94011235.0, + "step": 3650 + }, + { + "epoch": 0.4009444322424775, + "grad_norm": 2.1562705039978027, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6816772222518921, + "num_tokens": 94038333.0, + "step": 3651 + }, + { + "epoch": 0.40105424994509115, + "grad_norm": 2.074068069458008, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.700969934463501, + "num_tokens": 94065698.0, + "step": 3652 + }, + { + "epoch": 0.4011640676477048, + "grad_norm": 2.1477041244506836, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7088913917541504, + "num_tokens": 94093757.0, + "step": 3653 + }, + { + "epoch": 0.4012738853503185, + "grad_norm": 2.118433713912964, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7051883935928345, + "num_tokens": 94123844.0, + "step": 3654 + }, + { + "epoch": 0.40138370305293214, + "grad_norm": 2.2809860706329346, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6950662136077881, + "num_tokens": 94147954.0, + "step": 3655 + }, + { + "epoch": 0.4014935207555458, + "grad_norm": 2.2295820713043213, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7085312604904175, + "num_tokens": 94171823.0, + "step": 3656 + }, + { + "epoch": 0.40160333845815943, + "grad_norm": 2.1566808223724365, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7043213248252869, + "num_tokens": 94197995.0, + "step": 3657 + }, + { + "epoch": 0.40171315616077313, + "grad_norm": 2.1053378582000732, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6840336322784424, + "num_tokens": 94226359.0, + "step": 3658 + }, + { + "epoch": 0.4018229738633868, + "grad_norm": 2.2941250801086426, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.715013861656189, + "num_tokens": 94251641.0, + "step": 3659 + }, + { + "epoch": 0.4019327915660004, + "grad_norm": 2.328339099884033, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7086614966392517, + "num_tokens": 94277194.0, + "step": 3660 + }, + { + "epoch": 0.4020426092686141, + "grad_norm": 1.9468271732330322, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6984882354736328, + "num_tokens": 94308504.0, + "step": 3661 + }, + { + "epoch": 0.40215242697122777, + "grad_norm": 2.325423240661621, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7144768834114075, + "num_tokens": 94331451.0, + "step": 3662 + }, + { + "epoch": 0.4022622446738414, + "grad_norm": 2.5589683055877686, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7252542972564697, + "num_tokens": 94350507.0, + "step": 3663 + }, + { + "epoch": 0.40237206237645506, + "grad_norm": 2.413774013519287, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7008006572723389, + "num_tokens": 94372176.0, + "step": 3664 + }, + { + "epoch": 0.40248188007906877, + "grad_norm": 2.5493791103363037, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7128518223762512, + "num_tokens": 94391258.0, + "step": 3665 + }, + { + "epoch": 0.4025916977816824, + "grad_norm": 1.9874203205108643, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6922917366027832, + "num_tokens": 94426879.0, + "step": 3666 + }, + { + "epoch": 0.40270151548429606, + "grad_norm": 2.739048957824707, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.693120539188385, + "num_tokens": 94445692.0, + "step": 3667 + }, + { + "epoch": 0.4028113331869097, + "grad_norm": 2.071213483810425, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6966071724891663, + "num_tokens": 94475374.0, + "step": 3668 + }, + { + "epoch": 0.4029211508895234, + "grad_norm": 2.1407973766326904, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.689093828201294, + "num_tokens": 94501373.0, + "step": 3669 + }, + { + "epoch": 0.40303096859213705, + "grad_norm": 1.9060440063476562, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6726747155189514, + "num_tokens": 94534397.0, + "step": 3670 + }, + { + "epoch": 0.4031407862947507, + "grad_norm": 2.2483959197998047, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7218717336654663, + "num_tokens": 94557790.0, + "step": 3671 + }, + { + "epoch": 0.4032506039973644, + "grad_norm": 2.019371271133423, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7052488327026367, + "num_tokens": 94587931.0, + "step": 3672 + }, + { + "epoch": 0.40336042169997804, + "grad_norm": 2.2086734771728516, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6854634284973145, + "num_tokens": 94616466.0, + "step": 3673 + }, + { + "epoch": 0.4034702394025917, + "grad_norm": 2.0834779739379883, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7102592587471008, + "num_tokens": 94642773.0, + "step": 3674 + }, + { + "epoch": 0.40358005710520534, + "grad_norm": 2.512051820755005, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6963992714881897, + "num_tokens": 94664160.0, + "step": 3675 + }, + { + "epoch": 0.40368987480781904, + "grad_norm": 2.4766201972961426, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7011744976043701, + "num_tokens": 94687708.0, + "step": 3676 + }, + { + "epoch": 0.4037996925104327, + "grad_norm": 2.315326690673828, + "learning_rate": 1e-06, + "loss": 1.1133, + "mean_token_accuracy": 0.6717232465744019, + "num_tokens": 94711153.0, + "step": 3677 + }, + { + "epoch": 0.40390951021304633, + "grad_norm": 2.5798821449279785, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7188975214958191, + "num_tokens": 94732099.0, + "step": 3678 + }, + { + "epoch": 0.40401932791566003, + "grad_norm": 1.850015640258789, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6832390427589417, + "num_tokens": 94769169.0, + "step": 3679 + }, + { + "epoch": 0.4041291456182737, + "grad_norm": 2.031099319458008, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6906529664993286, + "num_tokens": 94801759.0, + "step": 3680 + }, + { + "epoch": 0.4042389633208873, + "grad_norm": 2.258801221847534, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7042937278747559, + "num_tokens": 94824750.0, + "step": 3681 + }, + { + "epoch": 0.40434878102350097, + "grad_norm": 2.201022148132324, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7058336734771729, + "num_tokens": 94849043.0, + "step": 3682 + }, + { + "epoch": 0.40445859872611467, + "grad_norm": 2.0523412227630615, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6859833002090454, + "num_tokens": 94878140.0, + "step": 3683 + }, + { + "epoch": 0.4045684164287283, + "grad_norm": 2.321016788482666, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7064018249511719, + "num_tokens": 94900604.0, + "step": 3684 + }, + { + "epoch": 0.40467823413134196, + "grad_norm": 1.9442780017852783, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6951988935470581, + "num_tokens": 94933212.0, + "step": 3685 + }, + { + "epoch": 0.4047880518339556, + "grad_norm": 2.3617422580718994, + "learning_rate": 1e-06, + "loss": 1.0861, + "mean_token_accuracy": 0.681697428226471, + "num_tokens": 94958804.0, + "step": 3686 + }, + { + "epoch": 0.4048978695365693, + "grad_norm": 2.4171955585479736, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6973528265953064, + "num_tokens": 94980736.0, + "step": 3687 + }, + { + "epoch": 0.40500768723918296, + "grad_norm": 2.1538658142089844, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6924943327903748, + "num_tokens": 95007094.0, + "step": 3688 + }, + { + "epoch": 0.4051175049417966, + "grad_norm": 2.114805221557617, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7360572814941406, + "num_tokens": 95033218.0, + "step": 3689 + }, + { + "epoch": 0.4052273226444103, + "grad_norm": 2.3129210472106934, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7007640600204468, + "num_tokens": 95056199.0, + "step": 3690 + }, + { + "epoch": 0.40533714034702395, + "grad_norm": 2.2914819717407227, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7226579189300537, + "num_tokens": 95080357.0, + "step": 3691 + }, + { + "epoch": 0.4054469580496376, + "grad_norm": 2.8132944107055664, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.721849799156189, + "num_tokens": 95096506.0, + "step": 3692 + }, + { + "epoch": 0.40555677575225124, + "grad_norm": 1.998534083366394, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.69114089012146, + "num_tokens": 95127083.0, + "step": 3693 + }, + { + "epoch": 0.40566659345486494, + "grad_norm": 2.4784767627716064, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7185355424880981, + "num_tokens": 95147878.0, + "step": 3694 + }, + { + "epoch": 0.4057764111574786, + "grad_norm": 2.1035079956054688, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6934882402420044, + "num_tokens": 95175285.0, + "step": 3695 + }, + { + "epoch": 0.40588622886009224, + "grad_norm": 2.194692611694336, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7058338522911072, + "num_tokens": 95199011.0, + "step": 3696 + }, + { + "epoch": 0.4059960465627059, + "grad_norm": 2.2803330421447754, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6970314979553223, + "num_tokens": 95223094.0, + "step": 3697 + }, + { + "epoch": 0.4061058642653196, + "grad_norm": 2.3042540550231934, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6959124803543091, + "num_tokens": 95247926.0, + "step": 3698 + }, + { + "epoch": 0.40621568196793323, + "grad_norm": 2.2941431999206543, + "learning_rate": 1e-06, + "loss": 1.0907, + "mean_token_accuracy": 0.6836967468261719, + "num_tokens": 95275913.0, + "step": 3699 + }, + { + "epoch": 0.4063254996705469, + "grad_norm": 2.137377977371216, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7072567939758301, + "num_tokens": 95302711.0, + "step": 3700 + }, + { + "epoch": 0.4064353173731606, + "grad_norm": 2.2526514530181885, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6816175580024719, + "num_tokens": 95329283.0, + "step": 3701 + }, + { + "epoch": 0.4065451350757742, + "grad_norm": 2.3125200271606445, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7276009321212769, + "num_tokens": 95352071.0, + "step": 3702 + }, + { + "epoch": 0.40665495277838787, + "grad_norm": 2.3008875846862793, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6910898685455322, + "num_tokens": 95377521.0, + "step": 3703 + }, + { + "epoch": 0.4067647704810015, + "grad_norm": 2.016850709915161, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7066909074783325, + "num_tokens": 95407145.0, + "step": 3704 + }, + { + "epoch": 0.4068745881836152, + "grad_norm": 2.4048285484313965, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6961414217948914, + "num_tokens": 95430664.0, + "step": 3705 + }, + { + "epoch": 0.40698440588622886, + "grad_norm": 2.4818267822265625, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7246696949005127, + "num_tokens": 95451836.0, + "step": 3706 + }, + { + "epoch": 0.4070942235888425, + "grad_norm": 2.3042144775390625, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7111314535140991, + "num_tokens": 95474651.0, + "step": 3707 + }, + { + "epoch": 0.4072040412914562, + "grad_norm": 1.968442440032959, + "learning_rate": 1e-06, + "loss": 1.0788, + "mean_token_accuracy": 0.6890038847923279, + "num_tokens": 95505931.0, + "step": 3708 + }, + { + "epoch": 0.40731385899406986, + "grad_norm": 2.310256004333496, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6986455321311951, + "num_tokens": 95530341.0, + "step": 3709 + }, + { + "epoch": 0.4074236766966835, + "grad_norm": 2.357114315032959, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7061722874641418, + "num_tokens": 95552812.0, + "step": 3710 + }, + { + "epoch": 0.40753349439929715, + "grad_norm": 2.2956972122192383, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7042834162712097, + "num_tokens": 95577796.0, + "step": 3711 + }, + { + "epoch": 0.40764331210191085, + "grad_norm": 2.3858959674835205, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.714162290096283, + "num_tokens": 95599825.0, + "step": 3712 + }, + { + "epoch": 0.4077531298045245, + "grad_norm": 2.3549211025238037, + "learning_rate": 1e-06, + "loss": 1.0916, + "mean_token_accuracy": 0.679597020149231, + "num_tokens": 95624704.0, + "step": 3713 + }, + { + "epoch": 0.40786294750713814, + "grad_norm": 2.526183843612671, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7093125581741333, + "num_tokens": 95644308.0, + "step": 3714 + }, + { + "epoch": 0.4079727652097518, + "grad_norm": 2.182586431503296, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7101513147354126, + "num_tokens": 95670393.0, + "step": 3715 + }, + { + "epoch": 0.4080825829123655, + "grad_norm": 2.3126819133758545, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.7014023065567017, + "num_tokens": 95692246.0, + "step": 3716 + }, + { + "epoch": 0.40819240061497913, + "grad_norm": 2.5364818572998047, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7245289087295532, + "num_tokens": 95712128.0, + "step": 3717 + }, + { + "epoch": 0.4083022183175928, + "grad_norm": 2.022965669631958, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6966023445129395, + "num_tokens": 95740659.0, + "step": 3718 + }, + { + "epoch": 0.4084120360202065, + "grad_norm": 2.232442617416382, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6927534937858582, + "num_tokens": 95765828.0, + "step": 3719 + }, + { + "epoch": 0.4085218537228201, + "grad_norm": 2.189073085784912, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6800618171691895, + "num_tokens": 95790999.0, + "step": 3720 + }, + { + "epoch": 0.4086316714254338, + "grad_norm": 2.1469743251800537, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7169424295425415, + "num_tokens": 95816392.0, + "step": 3721 + }, + { + "epoch": 0.4087414891280474, + "grad_norm": 2.076184034347534, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6999377012252808, + "num_tokens": 95846815.0, + "step": 3722 + }, + { + "epoch": 0.4088513068306611, + "grad_norm": 2.270193338394165, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6995217800140381, + "num_tokens": 95870196.0, + "step": 3723 + }, + { + "epoch": 0.40896112453327477, + "grad_norm": 2.0129525661468506, + "learning_rate": 1e-06, + "loss": 1.0729, + "mean_token_accuracy": 0.6821051836013794, + "num_tokens": 95901053.0, + "step": 3724 + }, + { + "epoch": 0.4090709422358884, + "grad_norm": 2.0944716930389404, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7100579142570496, + "num_tokens": 95928625.0, + "step": 3725 + }, + { + "epoch": 0.40918075993850206, + "grad_norm": 2.252178192138672, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6847711205482483, + "num_tokens": 95954988.0, + "step": 3726 + }, + { + "epoch": 0.40929057764111576, + "grad_norm": 2.365863084793091, + "learning_rate": 1e-06, + "loss": 1.0653, + "mean_token_accuracy": 0.6829464435577393, + "num_tokens": 95980049.0, + "step": 3727 + }, + { + "epoch": 0.4094003953437294, + "grad_norm": 2.3179073333740234, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.7026302218437195, + "num_tokens": 96003022.0, + "step": 3728 + }, + { + "epoch": 0.40951021304634305, + "grad_norm": 2.053649425506592, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.691387414932251, + "num_tokens": 96030854.0, + "step": 3729 + }, + { + "epoch": 0.40962003074895675, + "grad_norm": 2.190577745437622, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6936055421829224, + "num_tokens": 96056734.0, + "step": 3730 + }, + { + "epoch": 0.4097298484515704, + "grad_norm": 2.2957746982574463, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6945995688438416, + "num_tokens": 96081119.0, + "step": 3731 + }, + { + "epoch": 0.40983966615418405, + "grad_norm": 2.2794742584228516, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7040607333183289, + "num_tokens": 96106571.0, + "step": 3732 + }, + { + "epoch": 0.4099494838567977, + "grad_norm": 2.3139283657073975, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.731468677520752, + "num_tokens": 96128030.0, + "step": 3733 + }, + { + "epoch": 0.4100593015594114, + "grad_norm": 2.546938180923462, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7337102890014648, + "num_tokens": 96147891.0, + "step": 3734 + }, + { + "epoch": 0.41016911926202504, + "grad_norm": 2.484452962875366, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.721305251121521, + "num_tokens": 96169172.0, + "step": 3735 + }, + { + "epoch": 0.4102789369646387, + "grad_norm": 2.3286750316619873, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.696382462978363, + "num_tokens": 96192297.0, + "step": 3736 + }, + { + "epoch": 0.4103887546672524, + "grad_norm": 2.3780527114868164, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.714701235294342, + "num_tokens": 96212726.0, + "step": 3737 + }, + { + "epoch": 0.41049857236986603, + "grad_norm": 2.138406276702881, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7307551503181458, + "num_tokens": 96238205.0, + "step": 3738 + }, + { + "epoch": 0.4106083900724797, + "grad_norm": 2.0975053310394287, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6840617656707764, + "num_tokens": 96266652.0, + "step": 3739 + }, + { + "epoch": 0.4107182077750933, + "grad_norm": 2.1842844486236572, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7060171961784363, + "num_tokens": 96291848.0, + "step": 3740 + }, + { + "epoch": 0.410828025477707, + "grad_norm": 2.4127566814422607, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6926912069320679, + "num_tokens": 96313041.0, + "step": 3741 + }, + { + "epoch": 0.41093784318032067, + "grad_norm": 2.1710970401763916, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7038641571998596, + "num_tokens": 96338764.0, + "step": 3742 + }, + { + "epoch": 0.4110476608829343, + "grad_norm": 2.1141433715820312, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.68756103515625, + "num_tokens": 96369544.0, + "step": 3743 + }, + { + "epoch": 0.41115747858554796, + "grad_norm": 2.1981096267700195, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6876819133758545, + "num_tokens": 96397739.0, + "step": 3744 + }, + { + "epoch": 0.41126729628816167, + "grad_norm": 2.3571033477783203, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7015185952186584, + "num_tokens": 96419764.0, + "step": 3745 + }, + { + "epoch": 0.4113771139907753, + "grad_norm": 2.1224405765533447, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7096221446990967, + "num_tokens": 96445588.0, + "step": 3746 + }, + { + "epoch": 0.41148693169338896, + "grad_norm": 2.07231068611145, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6976033449172974, + "num_tokens": 96474243.0, + "step": 3747 + }, + { + "epoch": 0.41159674939600266, + "grad_norm": 2.0817389488220215, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6913878321647644, + "num_tokens": 96502237.0, + "step": 3748 + }, + { + "epoch": 0.4117065670986163, + "grad_norm": 1.99656343460083, + "learning_rate": 1e-06, + "loss": 1.0822, + "mean_token_accuracy": 0.673953652381897, + "num_tokens": 96533446.0, + "step": 3749 + }, + { + "epoch": 0.41181638480122995, + "grad_norm": 2.3451292514801025, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6995447874069214, + "num_tokens": 96557137.0, + "step": 3750 + }, + { + "epoch": 0.4119262025038436, + "grad_norm": 2.0920827388763428, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6843113303184509, + "num_tokens": 96586255.0, + "step": 3751 + }, + { + "epoch": 0.4120360202064573, + "grad_norm": 1.825789213180542, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7109280824661255, + "num_tokens": 96620821.0, + "step": 3752 + }, + { + "epoch": 0.41214583790907094, + "grad_norm": 2.1827216148376465, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7032816410064697, + "num_tokens": 96646661.0, + "step": 3753 + }, + { + "epoch": 0.4122556556116846, + "grad_norm": 2.224889039993286, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.7041794061660767, + "num_tokens": 96671945.0, + "step": 3754 + }, + { + "epoch": 0.4123654733142983, + "grad_norm": 2.221724510192871, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7093380689620972, + "num_tokens": 96697857.0, + "step": 3755 + }, + { + "epoch": 0.41247529101691194, + "grad_norm": 2.0284721851348877, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.728538453578949, + "num_tokens": 96723928.0, + "step": 3756 + }, + { + "epoch": 0.4125851087195256, + "grad_norm": 2.401465654373169, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7063263058662415, + "num_tokens": 96747350.0, + "step": 3757 + }, + { + "epoch": 0.41269492642213923, + "grad_norm": 2.2706220149993896, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6859647035598755, + "num_tokens": 96775984.0, + "step": 3758 + }, + { + "epoch": 0.41280474412475293, + "grad_norm": 2.001722812652588, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.685967206954956, + "num_tokens": 96807541.0, + "step": 3759 + }, + { + "epoch": 0.4129145618273666, + "grad_norm": 2.252915859222412, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.704378604888916, + "num_tokens": 96833357.0, + "step": 3760 + }, + { + "epoch": 0.4130243795299802, + "grad_norm": 2.312267780303955, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7078522443771362, + "num_tokens": 96857962.0, + "step": 3761 + }, + { + "epoch": 0.41313419723259387, + "grad_norm": 2.2120561599731445, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6944788098335266, + "num_tokens": 96885093.0, + "step": 3762 + }, + { + "epoch": 0.41324401493520757, + "grad_norm": 2.0831820964813232, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6819888353347778, + "num_tokens": 96915355.0, + "step": 3763 + }, + { + "epoch": 0.4133538326378212, + "grad_norm": 2.157968521118164, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.70053631067276, + "num_tokens": 96943531.0, + "step": 3764 + }, + { + "epoch": 0.41346365034043486, + "grad_norm": 2.183100938796997, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7099862098693848, + "num_tokens": 96968664.0, + "step": 3765 + }, + { + "epoch": 0.41357346804304856, + "grad_norm": 2.289577007293701, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6858023405075073, + "num_tokens": 96992963.0, + "step": 3766 + }, + { + "epoch": 0.4136832857456622, + "grad_norm": 2.126695156097412, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6921921968460083, + "num_tokens": 97020712.0, + "step": 3767 + }, + { + "epoch": 0.41379310344827586, + "grad_norm": 2.3678817749023438, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.707007884979248, + "num_tokens": 97046534.0, + "step": 3768 + }, + { + "epoch": 0.4139029211508895, + "grad_norm": 2.2612693309783936, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6994098424911499, + "num_tokens": 97070798.0, + "step": 3769 + }, + { + "epoch": 0.4140127388535032, + "grad_norm": 2.299227237701416, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6916425228118896, + "num_tokens": 97096226.0, + "step": 3770 + }, + { + "epoch": 0.41412255655611685, + "grad_norm": 2.097446918487549, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6965703964233398, + "num_tokens": 97123290.0, + "step": 3771 + }, + { + "epoch": 0.4142323742587305, + "grad_norm": 2.0656578540802, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.71995609998703, + "num_tokens": 97151929.0, + "step": 3772 + }, + { + "epoch": 0.41434219196134414, + "grad_norm": 2.477630138397217, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6936976909637451, + "num_tokens": 97173448.0, + "step": 3773 + }, + { + "epoch": 0.41445200966395784, + "grad_norm": 2.0925369262695312, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7064610719680786, + "num_tokens": 97202832.0, + "step": 3774 + }, + { + "epoch": 0.4145618273665715, + "grad_norm": 2.3755369186401367, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.725875973701477, + "num_tokens": 97226304.0, + "step": 3775 + }, + { + "epoch": 0.41467164506918514, + "grad_norm": 2.218980550765991, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.693030834197998, + "num_tokens": 97254583.0, + "step": 3776 + }, + { + "epoch": 0.41478146277179884, + "grad_norm": 2.2428982257843018, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7103520035743713, + "num_tokens": 97278426.0, + "step": 3777 + }, + { + "epoch": 0.4148912804744125, + "grad_norm": 2.0410635471343994, + "learning_rate": 1e-06, + "loss": 1.1015, + "mean_token_accuracy": 0.6706748008728027, + "num_tokens": 97309034.0, + "step": 3778 + }, + { + "epoch": 0.41500109817702613, + "grad_norm": 2.1527044773101807, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6992524862289429, + "num_tokens": 97336024.0, + "step": 3779 + }, + { + "epoch": 0.4151109158796398, + "grad_norm": 2.6391544342041016, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6995888948440552, + "num_tokens": 97356206.0, + "step": 3780 + }, + { + "epoch": 0.4152207335822535, + "grad_norm": 2.70109224319458, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7096561193466187, + "num_tokens": 97375112.0, + "step": 3781 + }, + { + "epoch": 0.4153305512848671, + "grad_norm": 2.0895698070526123, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.691648542881012, + "num_tokens": 97407115.0, + "step": 3782 + }, + { + "epoch": 0.41544036898748077, + "grad_norm": 2.2964282035827637, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7135714292526245, + "num_tokens": 97429375.0, + "step": 3783 + }, + { + "epoch": 0.41555018669009447, + "grad_norm": 2.2054243087768555, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6874538064002991, + "num_tokens": 97454060.0, + "step": 3784 + }, + { + "epoch": 0.4156600043927081, + "grad_norm": 2.4529786109924316, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.69737309217453, + "num_tokens": 97475978.0, + "step": 3785 + }, + { + "epoch": 0.41576982209532176, + "grad_norm": 2.123819589614868, + "learning_rate": 1e-06, + "loss": 1.0719, + "mean_token_accuracy": 0.6862756013870239, + "num_tokens": 97501457.0, + "step": 3786 + }, + { + "epoch": 0.4158796397979354, + "grad_norm": 1.9887988567352295, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7242565155029297, + "num_tokens": 97528844.0, + "step": 3787 + }, + { + "epoch": 0.4159894575005491, + "grad_norm": 2.0927767753601074, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.700004518032074, + "num_tokens": 97556821.0, + "step": 3788 + }, + { + "epoch": 0.41609927520316276, + "grad_norm": 2.4036827087402344, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6866105794906616, + "num_tokens": 97580851.0, + "step": 3789 + }, + { + "epoch": 0.4162090929057764, + "grad_norm": 2.394653558731079, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7246111631393433, + "num_tokens": 97601608.0, + "step": 3790 + }, + { + "epoch": 0.41631891060839005, + "grad_norm": 2.2580010890960693, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7152297496795654, + "num_tokens": 97622361.0, + "step": 3791 + }, + { + "epoch": 0.41642872831100375, + "grad_norm": 2.0523176193237305, + "learning_rate": 1e-06, + "loss": 1.0824, + "mean_token_accuracy": 0.6732823848724365, + "num_tokens": 97653024.0, + "step": 3792 + }, + { + "epoch": 0.4165385460136174, + "grad_norm": 1.9357304573059082, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7090226411819458, + "num_tokens": 97683780.0, + "step": 3793 + }, + { + "epoch": 0.41664836371623104, + "grad_norm": 2.2774085998535156, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7117115259170532, + "num_tokens": 97716559.0, + "step": 3794 + }, + { + "epoch": 0.41675818141884474, + "grad_norm": 2.174962282180786, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6873296499252319, + "num_tokens": 97743768.0, + "step": 3795 + }, + { + "epoch": 0.4168679991214584, + "grad_norm": 2.0300283432006836, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7009154558181763, + "num_tokens": 97772285.0, + "step": 3796 + }, + { + "epoch": 0.41697781682407203, + "grad_norm": 2.1846985816955566, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6938285827636719, + "num_tokens": 97798781.0, + "step": 3797 + }, + { + "epoch": 0.4170876345266857, + "grad_norm": 2.438328742980957, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7089094519615173, + "num_tokens": 97820003.0, + "step": 3798 + }, + { + "epoch": 0.4171974522292994, + "grad_norm": 2.681326150894165, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.718023955821991, + "num_tokens": 97837419.0, + "step": 3799 + }, + { + "epoch": 0.417307269931913, + "grad_norm": 2.0700602531433105, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7145993113517761, + "num_tokens": 97866620.0, + "step": 3800 + }, + { + "epoch": 0.4174170876345267, + "grad_norm": 2.401097536087036, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.6971728801727295, + "num_tokens": 97888755.0, + "step": 3801 + }, + { + "epoch": 0.4175269053371403, + "grad_norm": 2.093303680419922, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7060210704803467, + "num_tokens": 97916432.0, + "step": 3802 + }, + { + "epoch": 0.417636723039754, + "grad_norm": 2.3245656490325928, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7133121490478516, + "num_tokens": 97940186.0, + "step": 3803 + }, + { + "epoch": 0.41774654074236767, + "grad_norm": 2.2205355167388916, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7471156120300293, + "num_tokens": 97964726.0, + "step": 3804 + }, + { + "epoch": 0.4178563584449813, + "grad_norm": 2.140367269515991, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.7046671509742737, + "num_tokens": 97991592.0, + "step": 3805 + }, + { + "epoch": 0.417966176147595, + "grad_norm": 2.373044729232788, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7057058811187744, + "num_tokens": 98014208.0, + "step": 3806 + }, + { + "epoch": 0.41807599385020866, + "grad_norm": 1.9831610918045044, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6980545520782471, + "num_tokens": 98044742.0, + "step": 3807 + }, + { + "epoch": 0.4181858115528223, + "grad_norm": 2.2178096771240234, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7310467958450317, + "num_tokens": 98068581.0, + "step": 3808 + }, + { + "epoch": 0.41829562925543595, + "grad_norm": 2.42525315284729, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6940346956253052, + "num_tokens": 98090365.0, + "step": 3809 + }, + { + "epoch": 0.41840544695804965, + "grad_norm": 2.278174877166748, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6890369653701782, + "num_tokens": 98115099.0, + "step": 3810 + }, + { + "epoch": 0.4185152646606633, + "grad_norm": 1.8338361978530884, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7268063426017761, + "num_tokens": 98150220.0, + "step": 3811 + }, + { + "epoch": 0.41862508236327695, + "grad_norm": 2.2488560676574707, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.701461672782898, + "num_tokens": 98174970.0, + "step": 3812 + }, + { + "epoch": 0.41873490006589065, + "grad_norm": 2.160984516143799, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6828996539115906, + "num_tokens": 98202165.0, + "step": 3813 + }, + { + "epoch": 0.4188447177685043, + "grad_norm": 2.420830011367798, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6935960054397583, + "num_tokens": 98224598.0, + "step": 3814 + }, + { + "epoch": 0.41895453547111794, + "grad_norm": 1.902227759361267, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6883276104927063, + "num_tokens": 98257645.0, + "step": 3815 + }, + { + "epoch": 0.4190643531737316, + "grad_norm": 2.2253522872924805, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.696435809135437, + "num_tokens": 98281702.0, + "step": 3816 + }, + { + "epoch": 0.4191741708763453, + "grad_norm": 2.1727466583251953, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7134319543838501, + "num_tokens": 98307269.0, + "step": 3817 + }, + { + "epoch": 0.41928398857895893, + "grad_norm": 2.122563123703003, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7217371463775635, + "num_tokens": 98334480.0, + "step": 3818 + }, + { + "epoch": 0.4193938062815726, + "grad_norm": 2.5037505626678467, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6924813985824585, + "num_tokens": 98356266.0, + "step": 3819 + }, + { + "epoch": 0.4195036239841862, + "grad_norm": 2.3285961151123047, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.7026845812797546, + "num_tokens": 98381659.0, + "step": 3820 + }, + { + "epoch": 0.4196134416867999, + "grad_norm": 2.291578531265259, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7220187783241272, + "num_tokens": 98406060.0, + "step": 3821 + }, + { + "epoch": 0.41972325938941357, + "grad_norm": 2.4336795806884766, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7270791530609131, + "num_tokens": 98426420.0, + "step": 3822 + }, + { + "epoch": 0.4198330770920272, + "grad_norm": 2.3103129863739014, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7060010433197021, + "num_tokens": 98450026.0, + "step": 3823 + }, + { + "epoch": 0.4199428947946409, + "grad_norm": 2.2437446117401123, + "learning_rate": 1e-06, + "loss": 1.0934, + "mean_token_accuracy": 0.6750695109367371, + "num_tokens": 98475785.0, + "step": 3824 + }, + { + "epoch": 0.42005271249725457, + "grad_norm": 2.2564127445220947, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7304388880729675, + "num_tokens": 98500304.0, + "step": 3825 + }, + { + "epoch": 0.4201625301998682, + "grad_norm": 2.3722054958343506, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7301546335220337, + "num_tokens": 98522037.0, + "step": 3826 + }, + { + "epoch": 0.42027234790248186, + "grad_norm": 2.3082704544067383, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6910691857337952, + "num_tokens": 98548026.0, + "step": 3827 + }, + { + "epoch": 0.42038216560509556, + "grad_norm": 2.5428144931793213, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7054101824760437, + "num_tokens": 98568436.0, + "step": 3828 + }, + { + "epoch": 0.4204919833077092, + "grad_norm": 2.174363613128662, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6960198283195496, + "num_tokens": 98595592.0, + "step": 3829 + }, + { + "epoch": 0.42060180101032285, + "grad_norm": 2.283674716949463, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6835830211639404, + "num_tokens": 98619398.0, + "step": 3830 + }, + { + "epoch": 0.42071161871293655, + "grad_norm": 2.4406473636627197, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7145648002624512, + "num_tokens": 98641667.0, + "step": 3831 + }, + { + "epoch": 0.4208214364155502, + "grad_norm": 2.1792171001434326, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6985435485839844, + "num_tokens": 98666819.0, + "step": 3832 + }, + { + "epoch": 0.42093125411816384, + "grad_norm": 2.267683506011963, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7079718112945557, + "num_tokens": 98692488.0, + "step": 3833 + }, + { + "epoch": 0.4210410718207775, + "grad_norm": 2.418734550476074, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.691254734992981, + "num_tokens": 98716352.0, + "step": 3834 + }, + { + "epoch": 0.4211508895233912, + "grad_norm": 2.122225761413574, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7015895843505859, + "num_tokens": 98744745.0, + "step": 3835 + }, + { + "epoch": 0.42126070722600484, + "grad_norm": 2.2010419368743896, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7156627178192139, + "num_tokens": 98771079.0, + "step": 3836 + }, + { + "epoch": 0.4213705249286185, + "grad_norm": 2.1624743938446045, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6930469274520874, + "num_tokens": 98798475.0, + "step": 3837 + }, + { + "epoch": 0.42148034263123213, + "grad_norm": 2.125588893890381, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7162783145904541, + "num_tokens": 98826326.0, + "step": 3838 + }, + { + "epoch": 0.42159016033384583, + "grad_norm": 1.9538238048553467, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7103309035301208, + "num_tokens": 98855670.0, + "step": 3839 + }, + { + "epoch": 0.4216999780364595, + "grad_norm": 2.3467488288879395, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7025852203369141, + "num_tokens": 98879988.0, + "step": 3840 + }, + { + "epoch": 0.4218097957390731, + "grad_norm": 2.2770509719848633, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7083587646484375, + "num_tokens": 98902438.0, + "step": 3841 + }, + { + "epoch": 0.4219196134416868, + "grad_norm": 1.930797815322876, + "learning_rate": 1e-06, + "loss": 1.0745, + "mean_token_accuracy": 0.6834961771965027, + "num_tokens": 98937500.0, + "step": 3842 + }, + { + "epoch": 0.42202943114430047, + "grad_norm": 2.0289483070373535, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7207410335540771, + "num_tokens": 98966668.0, + "step": 3843 + }, + { + "epoch": 0.4221392488469141, + "grad_norm": 2.1869304180145264, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7017238736152649, + "num_tokens": 98991402.0, + "step": 3844 + }, + { + "epoch": 0.42224906654952776, + "grad_norm": 2.0222465991973877, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.6813955903053284, + "num_tokens": 99020907.0, + "step": 3845 + }, + { + "epoch": 0.42235888425214146, + "grad_norm": 2.1852846145629883, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7011620998382568, + "num_tokens": 99048129.0, + "step": 3846 + }, + { + "epoch": 0.4224687019547551, + "grad_norm": 2.13860821723938, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7050786018371582, + "num_tokens": 99072997.0, + "step": 3847 + }, + { + "epoch": 0.42257851965736876, + "grad_norm": 2.139744758605957, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.7096821069717407, + "num_tokens": 99100108.0, + "step": 3848 + }, + { + "epoch": 0.4226883373599824, + "grad_norm": 2.2017762660980225, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6976655721664429, + "num_tokens": 99127149.0, + "step": 3849 + }, + { + "epoch": 0.4227981550625961, + "grad_norm": 2.095644235610962, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6996554732322693, + "num_tokens": 99154865.0, + "step": 3850 + }, + { + "epoch": 0.42290797276520975, + "grad_norm": 2.349611520767212, + "learning_rate": 1e-06, + "loss": 1.0892, + "mean_token_accuracy": 0.6861823797225952, + "num_tokens": 99179363.0, + "step": 3851 + }, + { + "epoch": 0.4230177904678234, + "grad_norm": 2.297175168991089, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7234973907470703, + "num_tokens": 99201574.0, + "step": 3852 + }, + { + "epoch": 0.4231276081704371, + "grad_norm": 2.2420661449432373, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6900520920753479, + "num_tokens": 99226649.0, + "step": 3853 + }, + { + "epoch": 0.42323742587305074, + "grad_norm": 2.2297136783599854, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.706305742263794, + "num_tokens": 99252134.0, + "step": 3854 + }, + { + "epoch": 0.4233472435756644, + "grad_norm": 2.1557934284210205, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7238699197769165, + "num_tokens": 99278219.0, + "step": 3855 + }, + { + "epoch": 0.42345706127827804, + "grad_norm": 2.3756656646728516, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7038823366165161, + "num_tokens": 99301406.0, + "step": 3856 + }, + { + "epoch": 0.42356687898089174, + "grad_norm": 2.253614902496338, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.707371711730957, + "num_tokens": 99325111.0, + "step": 3857 + }, + { + "epoch": 0.4236766966835054, + "grad_norm": 2.2210445404052734, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.692817747592926, + "num_tokens": 99351175.0, + "step": 3858 + }, + { + "epoch": 0.42378651438611903, + "grad_norm": 2.253146171569824, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.719214916229248, + "num_tokens": 99375933.0, + "step": 3859 + }, + { + "epoch": 0.42389633208873273, + "grad_norm": 1.9564502239227295, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6948157548904419, + "num_tokens": 99406990.0, + "step": 3860 + }, + { + "epoch": 0.4240061497913464, + "grad_norm": 2.1133267879486084, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7037972211837769, + "num_tokens": 99432960.0, + "step": 3861 + }, + { + "epoch": 0.42411596749396, + "grad_norm": 2.0603325366973877, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7066550850868225, + "num_tokens": 99462583.0, + "step": 3862 + }, + { + "epoch": 0.42422578519657367, + "grad_norm": 2.686002731323242, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7134743928909302, + "num_tokens": 99481555.0, + "step": 3863 + }, + { + "epoch": 0.42433560289918737, + "grad_norm": 2.006033182144165, + "learning_rate": 1e-06, + "loss": 1.0543, + "mean_token_accuracy": 0.6791362762451172, + "num_tokens": 99512339.0, + "step": 3864 + }, + { + "epoch": 0.424445420601801, + "grad_norm": 2.3336048126220703, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7043861150741577, + "num_tokens": 99535316.0, + "step": 3865 + }, + { + "epoch": 0.42455523830441466, + "grad_norm": 2.0684471130371094, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6926730871200562, + "num_tokens": 99565234.0, + "step": 3866 + }, + { + "epoch": 0.4246650560070283, + "grad_norm": 2.054276943206787, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6924556493759155, + "num_tokens": 99594710.0, + "step": 3867 + }, + { + "epoch": 0.424774873709642, + "grad_norm": 2.0494232177734375, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7260864973068237, + "num_tokens": 99622556.0, + "step": 3868 + }, + { + "epoch": 0.42488469141225566, + "grad_norm": 2.392071485519409, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7264935970306396, + "num_tokens": 99643411.0, + "step": 3869 + }, + { + "epoch": 0.4249945091148693, + "grad_norm": 2.091646671295166, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6956048607826233, + "num_tokens": 99671025.0, + "step": 3870 + }, + { + "epoch": 0.425104326817483, + "grad_norm": 2.443901538848877, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7143837213516235, + "num_tokens": 99694558.0, + "step": 3871 + }, + { + "epoch": 0.42521414452009665, + "grad_norm": 1.974377989768982, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6934738755226135, + "num_tokens": 99724612.0, + "step": 3872 + }, + { + "epoch": 0.4253239622227103, + "grad_norm": 2.179954767227173, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7045800089836121, + "num_tokens": 99750517.0, + "step": 3873 + }, + { + "epoch": 0.42543377992532394, + "grad_norm": 2.4881839752197266, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7053583264350891, + "num_tokens": 99771745.0, + "step": 3874 + }, + { + "epoch": 0.42554359762793764, + "grad_norm": 2.099529504776001, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6926359534263611, + "num_tokens": 99801595.0, + "step": 3875 + }, + { + "epoch": 0.4256534153305513, + "grad_norm": 1.918192744255066, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6943329572677612, + "num_tokens": 99836614.0, + "step": 3876 + }, + { + "epoch": 0.42576323303316493, + "grad_norm": 2.192491292953491, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6995394825935364, + "num_tokens": 99861737.0, + "step": 3877 + }, + { + "epoch": 0.4258730507357786, + "grad_norm": 2.480292797088623, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7157601118087769, + "num_tokens": 99881245.0, + "step": 3878 + }, + { + "epoch": 0.4259828684383923, + "grad_norm": 2.314626693725586, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6865398287773132, + "num_tokens": 99905892.0, + "step": 3879 + }, + { + "epoch": 0.4260926861410059, + "grad_norm": 2.273200273513794, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6822601556777954, + "num_tokens": 99931787.0, + "step": 3880 + }, + { + "epoch": 0.4262025038436196, + "grad_norm": 2.0520613193511963, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7131116390228271, + "num_tokens": 99959031.0, + "step": 3881 + }, + { + "epoch": 0.4263123215462333, + "grad_norm": 2.1017305850982666, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6955141425132751, + "num_tokens": 99985411.0, + "step": 3882 + }, + { + "epoch": 0.4264221392488469, + "grad_norm": 2.1682772636413574, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6910509467124939, + "num_tokens": 100013400.0, + "step": 3883 + }, + { + "epoch": 0.42653195695146057, + "grad_norm": 2.431345224380493, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7230134010314941, + "num_tokens": 100035128.0, + "step": 3884 + }, + { + "epoch": 0.4266417746540742, + "grad_norm": 2.2757415771484375, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.688630998134613, + "num_tokens": 100062279.0, + "step": 3885 + }, + { + "epoch": 0.4267515923566879, + "grad_norm": 2.0396013259887695, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7162804007530212, + "num_tokens": 100088755.0, + "step": 3886 + }, + { + "epoch": 0.42686141005930156, + "grad_norm": 2.1354422569274902, + "learning_rate": 1e-06, + "loss": 1.1298, + "mean_token_accuracy": 0.6651373505592346, + "num_tokens": 100116619.0, + "step": 3887 + }, + { + "epoch": 0.4269712277619152, + "grad_norm": 2.237685203552246, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.686126708984375, + "num_tokens": 100142887.0, + "step": 3888 + }, + { + "epoch": 0.4270810454645289, + "grad_norm": 2.3187899589538574, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6965067982673645, + "num_tokens": 100166580.0, + "step": 3889 + }, + { + "epoch": 0.42719086316714255, + "grad_norm": 2.257658004760742, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6907580494880676, + "num_tokens": 100192451.0, + "step": 3890 + }, + { + "epoch": 0.4273006808697562, + "grad_norm": 2.3985986709594727, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.707036554813385, + "num_tokens": 100214318.0, + "step": 3891 + }, + { + "epoch": 0.42741049857236985, + "grad_norm": 2.4194369316101074, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7307144999504089, + "num_tokens": 100235249.0, + "step": 3892 + }, + { + "epoch": 0.42752031627498355, + "grad_norm": 2.4158341884613037, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7128109931945801, + "num_tokens": 100258502.0, + "step": 3893 + }, + { + "epoch": 0.4276301339775972, + "grad_norm": 1.9596147537231445, + "learning_rate": 1e-06, + "loss": 1.0967, + "mean_token_accuracy": 0.6765286922454834, + "num_tokens": 100292318.0, + "step": 3894 + }, + { + "epoch": 0.42773995168021084, + "grad_norm": 1.9904133081436157, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.697498083114624, + "num_tokens": 100324526.0, + "step": 3895 + }, + { + "epoch": 0.4278497693828245, + "grad_norm": 2.482205867767334, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7123410701751709, + "num_tokens": 100344690.0, + "step": 3896 + }, + { + "epoch": 0.4279595870854382, + "grad_norm": 2.5242979526519775, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.701655387878418, + "num_tokens": 100365374.0, + "step": 3897 + }, + { + "epoch": 0.42806940478805183, + "grad_norm": 1.9059054851531982, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6839662790298462, + "num_tokens": 100400228.0, + "step": 3898 + }, + { + "epoch": 0.4281792224906655, + "grad_norm": 2.353431224822998, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7065154314041138, + "num_tokens": 100423225.0, + "step": 3899 + }, + { + "epoch": 0.4282890401932792, + "grad_norm": 2.05259370803833, + "learning_rate": 1e-06, + "loss": 1.1151, + "mean_token_accuracy": 0.671126127243042, + "num_tokens": 100455897.0, + "step": 3900 + }, + { + "epoch": 0.4283988578958928, + "grad_norm": 2.5901801586151123, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7137159109115601, + "num_tokens": 100475443.0, + "step": 3901 + }, + { + "epoch": 0.42850867559850647, + "grad_norm": 1.863952875137329, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.713721752166748, + "num_tokens": 100509167.0, + "step": 3902 + }, + { + "epoch": 0.4286184933011201, + "grad_norm": 2.277771472930908, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6979600191116333, + "num_tokens": 100533354.0, + "step": 3903 + }, + { + "epoch": 0.4287283110037338, + "grad_norm": 2.1553056240081787, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.72135990858078, + "num_tokens": 100559758.0, + "step": 3904 + }, + { + "epoch": 0.42883812870634747, + "grad_norm": 2.18446946144104, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6825944185256958, + "num_tokens": 100587012.0, + "step": 3905 + }, + { + "epoch": 0.4289479464089611, + "grad_norm": 1.99862539768219, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6878224015235901, + "num_tokens": 100617548.0, + "step": 3906 + }, + { + "epoch": 0.4290577641115748, + "grad_norm": 2.0112192630767822, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6986352801322937, + "num_tokens": 100649395.0, + "step": 3907 + }, + { + "epoch": 0.42916758181418846, + "grad_norm": 2.4163589477539062, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7082729935646057, + "num_tokens": 100671892.0, + "step": 3908 + }, + { + "epoch": 0.4292773995168021, + "grad_norm": 2.1444332599639893, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6820229291915894, + "num_tokens": 100699482.0, + "step": 3909 + }, + { + "epoch": 0.42938721721941575, + "grad_norm": 2.2460432052612305, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7306302189826965, + "num_tokens": 100722497.0, + "step": 3910 + }, + { + "epoch": 0.42949703492202945, + "grad_norm": 2.438513994216919, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7225770354270935, + "num_tokens": 100744559.0, + "step": 3911 + }, + { + "epoch": 0.4296068526246431, + "grad_norm": 2.36588978767395, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7125284671783447, + "num_tokens": 100767654.0, + "step": 3912 + }, + { + "epoch": 0.42971667032725674, + "grad_norm": 2.038952589035034, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6930612325668335, + "num_tokens": 100794629.0, + "step": 3913 + }, + { + "epoch": 0.4298264880298704, + "grad_norm": 2.5288350582122803, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.70943683385849, + "num_tokens": 100815522.0, + "step": 3914 + }, + { + "epoch": 0.4299363057324841, + "grad_norm": 1.9420678615570068, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6932726502418518, + "num_tokens": 100848512.0, + "step": 3915 + }, + { + "epoch": 0.43004612343509774, + "grad_norm": 2.266817808151245, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7205708026885986, + "num_tokens": 100872490.0, + "step": 3916 + }, + { + "epoch": 0.4301559411377114, + "grad_norm": 1.9805023670196533, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7125425338745117, + "num_tokens": 100902388.0, + "step": 3917 + }, + { + "epoch": 0.4302657588403251, + "grad_norm": 1.9962787628173828, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.697142481803894, + "num_tokens": 100933950.0, + "step": 3918 + }, + { + "epoch": 0.43037557654293873, + "grad_norm": 2.6046950817108154, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7015892267227173, + "num_tokens": 100953476.0, + "step": 3919 + }, + { + "epoch": 0.4304853942455524, + "grad_norm": 2.2852587699890137, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6989878416061401, + "num_tokens": 100978579.0, + "step": 3920 + }, + { + "epoch": 0.430595211948166, + "grad_norm": 2.4373254776000977, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6781731843948364, + "num_tokens": 101003328.0, + "step": 3921 + }, + { + "epoch": 0.4307050296507797, + "grad_norm": 2.1859219074249268, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7008310556411743, + "num_tokens": 101030796.0, + "step": 3922 + }, + { + "epoch": 0.43081484735339337, + "grad_norm": 2.0199520587921143, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6945873498916626, + "num_tokens": 101060541.0, + "step": 3923 + }, + { + "epoch": 0.430924665056007, + "grad_norm": 2.09557843208313, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6979280710220337, + "num_tokens": 101087852.0, + "step": 3924 + }, + { + "epoch": 0.43103448275862066, + "grad_norm": 2.205646514892578, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7145206928253174, + "num_tokens": 101112107.0, + "step": 3925 + }, + { + "epoch": 0.43114430046123436, + "grad_norm": 2.036341667175293, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7068872451782227, + "num_tokens": 101143195.0, + "step": 3926 + }, + { + "epoch": 0.431254118163848, + "grad_norm": 2.318063735961914, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6971789598464966, + "num_tokens": 101168792.0, + "step": 3927 + }, + { + "epoch": 0.43136393586646166, + "grad_norm": 2.7328357696533203, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6918398141860962, + "num_tokens": 101191474.0, + "step": 3928 + }, + { + "epoch": 0.43147375356907536, + "grad_norm": 2.3088855743408203, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6868832111358643, + "num_tokens": 101215282.0, + "step": 3929 + }, + { + "epoch": 0.431583571271689, + "grad_norm": 2.0312929153442383, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6988438367843628, + "num_tokens": 101246024.0, + "step": 3930 + }, + { + "epoch": 0.43169338897430265, + "grad_norm": 2.620255470275879, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7168868780136108, + "num_tokens": 101264789.0, + "step": 3931 + }, + { + "epoch": 0.4318032066769163, + "grad_norm": 2.0298290252685547, + "learning_rate": 1e-06, + "loss": 1.0753, + "mean_token_accuracy": 0.6780264377593994, + "num_tokens": 101296307.0, + "step": 3932 + }, + { + "epoch": 0.43191302437953, + "grad_norm": 2.5164694786071777, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7041593790054321, + "num_tokens": 101315971.0, + "step": 3933 + }, + { + "epoch": 0.43202284208214364, + "grad_norm": 2.3104140758514404, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.7005573511123657, + "num_tokens": 101341719.0, + "step": 3934 + }, + { + "epoch": 0.4321326597847573, + "grad_norm": 2.0771074295043945, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7237249612808228, + "num_tokens": 101368386.0, + "step": 3935 + }, + { + "epoch": 0.432242477487371, + "grad_norm": 2.4314260482788086, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7039139866828918, + "num_tokens": 101388174.0, + "step": 3936 + }, + { + "epoch": 0.43235229518998464, + "grad_norm": 2.4515604972839355, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6966507434844971, + "num_tokens": 101409826.0, + "step": 3937 + }, + { + "epoch": 0.4324621128925983, + "grad_norm": 2.267864465713501, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7108546495437622, + "num_tokens": 101433839.0, + "step": 3938 + }, + { + "epoch": 0.43257193059521193, + "grad_norm": 2.4238758087158203, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6949402689933777, + "num_tokens": 101455407.0, + "step": 3939 + }, + { + "epoch": 0.43268174829782563, + "grad_norm": 2.0580949783325195, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7194444537162781, + "num_tokens": 101482335.0, + "step": 3940 + }, + { + "epoch": 0.4327915660004393, + "grad_norm": 2.4437384605407715, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6955724954605103, + "num_tokens": 101504170.0, + "step": 3941 + }, + { + "epoch": 0.4329013837030529, + "grad_norm": 2.495542049407959, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7070655226707458, + "num_tokens": 101523257.0, + "step": 3942 + }, + { + "epoch": 0.43301120140566657, + "grad_norm": 2.331328868865967, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6961788535118103, + "num_tokens": 101546738.0, + "step": 3943 + }, + { + "epoch": 0.43312101910828027, + "grad_norm": 2.3113460540771484, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7086038589477539, + "num_tokens": 101569753.0, + "step": 3944 + }, + { + "epoch": 0.4332308368108939, + "grad_norm": 2.418670177459717, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6967171430587769, + "num_tokens": 101593504.0, + "step": 3945 + }, + { + "epoch": 0.43334065451350756, + "grad_norm": 2.547701358795166, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7171459794044495, + "num_tokens": 101612389.0, + "step": 3946 + }, + { + "epoch": 0.43345047221612126, + "grad_norm": 2.16876482963562, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.6996749639511108, + "num_tokens": 101638044.0, + "step": 3947 + }, + { + "epoch": 0.4335602899187349, + "grad_norm": 2.2981417179107666, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7138744592666626, + "num_tokens": 101663891.0, + "step": 3948 + }, + { + "epoch": 0.43367010762134856, + "grad_norm": 2.4296627044677734, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7100763320922852, + "num_tokens": 101686221.0, + "step": 3949 + }, + { + "epoch": 0.4337799253239622, + "grad_norm": 2.3302621841430664, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6896848082542419, + "num_tokens": 101709843.0, + "step": 3950 + }, + { + "epoch": 0.4338897430265759, + "grad_norm": 2.3662948608398438, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7197931408882141, + "num_tokens": 101731317.0, + "step": 3951 + }, + { + "epoch": 0.43399956072918955, + "grad_norm": 2.0874993801116943, + "learning_rate": 1e-06, + "loss": 1.1406, + "mean_token_accuracy": 0.6586518287658691, + "num_tokens": 101762643.0, + "step": 3952 + }, + { + "epoch": 0.4341093784318032, + "grad_norm": 2.7236135005950928, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7179673314094543, + "num_tokens": 101780589.0, + "step": 3953 + }, + { + "epoch": 0.43421919613441684, + "grad_norm": 2.1889424324035645, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6947421431541443, + "num_tokens": 101806859.0, + "step": 3954 + }, + { + "epoch": 0.43432901383703054, + "grad_norm": 2.1367838382720947, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7228938341140747, + "num_tokens": 101831232.0, + "step": 3955 + }, + { + "epoch": 0.4344388315396442, + "grad_norm": 2.5239055156707764, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7144164443016052, + "num_tokens": 101850296.0, + "step": 3956 + }, + { + "epoch": 0.43454864924225783, + "grad_norm": 2.2161543369293213, + "learning_rate": 1e-06, + "loss": 1.0503, + "mean_token_accuracy": 0.6837509870529175, + "num_tokens": 101878522.0, + "step": 3957 + }, + { + "epoch": 0.43465846694487154, + "grad_norm": 2.37097430229187, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6848585605621338, + "num_tokens": 101901621.0, + "step": 3958 + }, + { + "epoch": 0.4347682846474852, + "grad_norm": 2.0441055297851562, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7089922428131104, + "num_tokens": 101930044.0, + "step": 3959 + }, + { + "epoch": 0.4348781023500988, + "grad_norm": 2.1836040019989014, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7021145820617676, + "num_tokens": 101957344.0, + "step": 3960 + }, + { + "epoch": 0.4349879200527125, + "grad_norm": 2.2552554607391357, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6851388216018677, + "num_tokens": 101981689.0, + "step": 3961 + }, + { + "epoch": 0.4350977377553262, + "grad_norm": 2.5744776725769043, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7034660577774048, + "num_tokens": 102001589.0, + "step": 3962 + }, + { + "epoch": 0.4352075554579398, + "grad_norm": 2.5024502277374268, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7300044298171997, + "num_tokens": 102021429.0, + "step": 3963 + }, + { + "epoch": 0.43531737316055347, + "grad_norm": 2.2102649211883545, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.69510817527771, + "num_tokens": 102046826.0, + "step": 3964 + }, + { + "epoch": 0.43542719086316717, + "grad_norm": 2.198240041732788, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7021706700325012, + "num_tokens": 102072643.0, + "step": 3965 + }, + { + "epoch": 0.4355370085657808, + "grad_norm": 2.1627840995788574, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6895313262939453, + "num_tokens": 102100552.0, + "step": 3966 + }, + { + "epoch": 0.43564682626839446, + "grad_norm": 2.0828466415405273, + "learning_rate": 1e-06, + "loss": 1.0759, + "mean_token_accuracy": 0.6779358386993408, + "num_tokens": 102130128.0, + "step": 3967 + }, + { + "epoch": 0.4357566439710081, + "grad_norm": 2.1375515460968018, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7158424854278564, + "num_tokens": 102156920.0, + "step": 3968 + }, + { + "epoch": 0.4358664616736218, + "grad_norm": 2.1737420558929443, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6912481784820557, + "num_tokens": 102184748.0, + "step": 3969 + }, + { + "epoch": 0.43597627937623545, + "grad_norm": 2.2898309230804443, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6889442205429077, + "num_tokens": 102210396.0, + "step": 3970 + }, + { + "epoch": 0.4360860970788491, + "grad_norm": 2.6321794986724854, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.721218466758728, + "num_tokens": 102229119.0, + "step": 3971 + }, + { + "epoch": 0.43619591478146275, + "grad_norm": 2.014099597930908, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7049857378005981, + "num_tokens": 102257368.0, + "step": 3972 + }, + { + "epoch": 0.43630573248407645, + "grad_norm": 1.9609724283218384, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6998693943023682, + "num_tokens": 102288711.0, + "step": 3973 + }, + { + "epoch": 0.4364155501866901, + "grad_norm": 2.092095375061035, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7053248882293701, + "num_tokens": 102315390.0, + "step": 3974 + }, + { + "epoch": 0.43652536788930374, + "grad_norm": 2.186718702316284, + "learning_rate": 1e-06, + "loss": 1.1329, + "mean_token_accuracy": 0.6703634858131409, + "num_tokens": 102340576.0, + "step": 3975 + }, + { + "epoch": 0.43663518559191744, + "grad_norm": 2.405301332473755, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.71343994140625, + "num_tokens": 102362614.0, + "step": 3976 + }, + { + "epoch": 0.4367450032945311, + "grad_norm": 2.4668490886688232, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7226957678794861, + "num_tokens": 102384611.0, + "step": 3977 + }, + { + "epoch": 0.43685482099714473, + "grad_norm": 2.0879721641540527, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6880863308906555, + "num_tokens": 102414291.0, + "step": 3978 + }, + { + "epoch": 0.4369646386997584, + "grad_norm": 2.4076170921325684, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7054308652877808, + "num_tokens": 102436491.0, + "step": 3979 + }, + { + "epoch": 0.4370744564023721, + "grad_norm": 2.2991080284118652, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7285411357879639, + "num_tokens": 102457903.0, + "step": 3980 + }, + { + "epoch": 0.4371842741049857, + "grad_norm": 2.1772079467773438, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7076929807662964, + "num_tokens": 102482883.0, + "step": 3981 + }, + { + "epoch": 0.43729409180759937, + "grad_norm": 2.145594835281372, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7116020321846008, + "num_tokens": 102510641.0, + "step": 3982 + }, + { + "epoch": 0.4374039095102131, + "grad_norm": 2.2807326316833496, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7018805742263794, + "num_tokens": 102534387.0, + "step": 3983 + }, + { + "epoch": 0.4375137272128267, + "grad_norm": 2.132246255874634, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7056987285614014, + "num_tokens": 102559979.0, + "step": 3984 + }, + { + "epoch": 0.43762354491544037, + "grad_norm": 2.422032356262207, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7055045366287231, + "num_tokens": 102582247.0, + "step": 3985 + }, + { + "epoch": 0.437733362618054, + "grad_norm": 2.151421308517456, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6886246800422668, + "num_tokens": 102608863.0, + "step": 3986 + }, + { + "epoch": 0.4378431803206677, + "grad_norm": 2.320772886276245, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.686324417591095, + "num_tokens": 102632826.0, + "step": 3987 + }, + { + "epoch": 0.43795299802328136, + "grad_norm": 2.4483046531677246, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6992133259773254, + "num_tokens": 102655014.0, + "step": 3988 + }, + { + "epoch": 0.438062815725895, + "grad_norm": 2.200467348098755, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7284928560256958, + "num_tokens": 102680466.0, + "step": 3989 + }, + { + "epoch": 0.43817263342850865, + "grad_norm": 2.1842668056488037, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6894611716270447, + "num_tokens": 102706520.0, + "step": 3990 + }, + { + "epoch": 0.43828245113112235, + "grad_norm": 2.1822750568389893, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7181072235107422, + "num_tokens": 102732000.0, + "step": 3991 + }, + { + "epoch": 0.438392268833736, + "grad_norm": 2.1588456630706787, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7073677778244019, + "num_tokens": 102758909.0, + "step": 3992 + }, + { + "epoch": 0.43850208653634964, + "grad_norm": 2.2948946952819824, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6930074691772461, + "num_tokens": 102784524.0, + "step": 3993 + }, + { + "epoch": 0.43861190423896335, + "grad_norm": 1.8013380765914917, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7060186862945557, + "num_tokens": 102819991.0, + "step": 3994 + }, + { + "epoch": 0.438721721941577, + "grad_norm": 2.5586700439453125, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.7051536440849304, + "num_tokens": 102840806.0, + "step": 3995 + }, + { + "epoch": 0.43883153964419064, + "grad_norm": 2.3021368980407715, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7161450386047363, + "num_tokens": 102862137.0, + "step": 3996 + }, + { + "epoch": 0.4389413573468043, + "grad_norm": 2.491365671157837, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7088992595672607, + "num_tokens": 102881979.0, + "step": 3997 + }, + { + "epoch": 0.439051175049418, + "grad_norm": 2.3776514530181885, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6814150810241699, + "num_tokens": 102905969.0, + "step": 3998 + }, + { + "epoch": 0.43916099275203163, + "grad_norm": 2.447347640991211, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7060372829437256, + "num_tokens": 102927292.0, + "step": 3999 + }, + { + "epoch": 0.4392708104546453, + "grad_norm": 2.1861021518707275, + "learning_rate": 1e-06, + "loss": 1.1334, + "mean_token_accuracy": 0.6719980835914612, + "num_tokens": 102956074.0, + "step": 4000 + }, + { + "epoch": 0.4393806281572589, + "grad_norm": 2.2434165477752686, + "learning_rate": 1e-06, + "loss": 1.0856, + "mean_token_accuracy": 0.6741212010383606, + "num_tokens": 102982902.0, + "step": 4001 + }, + { + "epoch": 0.4394904458598726, + "grad_norm": 2.16475772857666, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6824787855148315, + "num_tokens": 103008900.0, + "step": 4002 + }, + { + "epoch": 0.43960026356248627, + "grad_norm": 2.1531870365142822, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6865631341934204, + "num_tokens": 103037853.0, + "step": 4003 + }, + { + "epoch": 0.4397100812650999, + "grad_norm": 2.4570162296295166, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6997050046920776, + "num_tokens": 103058818.0, + "step": 4004 + }, + { + "epoch": 0.4398198989677136, + "grad_norm": 2.22141170501709, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6893113851547241, + "num_tokens": 103083998.0, + "step": 4005 + }, + { + "epoch": 0.43992971667032726, + "grad_norm": 2.267251968383789, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7029142379760742, + "num_tokens": 103108996.0, + "step": 4006 + }, + { + "epoch": 0.4400395343729409, + "grad_norm": 2.0112733840942383, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7106727957725525, + "num_tokens": 103138050.0, + "step": 4007 + }, + { + "epoch": 0.44014935207555456, + "grad_norm": 2.431870460510254, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6803272366523743, + "num_tokens": 103161763.0, + "step": 4008 + }, + { + "epoch": 0.44025916977816826, + "grad_norm": 2.200242519378662, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7027105093002319, + "num_tokens": 103187973.0, + "step": 4009 + }, + { + "epoch": 0.4403689874807819, + "grad_norm": 2.003202199935913, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.695116400718689, + "num_tokens": 103219046.0, + "step": 4010 + }, + { + "epoch": 0.44047880518339555, + "grad_norm": 2.1706058979034424, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6959890127182007, + "num_tokens": 103247541.0, + "step": 4011 + }, + { + "epoch": 0.44058862288600925, + "grad_norm": 2.1487014293670654, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7000219225883484, + "num_tokens": 103272534.0, + "step": 4012 + }, + { + "epoch": 0.4406984405886229, + "grad_norm": 2.499898910522461, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.7021883726119995, + "num_tokens": 103292070.0, + "step": 4013 + }, + { + "epoch": 0.44080825829123654, + "grad_norm": 2.1835851669311523, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6963542103767395, + "num_tokens": 103316841.0, + "step": 4014 + }, + { + "epoch": 0.4409180759938502, + "grad_norm": 2.025481700897217, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6951490640640259, + "num_tokens": 103346932.0, + "step": 4015 + }, + { + "epoch": 0.4410278936964639, + "grad_norm": 2.3005995750427246, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7218643426895142, + "num_tokens": 103370305.0, + "step": 4016 + }, + { + "epoch": 0.44113771139907754, + "grad_norm": 2.359930992126465, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7472615242004395, + "num_tokens": 103391775.0, + "step": 4017 + }, + { + "epoch": 0.4412475291016912, + "grad_norm": 1.921132206916809, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6952099204063416, + "num_tokens": 103423557.0, + "step": 4018 + }, + { + "epoch": 0.44135734680430483, + "grad_norm": 2.2487921714782715, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6963936686515808, + "num_tokens": 103449519.0, + "step": 4019 + }, + { + "epoch": 0.44146716450691853, + "grad_norm": 2.1701087951660156, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7466570138931274, + "num_tokens": 103475536.0, + "step": 4020 + }, + { + "epoch": 0.4415769822095322, + "grad_norm": 2.663985252380371, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7211411595344543, + "num_tokens": 103492279.0, + "step": 4021 + }, + { + "epoch": 0.4416867999121458, + "grad_norm": 2.15343976020813, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.7001014947891235, + "num_tokens": 103517831.0, + "step": 4022 + }, + { + "epoch": 0.4417966176147595, + "grad_norm": 2.096076011657715, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6800525784492493, + "num_tokens": 103546947.0, + "step": 4023 + }, + { + "epoch": 0.44190643531737317, + "grad_norm": 2.5248146057128906, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.6979792714118958, + "num_tokens": 103567976.0, + "step": 4024 + }, + { + "epoch": 0.4420162530199868, + "grad_norm": 2.247014284133911, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6951417922973633, + "num_tokens": 103594457.0, + "step": 4025 + }, + { + "epoch": 0.44212607072260046, + "grad_norm": 2.3420157432556152, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6815412044525146, + "num_tokens": 103619515.0, + "step": 4026 + }, + { + "epoch": 0.44223588842521416, + "grad_norm": 2.5954623222351074, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.707636833190918, + "num_tokens": 103639262.0, + "step": 4027 + }, + { + "epoch": 0.4423457061278278, + "grad_norm": 2.331247568130493, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7133511304855347, + "num_tokens": 103661368.0, + "step": 4028 + }, + { + "epoch": 0.44245552383044146, + "grad_norm": 2.328111171722412, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7059885263442993, + "num_tokens": 103684242.0, + "step": 4029 + }, + { + "epoch": 0.4425653415330551, + "grad_norm": 1.965271234512329, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7164977788925171, + "num_tokens": 103715861.0, + "step": 4030 + }, + { + "epoch": 0.4426751592356688, + "grad_norm": 2.094224214553833, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6907593607902527, + "num_tokens": 103745869.0, + "step": 4031 + }, + { + "epoch": 0.44278497693828245, + "grad_norm": 2.0192201137542725, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.697727620601654, + "num_tokens": 103775204.0, + "step": 4032 + }, + { + "epoch": 0.4428947946408961, + "grad_norm": 2.700960159301758, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6928646564483643, + "num_tokens": 103793661.0, + "step": 4033 + }, + { + "epoch": 0.4430046123435098, + "grad_norm": 2.3095765113830566, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7050563097000122, + "num_tokens": 103817478.0, + "step": 4034 + }, + { + "epoch": 0.44311443004612344, + "grad_norm": 2.2201995849609375, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7000494003295898, + "num_tokens": 103843386.0, + "step": 4035 + }, + { + "epoch": 0.4432242477487371, + "grad_norm": 2.2436208724975586, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7097482085227966, + "num_tokens": 103867997.0, + "step": 4036 + }, + { + "epoch": 0.44333406545135073, + "grad_norm": 2.142369270324707, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6955711245536804, + "num_tokens": 103895351.0, + "step": 4037 + }, + { + "epoch": 0.44344388315396444, + "grad_norm": 2.3585352897644043, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.694672703742981, + "num_tokens": 103919943.0, + "step": 4038 + }, + { + "epoch": 0.4435537008565781, + "grad_norm": 2.362112283706665, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.6896475553512573, + "num_tokens": 103942577.0, + "step": 4039 + }, + { + "epoch": 0.4436635185591917, + "grad_norm": 1.7643578052520752, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.703281044960022, + "num_tokens": 103978198.0, + "step": 4040 + }, + { + "epoch": 0.44377333626180543, + "grad_norm": 2.2839746475219727, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7127700448036194, + "num_tokens": 104003481.0, + "step": 4041 + }, + { + "epoch": 0.4438831539644191, + "grad_norm": 2.3426661491394043, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7002139091491699, + "num_tokens": 104027113.0, + "step": 4042 + }, + { + "epoch": 0.4439929716670327, + "grad_norm": 2.1787993907928467, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6915097236633301, + "num_tokens": 104053360.0, + "step": 4043 + }, + { + "epoch": 0.44410278936964637, + "grad_norm": 2.332312822341919, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.728919267654419, + "num_tokens": 104073894.0, + "step": 4044 + }, + { + "epoch": 0.44421260707226007, + "grad_norm": 2.3409574031829834, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7104072570800781, + "num_tokens": 104097409.0, + "step": 4045 + }, + { + "epoch": 0.4443224247748737, + "grad_norm": 2.3818881511688232, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6968114972114563, + "num_tokens": 104120527.0, + "step": 4046 + }, + { + "epoch": 0.44443224247748736, + "grad_norm": 2.223130226135254, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.692104160785675, + "num_tokens": 104147753.0, + "step": 4047 + }, + { + "epoch": 0.444542060180101, + "grad_norm": 2.316585063934326, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7048385143280029, + "num_tokens": 104169577.0, + "step": 4048 + }, + { + "epoch": 0.4446518778827147, + "grad_norm": 2.0524988174438477, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.7052168250083923, + "num_tokens": 104199328.0, + "step": 4049 + }, + { + "epoch": 0.44476169558532835, + "grad_norm": 2.143908739089966, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6956769227981567, + "num_tokens": 104226165.0, + "step": 4050 + }, + { + "epoch": 0.444871513287942, + "grad_norm": 2.2981200218200684, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6908201575279236, + "num_tokens": 104250442.0, + "step": 4051 + }, + { + "epoch": 0.4449813309905557, + "grad_norm": 2.249117851257324, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7040705680847168, + "num_tokens": 104274768.0, + "step": 4052 + }, + { + "epoch": 0.44509114869316935, + "grad_norm": 2.095470428466797, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6918485164642334, + "num_tokens": 104303028.0, + "step": 4053 + }, + { + "epoch": 0.445200966395783, + "grad_norm": 2.166135311126709, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7166626453399658, + "num_tokens": 104332229.0, + "step": 4054 + }, + { + "epoch": 0.44531078409839664, + "grad_norm": 2.2521414756774902, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7096738219261169, + "num_tokens": 104357879.0, + "step": 4055 + }, + { + "epoch": 0.44542060180101034, + "grad_norm": 2.1062076091766357, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7153946757316589, + "num_tokens": 104385432.0, + "step": 4056 + }, + { + "epoch": 0.445530419503624, + "grad_norm": 2.255462408065796, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.687814474105835, + "num_tokens": 104409840.0, + "step": 4057 + }, + { + "epoch": 0.44564023720623763, + "grad_norm": 2.327808141708374, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.706976056098938, + "num_tokens": 104433548.0, + "step": 4058 + }, + { + "epoch": 0.44575005490885133, + "grad_norm": 2.537221670150757, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6846703290939331, + "num_tokens": 104456090.0, + "step": 4059 + }, + { + "epoch": 0.445859872611465, + "grad_norm": 2.166015386581421, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6955404877662659, + "num_tokens": 104484248.0, + "step": 4060 + }, + { + "epoch": 0.4459696903140786, + "grad_norm": 2.271543264389038, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7027437686920166, + "num_tokens": 104508656.0, + "step": 4061 + }, + { + "epoch": 0.44607950801669227, + "grad_norm": 2.196089506149292, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7089373469352722, + "num_tokens": 104534642.0, + "step": 4062 + }, + { + "epoch": 0.446189325719306, + "grad_norm": 2.3131203651428223, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7074493765830994, + "num_tokens": 104558342.0, + "step": 4063 + }, + { + "epoch": 0.4462991434219196, + "grad_norm": 2.3804173469543457, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6939637064933777, + "num_tokens": 104582597.0, + "step": 4064 + }, + { + "epoch": 0.44640896112453327, + "grad_norm": 2.026526927947998, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7053696513175964, + "num_tokens": 104611970.0, + "step": 4065 + }, + { + "epoch": 0.4465187788271469, + "grad_norm": 2.074167251586914, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6816279888153076, + "num_tokens": 104640165.0, + "step": 4066 + }, + { + "epoch": 0.4466285965297606, + "grad_norm": 2.040369987487793, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.684686541557312, + "num_tokens": 104668888.0, + "step": 4067 + }, + { + "epoch": 0.44673841423237426, + "grad_norm": 2.3673694133758545, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7055587768554688, + "num_tokens": 104691464.0, + "step": 4068 + }, + { + "epoch": 0.4468482319349879, + "grad_norm": 2.6723945140838623, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7375389337539673, + "num_tokens": 104710953.0, + "step": 4069 + }, + { + "epoch": 0.4469580496376016, + "grad_norm": 2.18723201751709, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7013676166534424, + "num_tokens": 104739848.0, + "step": 4070 + }, + { + "epoch": 0.44706786734021525, + "grad_norm": 2.3598012924194336, + "learning_rate": 1e-06, + "loss": 1.0715, + "mean_token_accuracy": 0.6808609962463379, + "num_tokens": 104764711.0, + "step": 4071 + }, + { + "epoch": 0.4471776850428289, + "grad_norm": 2.165785312652588, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7031105160713196, + "num_tokens": 104789905.0, + "step": 4072 + }, + { + "epoch": 0.44728750274544254, + "grad_norm": 2.307379961013794, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7021558284759521, + "num_tokens": 104813377.0, + "step": 4073 + }, + { + "epoch": 0.44739732044805625, + "grad_norm": 2.4669833183288574, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6979498863220215, + "num_tokens": 104836678.0, + "step": 4074 + }, + { + "epoch": 0.4475071381506699, + "grad_norm": 2.3483734130859375, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6780400276184082, + "num_tokens": 104862477.0, + "step": 4075 + }, + { + "epoch": 0.44761695585328354, + "grad_norm": 2.360997438430786, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7173217535018921, + "num_tokens": 104885777.0, + "step": 4076 + }, + { + "epoch": 0.4477267735558972, + "grad_norm": 2.566833734512329, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7161040902137756, + "num_tokens": 104906286.0, + "step": 4077 + }, + { + "epoch": 0.4478365912585109, + "grad_norm": 2.296748638153076, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7053101658821106, + "num_tokens": 104931132.0, + "step": 4078 + }, + { + "epoch": 0.44794640896112453, + "grad_norm": 2.1154139041900635, + "learning_rate": 1e-06, + "loss": 1.0816, + "mean_token_accuracy": 0.6859956979751587, + "num_tokens": 104961705.0, + "step": 4079 + }, + { + "epoch": 0.4480562266637382, + "grad_norm": 2.351789951324463, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7013698816299438, + "num_tokens": 104986262.0, + "step": 4080 + }, + { + "epoch": 0.4481660443663519, + "grad_norm": 2.4632205963134766, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7098565101623535, + "num_tokens": 105006892.0, + "step": 4081 + }, + { + "epoch": 0.4482758620689655, + "grad_norm": 2.2161288261413574, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6875559091567993, + "num_tokens": 105031438.0, + "step": 4082 + }, + { + "epoch": 0.44838567977157917, + "grad_norm": 2.231351852416992, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7062898278236389, + "num_tokens": 105057267.0, + "step": 4083 + }, + { + "epoch": 0.4484954974741928, + "grad_norm": 2.198253870010376, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7095245718955994, + "num_tokens": 105082537.0, + "step": 4084 + }, + { + "epoch": 0.4486053151768065, + "grad_norm": 2.1465790271759033, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6937854290008545, + "num_tokens": 105109672.0, + "step": 4085 + }, + { + "epoch": 0.44871513287942016, + "grad_norm": 2.1983420848846436, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6951903104782104, + "num_tokens": 105135675.0, + "step": 4086 + }, + { + "epoch": 0.4488249505820338, + "grad_norm": 2.2575197219848633, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7160559296607971, + "num_tokens": 105160578.0, + "step": 4087 + }, + { + "epoch": 0.4489347682846475, + "grad_norm": 2.23724365234375, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7041769027709961, + "num_tokens": 105183454.0, + "step": 4088 + }, + { + "epoch": 0.44904458598726116, + "grad_norm": 2.4957902431488037, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7215940356254578, + "num_tokens": 105203516.0, + "step": 4089 + }, + { + "epoch": 0.4491544036898748, + "grad_norm": 2.4999279975891113, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.697914183139801, + "num_tokens": 105224979.0, + "step": 4090 + }, + { + "epoch": 0.44926422139248845, + "grad_norm": 2.135711431503296, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6836510300636292, + "num_tokens": 105253590.0, + "step": 4091 + }, + { + "epoch": 0.44937403909510215, + "grad_norm": 1.8101078271865845, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6997937560081482, + "num_tokens": 105291105.0, + "step": 4092 + }, + { + "epoch": 0.4494838567977158, + "grad_norm": 2.164022922515869, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7183656096458435, + "num_tokens": 105318804.0, + "step": 4093 + }, + { + "epoch": 0.44959367450032944, + "grad_norm": 2.2173478603363037, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7089030742645264, + "num_tokens": 105343982.0, + "step": 4094 + }, + { + "epoch": 0.4497034922029431, + "grad_norm": 2.010228395462036, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7089540958404541, + "num_tokens": 105373672.0, + "step": 4095 + }, + { + "epoch": 0.4498133099055568, + "grad_norm": 2.1733999252319336, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7334164977073669, + "num_tokens": 105398164.0, + "step": 4096 + }, + { + "epoch": 0.44992312760817044, + "grad_norm": 2.2562174797058105, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7075417637825012, + "num_tokens": 105424212.0, + "step": 4097 + }, + { + "epoch": 0.4500329453107841, + "grad_norm": 2.225198268890381, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.721147894859314, + "num_tokens": 105447927.0, + "step": 4098 + }, + { + "epoch": 0.4501427630133978, + "grad_norm": 2.0139076709747314, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6931570768356323, + "num_tokens": 105477155.0, + "step": 4099 + }, + { + "epoch": 0.45025258071601143, + "grad_norm": 2.283364772796631, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7061493396759033, + "num_tokens": 105504013.0, + "step": 4100 + }, + { + "epoch": 0.4503623984186251, + "grad_norm": 2.120023488998413, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7117515802383423, + "num_tokens": 105532357.0, + "step": 4101 + }, + { + "epoch": 0.4504722161212387, + "grad_norm": 1.932651162147522, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7231723070144653, + "num_tokens": 105559629.0, + "step": 4102 + }, + { + "epoch": 0.4505820338238524, + "grad_norm": 2.0862350463867188, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6925641298294067, + "num_tokens": 105587944.0, + "step": 4103 + }, + { + "epoch": 0.45069185152646607, + "grad_norm": 2.111138343811035, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6898998022079468, + "num_tokens": 105615532.0, + "step": 4104 + }, + { + "epoch": 0.4508016692290797, + "grad_norm": 2.462547540664673, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.6879454851150513, + "num_tokens": 105638353.0, + "step": 4105 + }, + { + "epoch": 0.45091148693169336, + "grad_norm": 2.266969919204712, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7128167748451233, + "num_tokens": 105660494.0, + "step": 4106 + }, + { + "epoch": 0.45102130463430706, + "grad_norm": 2.045720338821411, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6986753940582275, + "num_tokens": 105688801.0, + "step": 4107 + }, + { + "epoch": 0.4511311223369207, + "grad_norm": 2.3884236812591553, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.721264123916626, + "num_tokens": 105711935.0, + "step": 4108 + }, + { + "epoch": 0.45124094003953436, + "grad_norm": 2.2246599197387695, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7310881018638611, + "num_tokens": 105736577.0, + "step": 4109 + }, + { + "epoch": 0.45135075774214806, + "grad_norm": 2.304413080215454, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7057797312736511, + "num_tokens": 105759231.0, + "step": 4110 + }, + { + "epoch": 0.4514605754447617, + "grad_norm": 2.2543785572052, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6910349130630493, + "num_tokens": 105785415.0, + "step": 4111 + }, + { + "epoch": 0.45157039314737535, + "grad_norm": 2.07181978225708, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.7008310556411743, + "num_tokens": 105814141.0, + "step": 4112 + }, + { + "epoch": 0.451680210849989, + "grad_norm": 2.495086431503296, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6966228485107422, + "num_tokens": 105835873.0, + "step": 4113 + }, + { + "epoch": 0.4517900285526027, + "grad_norm": 1.9973113536834717, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6868984699249268, + "num_tokens": 105868741.0, + "step": 4114 + }, + { + "epoch": 0.45189984625521634, + "grad_norm": 1.9435774087905884, + "learning_rate": 1e-06, + "loss": 1.0835, + "mean_token_accuracy": 0.6751309633255005, + "num_tokens": 105904708.0, + "step": 4115 + }, + { + "epoch": 0.45200966395783, + "grad_norm": 2.5016839504241943, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6983170509338379, + "num_tokens": 105927977.0, + "step": 4116 + }, + { + "epoch": 0.4521194816604437, + "grad_norm": 2.048316717147827, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6938179731369019, + "num_tokens": 105957713.0, + "step": 4117 + }, + { + "epoch": 0.45222929936305734, + "grad_norm": 2.2400615215301514, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7069880962371826, + "num_tokens": 105981596.0, + "step": 4118 + }, + { + "epoch": 0.452339117065671, + "grad_norm": 2.1101880073547363, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.680435061454773, + "num_tokens": 106012154.0, + "step": 4119 + }, + { + "epoch": 0.4524489347682846, + "grad_norm": 2.051976203918457, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6913524866104126, + "num_tokens": 106040671.0, + "step": 4120 + }, + { + "epoch": 0.45255875247089833, + "grad_norm": 2.0459039211273193, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7243633270263672, + "num_tokens": 106066740.0, + "step": 4121 + }, + { + "epoch": 0.452668570173512, + "grad_norm": 2.0648415088653564, + "learning_rate": 1e-06, + "loss": 1.1586, + "mean_token_accuracy": 0.6503161191940308, + "num_tokens": 106099801.0, + "step": 4122 + }, + { + "epoch": 0.4527783878761256, + "grad_norm": 2.071255683898926, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7078797817230225, + "num_tokens": 106127563.0, + "step": 4123 + }, + { + "epoch": 0.45288820557873927, + "grad_norm": 2.465179204940796, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7166441679000854, + "num_tokens": 106147243.0, + "step": 4124 + }, + { + "epoch": 0.45299802328135297, + "grad_norm": 2.072357177734375, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7049170732498169, + "num_tokens": 106175165.0, + "step": 4125 + }, + { + "epoch": 0.4531078409839666, + "grad_norm": 2.3314599990844727, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7035865783691406, + "num_tokens": 106199252.0, + "step": 4126 + }, + { + "epoch": 0.45321765868658026, + "grad_norm": 2.3927295207977295, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7060651779174805, + "num_tokens": 106221548.0, + "step": 4127 + }, + { + "epoch": 0.45332747638919396, + "grad_norm": 2.226109027862549, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7047498226165771, + "num_tokens": 106246409.0, + "step": 4128 + }, + { + "epoch": 0.4534372940918076, + "grad_norm": 2.2180585861206055, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6875901818275452, + "num_tokens": 106271873.0, + "step": 4129 + }, + { + "epoch": 0.45354711179442125, + "grad_norm": 2.3688948154449463, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.717851996421814, + "num_tokens": 106294307.0, + "step": 4130 + }, + { + "epoch": 0.4536569294970349, + "grad_norm": 2.274724006652832, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6961165070533752, + "num_tokens": 106317577.0, + "step": 4131 + }, + { + "epoch": 0.4537667471996486, + "grad_norm": 2.2163875102996826, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7045587301254272, + "num_tokens": 106344042.0, + "step": 4132 + }, + { + "epoch": 0.45387656490226225, + "grad_norm": 2.5631511211395264, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.6993846297264099, + "num_tokens": 106363940.0, + "step": 4133 + }, + { + "epoch": 0.4539863826048759, + "grad_norm": 2.1316182613372803, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6880096197128296, + "num_tokens": 106391976.0, + "step": 4134 + }, + { + "epoch": 0.4540962003074896, + "grad_norm": 2.226489782333374, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7048019170761108, + "num_tokens": 106417719.0, + "step": 4135 + }, + { + "epoch": 0.45420601801010324, + "grad_norm": 2.242453098297119, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.703808605670929, + "num_tokens": 106441841.0, + "step": 4136 + }, + { + "epoch": 0.4543158357127169, + "grad_norm": 2.2973098754882812, + "learning_rate": 1e-06, + "loss": 1.0928, + "mean_token_accuracy": 0.6812832355499268, + "num_tokens": 106467146.0, + "step": 4137 + }, + { + "epoch": 0.45442565341533053, + "grad_norm": 2.427777051925659, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7249535322189331, + "num_tokens": 106486311.0, + "step": 4138 + }, + { + "epoch": 0.45453547111794423, + "grad_norm": 2.4875991344451904, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7118375301361084, + "num_tokens": 106507570.0, + "step": 4139 + }, + { + "epoch": 0.4546452888205579, + "grad_norm": 2.187714099884033, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6920658349990845, + "num_tokens": 106534278.0, + "step": 4140 + }, + { + "epoch": 0.4547551065231715, + "grad_norm": 2.0240581035614014, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6913079023361206, + "num_tokens": 106563114.0, + "step": 4141 + }, + { + "epoch": 0.45486492422578517, + "grad_norm": 2.2334694862365723, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6957458853721619, + "num_tokens": 106586541.0, + "step": 4142 + }, + { + "epoch": 0.4549747419283989, + "grad_norm": 2.250046730041504, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7256482839584351, + "num_tokens": 106609335.0, + "step": 4143 + }, + { + "epoch": 0.4550845596310125, + "grad_norm": 2.4558632373809814, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6886493563652039, + "num_tokens": 106632133.0, + "step": 4144 + }, + { + "epoch": 0.45519437733362617, + "grad_norm": 2.1369640827178955, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7060213088989258, + "num_tokens": 106657637.0, + "step": 4145 + }, + { + "epoch": 0.45530419503623987, + "grad_norm": 2.4671847820281982, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7122036814689636, + "num_tokens": 106678038.0, + "step": 4146 + }, + { + "epoch": 0.4554140127388535, + "grad_norm": 2.025721311569214, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6940391063690186, + "num_tokens": 106708995.0, + "step": 4147 + }, + { + "epoch": 0.45552383044146716, + "grad_norm": 2.050105333328247, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7026830911636353, + "num_tokens": 106738332.0, + "step": 4148 + }, + { + "epoch": 0.4556336481440808, + "grad_norm": 2.533297061920166, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7135378122329712, + "num_tokens": 106757474.0, + "step": 4149 + }, + { + "epoch": 0.4557434658466945, + "grad_norm": 2.0950708389282227, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7020453214645386, + "num_tokens": 106785161.0, + "step": 4150 + }, + { + "epoch": 0.45585328354930815, + "grad_norm": 2.420546531677246, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7158055901527405, + "num_tokens": 106806427.0, + "step": 4151 + }, + { + "epoch": 0.4559631012519218, + "grad_norm": 2.272925853729248, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7027692198753357, + "num_tokens": 106831411.0, + "step": 4152 + }, + { + "epoch": 0.45607291895453544, + "grad_norm": 2.0737407207489014, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.7171302437782288, + "num_tokens": 106859579.0, + "step": 4153 + }, + { + "epoch": 0.45618273665714915, + "grad_norm": 2.23749041557312, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7106225490570068, + "num_tokens": 106884309.0, + "step": 4154 + }, + { + "epoch": 0.4562925543597628, + "grad_norm": 2.0256447792053223, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7142086029052734, + "num_tokens": 106912462.0, + "step": 4155 + }, + { + "epoch": 0.45640237206237644, + "grad_norm": 2.056063652038574, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7181805968284607, + "num_tokens": 106941965.0, + "step": 4156 + }, + { + "epoch": 0.45651218976499014, + "grad_norm": 2.2421257495880127, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7352457642555237, + "num_tokens": 106964705.0, + "step": 4157 + }, + { + "epoch": 0.4566220074676038, + "grad_norm": 2.0156800746917725, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.7051208019256592, + "num_tokens": 106993763.0, + "step": 4158 + }, + { + "epoch": 0.45673182517021743, + "grad_norm": 2.4872097969055176, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7109780311584473, + "num_tokens": 107012972.0, + "step": 4159 + }, + { + "epoch": 0.4568416428728311, + "grad_norm": 2.130873441696167, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7082475423812866, + "num_tokens": 107039156.0, + "step": 4160 + }, + { + "epoch": 0.4569514605754448, + "grad_norm": 1.7286005020141602, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7212393283843994, + "num_tokens": 107073866.0, + "step": 4161 + }, + { + "epoch": 0.4570612782780584, + "grad_norm": 2.091343641281128, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6990393400192261, + "num_tokens": 107100051.0, + "step": 4162 + }, + { + "epoch": 0.45717109598067207, + "grad_norm": 2.0357213020324707, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7141836881637573, + "num_tokens": 107127085.0, + "step": 4163 + }, + { + "epoch": 0.4572809136832858, + "grad_norm": 2.391148328781128, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7282083630561829, + "num_tokens": 107149230.0, + "step": 4164 + }, + { + "epoch": 0.4573907313858994, + "grad_norm": 2.461620569229126, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7075755596160889, + "num_tokens": 107170014.0, + "step": 4165 + }, + { + "epoch": 0.45750054908851306, + "grad_norm": 2.020624876022339, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.6790662407875061, + "num_tokens": 107200539.0, + "step": 4166 + }, + { + "epoch": 0.4576103667911267, + "grad_norm": 2.1487479209899902, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.7067302465438843, + "num_tokens": 107226131.0, + "step": 4167 + }, + { + "epoch": 0.4577201844937404, + "grad_norm": 2.013559579849243, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.701553463935852, + "num_tokens": 107256022.0, + "step": 4168 + }, + { + "epoch": 0.45783000219635406, + "grad_norm": 2.229081630706787, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7014517784118652, + "num_tokens": 107281057.0, + "step": 4169 + }, + { + "epoch": 0.4579398198989677, + "grad_norm": 2.3726046085357666, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7332923412322998, + "num_tokens": 107301786.0, + "step": 4170 + }, + { + "epoch": 0.45804963760158135, + "grad_norm": 2.377619504928589, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7185350656509399, + "num_tokens": 107322295.0, + "step": 4171 + }, + { + "epoch": 0.45815945530419505, + "grad_norm": 2.365068197250366, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7194226384162903, + "num_tokens": 107343314.0, + "step": 4172 + }, + { + "epoch": 0.4582692730068087, + "grad_norm": 2.1594035625457764, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6910192966461182, + "num_tokens": 107370904.0, + "step": 4173 + }, + { + "epoch": 0.45837909070942234, + "grad_norm": 1.9954975843429565, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7051412463188171, + "num_tokens": 107399845.0, + "step": 4174 + }, + { + "epoch": 0.45848890841203604, + "grad_norm": 2.1340456008911133, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7017403244972229, + "num_tokens": 107426886.0, + "step": 4175 + }, + { + "epoch": 0.4585987261146497, + "grad_norm": 2.2519938945770264, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.698452353477478, + "num_tokens": 107453578.0, + "step": 4176 + }, + { + "epoch": 0.45870854381726334, + "grad_norm": 2.3529367446899414, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6936224699020386, + "num_tokens": 107476287.0, + "step": 4177 + }, + { + "epoch": 0.458818361519877, + "grad_norm": 2.2367265224456787, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7096617221832275, + "num_tokens": 107499937.0, + "step": 4178 + }, + { + "epoch": 0.4589281792224907, + "grad_norm": 2.1896772384643555, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6937469840049744, + "num_tokens": 107526076.0, + "step": 4179 + }, + { + "epoch": 0.45903799692510433, + "grad_norm": 2.2817575931549072, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7245648503303528, + "num_tokens": 107548187.0, + "step": 4180 + }, + { + "epoch": 0.459147814627718, + "grad_norm": 2.4183297157287598, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7081661820411682, + "num_tokens": 107570339.0, + "step": 4181 + }, + { + "epoch": 0.4592576323303316, + "grad_norm": 2.0797600746154785, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6938987970352173, + "num_tokens": 107598615.0, + "step": 4182 + }, + { + "epoch": 0.4593674500329453, + "grad_norm": 2.312770128250122, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7013229727745056, + "num_tokens": 107623320.0, + "step": 4183 + }, + { + "epoch": 0.45947726773555897, + "grad_norm": 2.3005175590515137, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.702140748500824, + "num_tokens": 107648208.0, + "step": 4184 + }, + { + "epoch": 0.4595870854381726, + "grad_norm": 2.102529764175415, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6993187069892883, + "num_tokens": 107677891.0, + "step": 4185 + }, + { + "epoch": 0.4596969031407863, + "grad_norm": 2.3707268238067627, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7006720304489136, + "num_tokens": 107701472.0, + "step": 4186 + }, + { + "epoch": 0.45980672084339996, + "grad_norm": 2.1655704975128174, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.6872327327728271, + "num_tokens": 107727476.0, + "step": 4187 + }, + { + "epoch": 0.4599165385460136, + "grad_norm": 2.135559320449829, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7061001062393188, + "num_tokens": 107757978.0, + "step": 4188 + }, + { + "epoch": 0.46002635624862726, + "grad_norm": 2.4944303035736084, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7056986689567566, + "num_tokens": 107779783.0, + "step": 4189 + }, + { + "epoch": 0.46013617395124096, + "grad_norm": 2.3250317573547363, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7020100951194763, + "num_tokens": 107803785.0, + "step": 4190 + }, + { + "epoch": 0.4602459916538546, + "grad_norm": 2.1058948040008545, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6981573104858398, + "num_tokens": 107832256.0, + "step": 4191 + }, + { + "epoch": 0.46035580935646825, + "grad_norm": 2.1894476413726807, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6919292211532593, + "num_tokens": 107860024.0, + "step": 4192 + }, + { + "epoch": 0.46046562705908195, + "grad_norm": 2.369921922683716, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7121583223342896, + "num_tokens": 107883440.0, + "step": 4193 + }, + { + "epoch": 0.4605754447616956, + "grad_norm": 2.257908582687378, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7343305349349976, + "num_tokens": 107906399.0, + "step": 4194 + }, + { + "epoch": 0.46068526246430924, + "grad_norm": 1.9453336000442505, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6778960227966309, + "num_tokens": 107937697.0, + "step": 4195 + }, + { + "epoch": 0.4607950801669229, + "grad_norm": 2.2270073890686035, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6948834657669067, + "num_tokens": 107963099.0, + "step": 4196 + }, + { + "epoch": 0.4609048978695366, + "grad_norm": 2.4063844680786133, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.705304741859436, + "num_tokens": 107986158.0, + "step": 4197 + }, + { + "epoch": 0.46101471557215024, + "grad_norm": 2.215433120727539, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7073580026626587, + "num_tokens": 108011574.0, + "step": 4198 + }, + { + "epoch": 0.4611245332747639, + "grad_norm": 2.4396398067474365, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7006404399871826, + "num_tokens": 108034737.0, + "step": 4199 + }, + { + "epoch": 0.4612343509773775, + "grad_norm": 2.2928261756896973, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7095326781272888, + "num_tokens": 108058345.0, + "step": 4200 + }, + { + "epoch": 0.46134416867999123, + "grad_norm": 1.9247033596038818, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6840331554412842, + "num_tokens": 108092081.0, + "step": 4201 + }, + { + "epoch": 0.4614539863826049, + "grad_norm": 1.8224836587905884, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6878586411476135, + "num_tokens": 108128108.0, + "step": 4202 + }, + { + "epoch": 0.4615638040852185, + "grad_norm": 2.2184269428253174, + "learning_rate": 1e-06, + "loss": 1.1465, + "mean_token_accuracy": 0.6581050157546997, + "num_tokens": 108154367.0, + "step": 4203 + }, + { + "epoch": 0.4616736217878322, + "grad_norm": 2.013798952102661, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.715083122253418, + "num_tokens": 108186487.0, + "step": 4204 + }, + { + "epoch": 0.46178343949044587, + "grad_norm": 2.053152322769165, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6842949390411377, + "num_tokens": 108217650.0, + "step": 4205 + }, + { + "epoch": 0.4618932571930595, + "grad_norm": 2.471484899520874, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6981774568557739, + "num_tokens": 108240773.0, + "step": 4206 + }, + { + "epoch": 0.46200307489567316, + "grad_norm": 2.2477355003356934, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6946083903312683, + "num_tokens": 108265428.0, + "step": 4207 + }, + { + "epoch": 0.46211289259828686, + "grad_norm": 2.025912284851074, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7117699980735779, + "num_tokens": 108294088.0, + "step": 4208 + }, + { + "epoch": 0.4622227103009005, + "grad_norm": 2.052419900894165, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7014681100845337, + "num_tokens": 108322754.0, + "step": 4209 + }, + { + "epoch": 0.46233252800351415, + "grad_norm": 2.120410919189453, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7005203366279602, + "num_tokens": 108347918.0, + "step": 4210 + }, + { + "epoch": 0.46244234570612786, + "grad_norm": 2.3905110359191895, + "learning_rate": 1e-06, + "loss": 1.1171, + "mean_token_accuracy": 0.6705873012542725, + "num_tokens": 108370353.0, + "step": 4211 + }, + { + "epoch": 0.4625521634087415, + "grad_norm": 2.1271941661834717, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6957350969314575, + "num_tokens": 108396712.0, + "step": 4212 + }, + { + "epoch": 0.46266198111135515, + "grad_norm": 2.1579034328460693, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.702302873134613, + "num_tokens": 108423425.0, + "step": 4213 + }, + { + "epoch": 0.4627717988139688, + "grad_norm": 2.1464931964874268, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7260921001434326, + "num_tokens": 108447918.0, + "step": 4214 + }, + { + "epoch": 0.4628816165165825, + "grad_norm": 2.2933714389801025, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7167505025863647, + "num_tokens": 108473075.0, + "step": 4215 + }, + { + "epoch": 0.46299143421919614, + "grad_norm": 1.9883337020874023, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.723773717880249, + "num_tokens": 108501121.0, + "step": 4216 + }, + { + "epoch": 0.4631012519218098, + "grad_norm": 2.201474666595459, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7142657041549683, + "num_tokens": 108525481.0, + "step": 4217 + }, + { + "epoch": 0.46321106962442343, + "grad_norm": 2.3986570835113525, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7215425968170166, + "num_tokens": 108548806.0, + "step": 4218 + }, + { + "epoch": 0.46332088732703713, + "grad_norm": 2.0627634525299072, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7010596990585327, + "num_tokens": 108578241.0, + "step": 4219 + }, + { + "epoch": 0.4634307050296508, + "grad_norm": 1.8588038682937622, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6957991123199463, + "num_tokens": 108614997.0, + "step": 4220 + }, + { + "epoch": 0.4635405227322644, + "grad_norm": 2.3229098320007324, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7067729234695435, + "num_tokens": 108638223.0, + "step": 4221 + }, + { + "epoch": 0.4636503404348781, + "grad_norm": 2.2462387084960938, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7157759666442871, + "num_tokens": 108662986.0, + "step": 4222 + }, + { + "epoch": 0.4637601581374918, + "grad_norm": 2.031240940093994, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.695717453956604, + "num_tokens": 108692881.0, + "step": 4223 + }, + { + "epoch": 0.4638699758401054, + "grad_norm": 2.079963207244873, + "learning_rate": 1e-06, + "loss": 1.1238, + "mean_token_accuracy": 0.6686476469039917, + "num_tokens": 108722729.0, + "step": 4224 + }, + { + "epoch": 0.46397979354271907, + "grad_norm": 2.22733473777771, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6989899277687073, + "num_tokens": 108748125.0, + "step": 4225 + }, + { + "epoch": 0.46408961124533277, + "grad_norm": 2.1273038387298584, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6988425254821777, + "num_tokens": 108777915.0, + "step": 4226 + }, + { + "epoch": 0.4641994289479464, + "grad_norm": 2.161548376083374, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7315571308135986, + "num_tokens": 108803144.0, + "step": 4227 + }, + { + "epoch": 0.46430924665056006, + "grad_norm": 2.3956995010375977, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6888934969902039, + "num_tokens": 108824680.0, + "step": 4228 + }, + { + "epoch": 0.4644190643531737, + "grad_norm": 2.231592893600464, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.699148416519165, + "num_tokens": 108849506.0, + "step": 4229 + }, + { + "epoch": 0.4645288820557874, + "grad_norm": 1.9956539869308472, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6971306800842285, + "num_tokens": 108878839.0, + "step": 4230 + }, + { + "epoch": 0.46463869975840105, + "grad_norm": 2.3659071922302246, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7241708040237427, + "num_tokens": 108899883.0, + "step": 4231 + }, + { + "epoch": 0.4647485174610147, + "grad_norm": 2.3980209827423096, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7079658508300781, + "num_tokens": 108923504.0, + "step": 4232 + }, + { + "epoch": 0.4648583351636284, + "grad_norm": 2.120548725128174, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7212988138198853, + "num_tokens": 108950611.0, + "step": 4233 + }, + { + "epoch": 0.46496815286624205, + "grad_norm": 2.0286917686462402, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7009799480438232, + "num_tokens": 108978274.0, + "step": 4234 + }, + { + "epoch": 0.4650779705688557, + "grad_norm": 2.068984270095825, + "learning_rate": 1e-06, + "loss": 1.126, + "mean_token_accuracy": 0.6726008057594299, + "num_tokens": 109008703.0, + "step": 4235 + }, + { + "epoch": 0.46518778827146934, + "grad_norm": 1.9640517234802246, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7231441140174866, + "num_tokens": 109038187.0, + "step": 4236 + }, + { + "epoch": 0.46529760597408304, + "grad_norm": 2.2897932529449463, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7012794017791748, + "num_tokens": 109062990.0, + "step": 4237 + }, + { + "epoch": 0.4654074236766967, + "grad_norm": 2.0418641567230225, + "learning_rate": 1e-06, + "loss": 1.0659, + "mean_token_accuracy": 0.6787682771682739, + "num_tokens": 109092601.0, + "step": 4238 + }, + { + "epoch": 0.46551724137931033, + "grad_norm": 2.1685879230499268, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.7072845697402954, + "num_tokens": 109120279.0, + "step": 4239 + }, + { + "epoch": 0.46562705908192403, + "grad_norm": 2.358358383178711, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7029826641082764, + "num_tokens": 109145534.0, + "step": 4240 + }, + { + "epoch": 0.4657368767845377, + "grad_norm": 2.1886518001556396, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6943995952606201, + "num_tokens": 109170233.0, + "step": 4241 + }, + { + "epoch": 0.4658466944871513, + "grad_norm": 2.3123550415039062, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7151255011558533, + "num_tokens": 109191440.0, + "step": 4242 + }, + { + "epoch": 0.46595651218976497, + "grad_norm": 2.334563732147217, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7012783288955688, + "num_tokens": 109214823.0, + "step": 4243 + }, + { + "epoch": 0.4660663298923787, + "grad_norm": 2.186272382736206, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.696841835975647, + "num_tokens": 109241495.0, + "step": 4244 + }, + { + "epoch": 0.4661761475949923, + "grad_norm": 1.9850187301635742, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7084932327270508, + "num_tokens": 109270865.0, + "step": 4245 + }, + { + "epoch": 0.46628596529760596, + "grad_norm": 2.5818240642547607, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7053174376487732, + "num_tokens": 109290885.0, + "step": 4246 + }, + { + "epoch": 0.4663957830002196, + "grad_norm": 1.9512052536010742, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7018462419509888, + "num_tokens": 109320961.0, + "step": 4247 + }, + { + "epoch": 0.4665056007028333, + "grad_norm": 2.546543598175049, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7041499018669128, + "num_tokens": 109342695.0, + "step": 4248 + }, + { + "epoch": 0.46661541840544696, + "grad_norm": 2.145747184753418, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.690366268157959, + "num_tokens": 109371180.0, + "step": 4249 + }, + { + "epoch": 0.4667252361080606, + "grad_norm": 2.367119312286377, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7015681862831116, + "num_tokens": 109393475.0, + "step": 4250 + }, + { + "epoch": 0.4668350538106743, + "grad_norm": 2.4301860332489014, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6933331489562988, + "num_tokens": 109416288.0, + "step": 4251 + }, + { + "epoch": 0.46694487151328795, + "grad_norm": 2.421767234802246, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7219958901405334, + "num_tokens": 109436149.0, + "step": 4252 + }, + { + "epoch": 0.4670546892159016, + "grad_norm": 2.097336530685425, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7093194723129272, + "num_tokens": 109462294.0, + "step": 4253 + }, + { + "epoch": 0.46716450691851524, + "grad_norm": 2.129885673522949, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7068586945533752, + "num_tokens": 109486708.0, + "step": 4254 + }, + { + "epoch": 0.46727432462112894, + "grad_norm": 2.340397834777832, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7120967507362366, + "num_tokens": 109508285.0, + "step": 4255 + }, + { + "epoch": 0.4673841423237426, + "grad_norm": 2.606919765472412, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7200819253921509, + "num_tokens": 109532288.0, + "step": 4256 + }, + { + "epoch": 0.46749396002635624, + "grad_norm": 2.088761806488037, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6858108639717102, + "num_tokens": 109560817.0, + "step": 4257 + }, + { + "epoch": 0.4676037777289699, + "grad_norm": 2.3699300289154053, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7124176025390625, + "num_tokens": 109583107.0, + "step": 4258 + }, + { + "epoch": 0.4677135954315836, + "grad_norm": 2.2889809608459473, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6976605653762817, + "num_tokens": 109607925.0, + "step": 4259 + }, + { + "epoch": 0.46782341313419723, + "grad_norm": 2.2841246128082275, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7071943283081055, + "num_tokens": 109631005.0, + "step": 4260 + }, + { + "epoch": 0.4679332308368109, + "grad_norm": 2.127159595489502, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6909945011138916, + "num_tokens": 109657873.0, + "step": 4261 + }, + { + "epoch": 0.4680430485394246, + "grad_norm": 2.4532687664031982, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7031639814376831, + "num_tokens": 109678463.0, + "step": 4262 + }, + { + "epoch": 0.4681528662420382, + "grad_norm": 2.5551083087921143, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7041234970092773, + "num_tokens": 109698252.0, + "step": 4263 + }, + { + "epoch": 0.46826268394465187, + "grad_norm": 2.2852306365966797, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7024440765380859, + "num_tokens": 109722624.0, + "step": 4264 + }, + { + "epoch": 0.4683725016472655, + "grad_norm": 2.2600765228271484, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7048035264015198, + "num_tokens": 109748451.0, + "step": 4265 + }, + { + "epoch": 0.4684823193498792, + "grad_norm": 2.1478781700134277, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7164762616157532, + "num_tokens": 109775970.0, + "step": 4266 + }, + { + "epoch": 0.46859213705249286, + "grad_norm": 2.5528666973114014, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7277848720550537, + "num_tokens": 109795635.0, + "step": 4267 + }, + { + "epoch": 0.4687019547551065, + "grad_norm": 2.5405499935150146, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7118114233016968, + "num_tokens": 109815558.0, + "step": 4268 + }, + { + "epoch": 0.4688117724577202, + "grad_norm": 2.171945571899414, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7097290754318237, + "num_tokens": 109841907.0, + "step": 4269 + }, + { + "epoch": 0.46892159016033386, + "grad_norm": 2.264907121658325, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.705663800239563, + "num_tokens": 109865648.0, + "step": 4270 + }, + { + "epoch": 0.4690314078629475, + "grad_norm": 2.2394967079162598, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7043246626853943, + "num_tokens": 109890061.0, + "step": 4271 + }, + { + "epoch": 0.46914122556556115, + "grad_norm": 2.1985793113708496, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6897460222244263, + "num_tokens": 109915167.0, + "step": 4272 + }, + { + "epoch": 0.46925104326817485, + "grad_norm": 2.0341851711273193, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7147351503372192, + "num_tokens": 109942571.0, + "step": 4273 + }, + { + "epoch": 0.4693608609707885, + "grad_norm": 2.264305353164673, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7053849697113037, + "num_tokens": 109966312.0, + "step": 4274 + }, + { + "epoch": 0.46947067867340214, + "grad_norm": 1.9957767724990845, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7099313139915466, + "num_tokens": 109993144.0, + "step": 4275 + }, + { + "epoch": 0.4695804963760158, + "grad_norm": 2.25793194770813, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6974001526832581, + "num_tokens": 110018484.0, + "step": 4276 + }, + { + "epoch": 0.4696903140786295, + "grad_norm": 2.5202205181121826, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7012207508087158, + "num_tokens": 110039220.0, + "step": 4277 + }, + { + "epoch": 0.46980013178124314, + "grad_norm": 2.0276997089385986, + "learning_rate": 1e-06, + "loss": 1.0747, + "mean_token_accuracy": 0.6768745183944702, + "num_tokens": 110068674.0, + "step": 4278 + }, + { + "epoch": 0.4699099494838568, + "grad_norm": 2.2561280727386475, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7088009119033813, + "num_tokens": 110093167.0, + "step": 4279 + }, + { + "epoch": 0.4700197671864705, + "grad_norm": 2.0156211853027344, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6829625368118286, + "num_tokens": 110121996.0, + "step": 4280 + }, + { + "epoch": 0.47012958488908413, + "grad_norm": 2.4472880363464355, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7003516554832458, + "num_tokens": 110143648.0, + "step": 4281 + }, + { + "epoch": 0.4702394025916978, + "grad_norm": 2.022451162338257, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6922097206115723, + "num_tokens": 110172986.0, + "step": 4282 + }, + { + "epoch": 0.4703492202943114, + "grad_norm": 2.4090378284454346, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.724341630935669, + "num_tokens": 110196325.0, + "step": 4283 + }, + { + "epoch": 0.4704590379969251, + "grad_norm": 2.2085461616516113, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7099874019622803, + "num_tokens": 110222174.0, + "step": 4284 + }, + { + "epoch": 0.47056885569953877, + "grad_norm": 2.2794885635375977, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6895007491111755, + "num_tokens": 110245944.0, + "step": 4285 + }, + { + "epoch": 0.4706786734021524, + "grad_norm": 2.2466869354248047, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6909703612327576, + "num_tokens": 110272476.0, + "step": 4286 + }, + { + "epoch": 0.4707884911047661, + "grad_norm": 2.3436172008514404, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6908644437789917, + "num_tokens": 110297567.0, + "step": 4287 + }, + { + "epoch": 0.47089830880737976, + "grad_norm": 2.1620633602142334, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6829243898391724, + "num_tokens": 110326323.0, + "step": 4288 + }, + { + "epoch": 0.4710081265099934, + "grad_norm": 1.841981291770935, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6988533139228821, + "num_tokens": 110359959.0, + "step": 4289 + }, + { + "epoch": 0.47111794421260705, + "grad_norm": 2.314532518386841, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6967021226882935, + "num_tokens": 110386388.0, + "step": 4290 + }, + { + "epoch": 0.47122776191522076, + "grad_norm": 2.3753061294555664, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7127649784088135, + "num_tokens": 110407345.0, + "step": 4291 + }, + { + "epoch": 0.4713375796178344, + "grad_norm": 1.9643696546554565, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.6947307586669922, + "num_tokens": 110439542.0, + "step": 4292 + }, + { + "epoch": 0.47144739732044805, + "grad_norm": 2.024665594100952, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7018822431564331, + "num_tokens": 110472069.0, + "step": 4293 + }, + { + "epoch": 0.4715572150230617, + "grad_norm": 2.1371688842773438, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6982601284980774, + "num_tokens": 110498050.0, + "step": 4294 + }, + { + "epoch": 0.4716670327256754, + "grad_norm": 2.1417720317840576, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7194226980209351, + "num_tokens": 110521684.0, + "step": 4295 + }, + { + "epoch": 0.47177685042828904, + "grad_norm": 2.106369733810425, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6999293565750122, + "num_tokens": 110551696.0, + "step": 4296 + }, + { + "epoch": 0.4718866681309027, + "grad_norm": 2.1753697395324707, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7299716472625732, + "num_tokens": 110578496.0, + "step": 4297 + }, + { + "epoch": 0.4719964858335164, + "grad_norm": 2.19069766998291, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6925477385520935, + "num_tokens": 110603978.0, + "step": 4298 + }, + { + "epoch": 0.47210630353613003, + "grad_norm": 2.2900874614715576, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6903765201568604, + "num_tokens": 110630661.0, + "step": 4299 + }, + { + "epoch": 0.4722161212387437, + "grad_norm": 2.0270798206329346, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6883503198623657, + "num_tokens": 110661383.0, + "step": 4300 + }, + { + "epoch": 0.4723259389413573, + "grad_norm": 2.066375494003296, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7098567485809326, + "num_tokens": 110689321.0, + "step": 4301 + }, + { + "epoch": 0.472435756643971, + "grad_norm": 2.398437976837158, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7159664630889893, + "num_tokens": 110711962.0, + "step": 4302 + }, + { + "epoch": 0.4725455743465847, + "grad_norm": 2.147690534591675, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6976915001869202, + "num_tokens": 110739707.0, + "step": 4303 + }, + { + "epoch": 0.4726553920491983, + "grad_norm": 2.1270806789398193, + "learning_rate": 1e-06, + "loss": 1.11, + "mean_token_accuracy": 0.671326220035553, + "num_tokens": 110771473.0, + "step": 4304 + }, + { + "epoch": 0.47276520975181197, + "grad_norm": 2.1777822971343994, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6800601482391357, + "num_tokens": 110801163.0, + "step": 4305 + }, + { + "epoch": 0.47287502745442567, + "grad_norm": 2.24869966506958, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7100377082824707, + "num_tokens": 110827883.0, + "step": 4306 + }, + { + "epoch": 0.4729848451570393, + "grad_norm": 2.281506061553955, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6955533027648926, + "num_tokens": 110854280.0, + "step": 4307 + }, + { + "epoch": 0.47309466285965296, + "grad_norm": 2.403143882751465, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7262231111526489, + "num_tokens": 110876346.0, + "step": 4308 + }, + { + "epoch": 0.47320448056226666, + "grad_norm": 2.464698076248169, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6931010484695435, + "num_tokens": 110898149.0, + "step": 4309 + }, + { + "epoch": 0.4733142982648803, + "grad_norm": 2.0078022480010986, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7096003890037537, + "num_tokens": 110927301.0, + "step": 4310 + }, + { + "epoch": 0.47342411596749395, + "grad_norm": 2.1199164390563965, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.6986745595932007, + "num_tokens": 110953398.0, + "step": 4311 + }, + { + "epoch": 0.4735339336701076, + "grad_norm": 2.117255687713623, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7064855694770813, + "num_tokens": 110979306.0, + "step": 4312 + }, + { + "epoch": 0.4736437513727213, + "grad_norm": 2.1886322498321533, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7233668565750122, + "num_tokens": 111005062.0, + "step": 4313 + }, + { + "epoch": 0.47375356907533495, + "grad_norm": 1.8629319667816162, + "learning_rate": 1e-06, + "loss": 1.0785, + "mean_token_accuracy": 0.6863786578178406, + "num_tokens": 111037440.0, + "step": 4314 + }, + { + "epoch": 0.4738633867779486, + "grad_norm": 2.3496453762054443, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7083075046539307, + "num_tokens": 111060748.0, + "step": 4315 + }, + { + "epoch": 0.4739732044805623, + "grad_norm": 2.4310476779937744, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7165724039077759, + "num_tokens": 111082683.0, + "step": 4316 + }, + { + "epoch": 0.47408302218317594, + "grad_norm": 2.0720951557159424, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6863459348678589, + "num_tokens": 111112223.0, + "step": 4317 + }, + { + "epoch": 0.4741928398857896, + "grad_norm": 2.0512006282806396, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6971104145050049, + "num_tokens": 111139272.0, + "step": 4318 + }, + { + "epoch": 0.47430265758840323, + "grad_norm": 2.04998517036438, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6817243099212646, + "num_tokens": 111170050.0, + "step": 4319 + }, + { + "epoch": 0.47441247529101693, + "grad_norm": 2.008112668991089, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7135761976242065, + "num_tokens": 111197742.0, + "step": 4320 + }, + { + "epoch": 0.4745222929936306, + "grad_norm": 2.1130454540252686, + "learning_rate": 1e-06, + "loss": 1.0879, + "mean_token_accuracy": 0.6767585277557373, + "num_tokens": 111228732.0, + "step": 4321 + }, + { + "epoch": 0.4746321106962442, + "grad_norm": 2.100757122039795, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7191748023033142, + "num_tokens": 111254580.0, + "step": 4322 + }, + { + "epoch": 0.47474192839885787, + "grad_norm": 2.232658863067627, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.6993527412414551, + "num_tokens": 111279442.0, + "step": 4323 + }, + { + "epoch": 0.4748517461014716, + "grad_norm": 2.243195056915283, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6930617094039917, + "num_tokens": 111304487.0, + "step": 4324 + }, + { + "epoch": 0.4749615638040852, + "grad_norm": 2.725137233734131, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.7037341594696045, + "num_tokens": 111329723.0, + "step": 4325 + }, + { + "epoch": 0.47507138150669886, + "grad_norm": 2.558347463607788, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7128934860229492, + "num_tokens": 111348659.0, + "step": 4326 + }, + { + "epoch": 0.47518119920931257, + "grad_norm": 2.0577759742736816, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6832666397094727, + "num_tokens": 111377529.0, + "step": 4327 + }, + { + "epoch": 0.4752910169119262, + "grad_norm": 1.9824514389038086, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6950604915618896, + "num_tokens": 111410903.0, + "step": 4328 + }, + { + "epoch": 0.47540083461453986, + "grad_norm": 2.425431251525879, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6932590007781982, + "num_tokens": 111434239.0, + "step": 4329 + }, + { + "epoch": 0.4755106523171535, + "grad_norm": 2.2336556911468506, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.687084436416626, + "num_tokens": 111459183.0, + "step": 4330 + }, + { + "epoch": 0.4756204700197672, + "grad_norm": 2.1909780502319336, + "learning_rate": 1e-06, + "loss": 1.081, + "mean_token_accuracy": 0.6787938475608826, + "num_tokens": 111485319.0, + "step": 4331 + }, + { + "epoch": 0.47573028772238085, + "grad_norm": 2.087756872177124, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7149850726127625, + "num_tokens": 111512268.0, + "step": 4332 + }, + { + "epoch": 0.4758401054249945, + "grad_norm": 2.4516937732696533, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7038544416427612, + "num_tokens": 111535238.0, + "step": 4333 + }, + { + "epoch": 0.47594992312760814, + "grad_norm": 2.1573326587677, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7189403772354126, + "num_tokens": 111559814.0, + "step": 4334 + }, + { + "epoch": 0.47605974083022184, + "grad_norm": 2.2900283336639404, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7068508267402649, + "num_tokens": 111581984.0, + "step": 4335 + }, + { + "epoch": 0.4761695585328355, + "grad_norm": 2.3423497676849365, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6866470575332642, + "num_tokens": 111607450.0, + "step": 4336 + }, + { + "epoch": 0.47627937623544914, + "grad_norm": 2.219341516494751, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7194325923919678, + "num_tokens": 111631163.0, + "step": 4337 + }, + { + "epoch": 0.47638919393806284, + "grad_norm": 2.0771236419677734, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6933398246765137, + "num_tokens": 111658868.0, + "step": 4338 + }, + { + "epoch": 0.4764990116406765, + "grad_norm": 2.677600860595703, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7157908082008362, + "num_tokens": 111678923.0, + "step": 4339 + }, + { + "epoch": 0.47660882934329013, + "grad_norm": 2.5079617500305176, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7070145010948181, + "num_tokens": 111700486.0, + "step": 4340 + }, + { + "epoch": 0.4767186470459038, + "grad_norm": 2.6900084018707275, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7122882008552551, + "num_tokens": 111718326.0, + "step": 4341 + }, + { + "epoch": 0.4768284647485175, + "grad_norm": 1.9504340887069702, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7190437316894531, + "num_tokens": 111748316.0, + "step": 4342 + }, + { + "epoch": 0.4769382824511311, + "grad_norm": 2.5179429054260254, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7251537442207336, + "num_tokens": 111769539.0, + "step": 4343 + }, + { + "epoch": 0.47704810015374477, + "grad_norm": 2.443816900253296, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6898762583732605, + "num_tokens": 111791374.0, + "step": 4344 + }, + { + "epoch": 0.47715791785635847, + "grad_norm": 1.971832513809204, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6859737038612366, + "num_tokens": 111823477.0, + "step": 4345 + }, + { + "epoch": 0.4772677355589721, + "grad_norm": 2.123420000076294, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7220264077186584, + "num_tokens": 111848975.0, + "step": 4346 + }, + { + "epoch": 0.47737755326158576, + "grad_norm": 2.280766487121582, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7079474329948425, + "num_tokens": 111874305.0, + "step": 4347 + }, + { + "epoch": 0.4774873709641994, + "grad_norm": 2.0865817070007324, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7060656547546387, + "num_tokens": 111901076.0, + "step": 4348 + }, + { + "epoch": 0.4775971886668131, + "grad_norm": 2.4480042457580566, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7138358950614929, + "num_tokens": 111920279.0, + "step": 4349 + }, + { + "epoch": 0.47770700636942676, + "grad_norm": 1.88174307346344, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6956081390380859, + "num_tokens": 111953217.0, + "step": 4350 + }, + { + "epoch": 0.4778168240720404, + "grad_norm": 2.3343710899353027, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7041569948196411, + "num_tokens": 111978097.0, + "step": 4351 + }, + { + "epoch": 0.47792664177465405, + "grad_norm": 2.1444764137268066, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6911031007766724, + "num_tokens": 112004066.0, + "step": 4352 + }, + { + "epoch": 0.47803645947726775, + "grad_norm": 2.0132668018341064, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6962544918060303, + "num_tokens": 112037321.0, + "step": 4353 + }, + { + "epoch": 0.4781462771798814, + "grad_norm": 2.486809730529785, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7040888071060181, + "num_tokens": 112058838.0, + "step": 4354 + }, + { + "epoch": 0.47825609488249504, + "grad_norm": 2.4616594314575195, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.71638423204422, + "num_tokens": 112078619.0, + "step": 4355 + }, + { + "epoch": 0.47836591258510874, + "grad_norm": 2.1912333965301514, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7220948934555054, + "num_tokens": 112102420.0, + "step": 4356 + }, + { + "epoch": 0.4784757302877224, + "grad_norm": 2.105827569961548, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7006369233131409, + "num_tokens": 112129401.0, + "step": 4357 + }, + { + "epoch": 0.47858554799033604, + "grad_norm": 2.04931902885437, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6968439817428589, + "num_tokens": 112160202.0, + "step": 4358 + }, + { + "epoch": 0.4786953656929497, + "grad_norm": 2.3125264644622803, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.703117847442627, + "num_tokens": 112184562.0, + "step": 4359 + }, + { + "epoch": 0.4788051833955634, + "grad_norm": 2.150343179702759, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6995538473129272, + "num_tokens": 112212945.0, + "step": 4360 + }, + { + "epoch": 0.47891500109817703, + "grad_norm": 2.6125690937042236, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7176284790039062, + "num_tokens": 112231420.0, + "step": 4361 + }, + { + "epoch": 0.4790248188007907, + "grad_norm": 2.4546966552734375, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7107006311416626, + "num_tokens": 112251529.0, + "step": 4362 + }, + { + "epoch": 0.4791346365034044, + "grad_norm": 2.3849644660949707, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6959314346313477, + "num_tokens": 112275321.0, + "step": 4363 + }, + { + "epoch": 0.479244454206018, + "grad_norm": 2.298893451690674, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7181730270385742, + "num_tokens": 112298745.0, + "step": 4364 + }, + { + "epoch": 0.47935427190863167, + "grad_norm": 2.1123764514923096, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7069023251533508, + "num_tokens": 112326420.0, + "step": 4365 + }, + { + "epoch": 0.4794640896112453, + "grad_norm": 2.1152098178863525, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6899218559265137, + "num_tokens": 112354377.0, + "step": 4366 + }, + { + "epoch": 0.479573907313859, + "grad_norm": 2.0659492015838623, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6844737529754639, + "num_tokens": 112383444.0, + "step": 4367 + }, + { + "epoch": 0.47968372501647266, + "grad_norm": 2.24727201461792, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.683117151260376, + "num_tokens": 112407680.0, + "step": 4368 + }, + { + "epoch": 0.4797935427190863, + "grad_norm": 2.0603530406951904, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6836910843849182, + "num_tokens": 112437234.0, + "step": 4369 + }, + { + "epoch": 0.47990336042169995, + "grad_norm": 2.1510801315307617, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6898093819618225, + "num_tokens": 112464767.0, + "step": 4370 + }, + { + "epoch": 0.48001317812431366, + "grad_norm": 2.188535690307617, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7282847166061401, + "num_tokens": 112488671.0, + "step": 4371 + }, + { + "epoch": 0.4801229958269273, + "grad_norm": 2.2360286712646484, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7028537392616272, + "num_tokens": 112514874.0, + "step": 4372 + }, + { + "epoch": 0.48023281352954095, + "grad_norm": 2.497755527496338, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6899157762527466, + "num_tokens": 112537030.0, + "step": 4373 + }, + { + "epoch": 0.48034263123215465, + "grad_norm": 2.1174521446228027, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6804750561714172, + "num_tokens": 112566214.0, + "step": 4374 + }, + { + "epoch": 0.4804524489347683, + "grad_norm": 2.0618021488189697, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7057396173477173, + "num_tokens": 112593680.0, + "step": 4375 + }, + { + "epoch": 0.48056226663738194, + "grad_norm": 2.0781617164611816, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6977782249450684, + "num_tokens": 112621156.0, + "step": 4376 + }, + { + "epoch": 0.4806720843399956, + "grad_norm": 2.37117338180542, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7148850560188293, + "num_tokens": 112642845.0, + "step": 4377 + }, + { + "epoch": 0.4807819020426093, + "grad_norm": 2.1626272201538086, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.723465085029602, + "num_tokens": 112668344.0, + "step": 4378 + }, + { + "epoch": 0.48089171974522293, + "grad_norm": 2.4233999252319336, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7110595703125, + "num_tokens": 112690867.0, + "step": 4379 + }, + { + "epoch": 0.4810015374478366, + "grad_norm": 2.4994730949401855, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6923074722290039, + "num_tokens": 112711735.0, + "step": 4380 + }, + { + "epoch": 0.4811113551504502, + "grad_norm": 2.178144693374634, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7052528262138367, + "num_tokens": 112737277.0, + "step": 4381 + }, + { + "epoch": 0.4812211728530639, + "grad_norm": 2.0706210136413574, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7017073035240173, + "num_tokens": 112763744.0, + "step": 4382 + }, + { + "epoch": 0.4813309905556776, + "grad_norm": 2.3243987560272217, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7089828252792358, + "num_tokens": 112786349.0, + "step": 4383 + }, + { + "epoch": 0.4814408082582912, + "grad_norm": 2.2749085426330566, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6965429782867432, + "num_tokens": 112812026.0, + "step": 4384 + }, + { + "epoch": 0.4815506259609049, + "grad_norm": 2.393900156021118, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6990426182746887, + "num_tokens": 112833690.0, + "step": 4385 + }, + { + "epoch": 0.48166044366351857, + "grad_norm": 2.148500442504883, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.734052300453186, + "num_tokens": 112858899.0, + "step": 4386 + }, + { + "epoch": 0.4817702613661322, + "grad_norm": 2.6280910968780518, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.707461953163147, + "num_tokens": 112878551.0, + "step": 4387 + }, + { + "epoch": 0.48188007906874586, + "grad_norm": 2.2228891849517822, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7164142727851868, + "num_tokens": 112901161.0, + "step": 4388 + }, + { + "epoch": 0.48198989677135956, + "grad_norm": 2.2292733192443848, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7113736867904663, + "num_tokens": 112924644.0, + "step": 4389 + }, + { + "epoch": 0.4820997144739732, + "grad_norm": 2.4643256664276123, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7007026672363281, + "num_tokens": 112946876.0, + "step": 4390 + }, + { + "epoch": 0.48220953217658685, + "grad_norm": 2.1862919330596924, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.713060736656189, + "num_tokens": 112972773.0, + "step": 4391 + }, + { + "epoch": 0.48231934987920055, + "grad_norm": 2.0899417400360107, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6948294043540955, + "num_tokens": 113001079.0, + "step": 4392 + }, + { + "epoch": 0.4824291675818142, + "grad_norm": 2.473674774169922, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7118823528289795, + "num_tokens": 113021235.0, + "step": 4393 + }, + { + "epoch": 0.48253898528442785, + "grad_norm": 2.787654161453247, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7047756910324097, + "num_tokens": 113038565.0, + "step": 4394 + }, + { + "epoch": 0.4826488029870415, + "grad_norm": 2.007056713104248, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7036321759223938, + "num_tokens": 113067176.0, + "step": 4395 + }, + { + "epoch": 0.4827586206896552, + "grad_norm": 2.0174572467803955, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.722385823726654, + "num_tokens": 113093162.0, + "step": 4396 + }, + { + "epoch": 0.48286843839226884, + "grad_norm": 2.4688220024108887, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7092171907424927, + "num_tokens": 113114983.0, + "step": 4397 + }, + { + "epoch": 0.4829782560948825, + "grad_norm": 2.506385564804077, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6923393607139587, + "num_tokens": 113136225.0, + "step": 4398 + }, + { + "epoch": 0.48308807379749613, + "grad_norm": 2.518883228302002, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.7053619027137756, + "num_tokens": 113156902.0, + "step": 4399 + }, + { + "epoch": 0.48319789150010983, + "grad_norm": 2.282254457473755, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7284130454063416, + "num_tokens": 113179291.0, + "step": 4400 + }, + { + "epoch": 0.4833077092027235, + "grad_norm": 2.098947763442993, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7090242505073547, + "num_tokens": 113206459.0, + "step": 4401 + }, + { + "epoch": 0.4834175269053371, + "grad_norm": 2.1594555377960205, + "learning_rate": 1e-06, + "loss": 1.0752, + "mean_token_accuracy": 0.6728915572166443, + "num_tokens": 113232184.0, + "step": 4402 + }, + { + "epoch": 0.4835273446079508, + "grad_norm": 2.155226945877075, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6777046918869019, + "num_tokens": 113257980.0, + "step": 4403 + }, + { + "epoch": 0.4836371623105645, + "grad_norm": 2.0344998836517334, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7038010358810425, + "num_tokens": 113285713.0, + "step": 4404 + }, + { + "epoch": 0.4837469800131781, + "grad_norm": 2.0278072357177734, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7174049615859985, + "num_tokens": 113313784.0, + "step": 4405 + }, + { + "epoch": 0.48385679771579176, + "grad_norm": 2.081880807876587, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7111939191818237, + "num_tokens": 113342882.0, + "step": 4406 + }, + { + "epoch": 0.48396661541840547, + "grad_norm": 2.244673013687134, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6882971525192261, + "num_tokens": 113370727.0, + "step": 4407 + }, + { + "epoch": 0.4840764331210191, + "grad_norm": 2.218238115310669, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.7009506225585938, + "num_tokens": 113395569.0, + "step": 4408 + }, + { + "epoch": 0.48418625082363276, + "grad_norm": 2.348823070526123, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6948670148849487, + "num_tokens": 113418109.0, + "step": 4409 + }, + { + "epoch": 0.4842960685262464, + "grad_norm": 2.1464486122131348, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6997777223587036, + "num_tokens": 113447149.0, + "step": 4410 + }, + { + "epoch": 0.4844058862288601, + "grad_norm": 2.4152936935424805, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.721148669719696, + "num_tokens": 113469948.0, + "step": 4411 + }, + { + "epoch": 0.48451570393147375, + "grad_norm": 2.331371307373047, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7159896492958069, + "num_tokens": 113493856.0, + "step": 4412 + }, + { + "epoch": 0.4846255216340874, + "grad_norm": 2.352212905883789, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7090951204299927, + "num_tokens": 113515261.0, + "step": 4413 + }, + { + "epoch": 0.4847353393367011, + "grad_norm": 2.36617374420166, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7103087902069092, + "num_tokens": 113538642.0, + "step": 4414 + }, + { + "epoch": 0.48484515703931474, + "grad_norm": 2.2444326877593994, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6992135047912598, + "num_tokens": 113564440.0, + "step": 4415 + }, + { + "epoch": 0.4849549747419284, + "grad_norm": 2.6751925945281982, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7188675403594971, + "num_tokens": 113582229.0, + "step": 4416 + }, + { + "epoch": 0.48506479244454204, + "grad_norm": 2.387087345123291, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7094843983650208, + "num_tokens": 113603802.0, + "step": 4417 + }, + { + "epoch": 0.48517461014715574, + "grad_norm": 2.178788185119629, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6992934942245483, + "num_tokens": 113628812.0, + "step": 4418 + }, + { + "epoch": 0.4852844278497694, + "grad_norm": 2.1920037269592285, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7231171131134033, + "num_tokens": 113653103.0, + "step": 4419 + }, + { + "epoch": 0.48539424555238303, + "grad_norm": 2.114600419998169, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6933526992797852, + "num_tokens": 113679337.0, + "step": 4420 + }, + { + "epoch": 0.48550406325499673, + "grad_norm": 2.377002477645874, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6956688165664673, + "num_tokens": 113701446.0, + "step": 4421 + }, + { + "epoch": 0.4856138809576104, + "grad_norm": 2.50148606300354, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6955850124359131, + "num_tokens": 113723703.0, + "step": 4422 + }, + { + "epoch": 0.485723698660224, + "grad_norm": 2.0439393520355225, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6870748400688171, + "num_tokens": 113752974.0, + "step": 4423 + }, + { + "epoch": 0.48583351636283767, + "grad_norm": 2.361250877380371, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7392861843109131, + "num_tokens": 113775319.0, + "step": 4424 + }, + { + "epoch": 0.48594333406545137, + "grad_norm": 2.256436824798584, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7007668018341064, + "num_tokens": 113798051.0, + "step": 4425 + }, + { + "epoch": 0.486053151768065, + "grad_norm": 2.5663301944732666, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7112998962402344, + "num_tokens": 113817181.0, + "step": 4426 + }, + { + "epoch": 0.48616296947067866, + "grad_norm": 1.8915133476257324, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6865893602371216, + "num_tokens": 113848539.0, + "step": 4427 + }, + { + "epoch": 0.4862727871732923, + "grad_norm": 2.09706974029541, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7021694183349609, + "num_tokens": 113877054.0, + "step": 4428 + }, + { + "epoch": 0.486382604875906, + "grad_norm": 2.138453245162964, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7030972242355347, + "num_tokens": 113903360.0, + "step": 4429 + }, + { + "epoch": 0.48649242257851966, + "grad_norm": 2.228130578994751, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6978169679641724, + "num_tokens": 113929499.0, + "step": 4430 + }, + { + "epoch": 0.4866022402811333, + "grad_norm": 2.0042006969451904, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7028989791870117, + "num_tokens": 113958736.0, + "step": 4431 + }, + { + "epoch": 0.486712057983747, + "grad_norm": 1.9762184619903564, + "learning_rate": 1e-06, + "loss": 1.1078, + "mean_token_accuracy": 0.6748388409614563, + "num_tokens": 113992316.0, + "step": 4432 + }, + { + "epoch": 0.48682187568636065, + "grad_norm": 2.2468972206115723, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6812707185745239, + "num_tokens": 114016809.0, + "step": 4433 + }, + { + "epoch": 0.4869316933889743, + "grad_norm": 2.0594642162323, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7063767910003662, + "num_tokens": 114045141.0, + "step": 4434 + }, + { + "epoch": 0.48704151109158794, + "grad_norm": 2.396573305130005, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7048946619033813, + "num_tokens": 114069390.0, + "step": 4435 + }, + { + "epoch": 0.48715132879420164, + "grad_norm": 1.876959204673767, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.7000865936279297, + "num_tokens": 114101690.0, + "step": 4436 + }, + { + "epoch": 0.4872611464968153, + "grad_norm": 2.288085460662842, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6898123025894165, + "num_tokens": 114124270.0, + "step": 4437 + }, + { + "epoch": 0.48737096419942894, + "grad_norm": 1.9833160638809204, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7009159922599792, + "num_tokens": 114156556.0, + "step": 4438 + }, + { + "epoch": 0.48748078190204264, + "grad_norm": 2.1306445598602295, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6909009218215942, + "num_tokens": 114185060.0, + "step": 4439 + }, + { + "epoch": 0.4875905996046563, + "grad_norm": 2.2098164558410645, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.701028048992157, + "num_tokens": 114210510.0, + "step": 4440 + }, + { + "epoch": 0.48770041730726993, + "grad_norm": 2.1147043704986572, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7331465482711792, + "num_tokens": 114235495.0, + "step": 4441 + }, + { + "epoch": 0.4878102350098836, + "grad_norm": 2.2380359172821045, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7144598960876465, + "num_tokens": 114259833.0, + "step": 4442 + }, + { + "epoch": 0.4879200527124973, + "grad_norm": 2.1215410232543945, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6986345052719116, + "num_tokens": 114287154.0, + "step": 4443 + }, + { + "epoch": 0.4880298704151109, + "grad_norm": 2.466214656829834, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7281641364097595, + "num_tokens": 114305846.0, + "step": 4444 + }, + { + "epoch": 0.48813968811772457, + "grad_norm": 2.1388731002807617, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.695172905921936, + "num_tokens": 114331967.0, + "step": 4445 + }, + { + "epoch": 0.4882495058203382, + "grad_norm": 2.4042446613311768, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7019756436347961, + "num_tokens": 114354249.0, + "step": 4446 + }, + { + "epoch": 0.4883593235229519, + "grad_norm": 2.381514310836792, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.706925094127655, + "num_tokens": 114377198.0, + "step": 4447 + }, + { + "epoch": 0.48846914122556556, + "grad_norm": 2.4447124004364014, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7207511067390442, + "num_tokens": 114397002.0, + "step": 4448 + }, + { + "epoch": 0.4885789589281792, + "grad_norm": 2.1179161071777344, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7000384330749512, + "num_tokens": 114423733.0, + "step": 4449 + }, + { + "epoch": 0.4886887766307929, + "grad_norm": 2.36293363571167, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7208021283149719, + "num_tokens": 114445367.0, + "step": 4450 + }, + { + "epoch": 0.48879859433340656, + "grad_norm": 2.1062963008880615, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6961908340454102, + "num_tokens": 114474159.0, + "step": 4451 + }, + { + "epoch": 0.4889084120360202, + "grad_norm": 2.0731725692749023, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6813037991523743, + "num_tokens": 114502273.0, + "step": 4452 + }, + { + "epoch": 0.48901822973863385, + "grad_norm": 1.967606782913208, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7121783494949341, + "num_tokens": 114531872.0, + "step": 4453 + }, + { + "epoch": 0.48912804744124755, + "grad_norm": 2.166825294494629, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.721489429473877, + "num_tokens": 114555644.0, + "step": 4454 + }, + { + "epoch": 0.4892378651438612, + "grad_norm": 2.1452717781066895, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6841420531272888, + "num_tokens": 114583887.0, + "step": 4455 + }, + { + "epoch": 0.48934768284647484, + "grad_norm": 2.4254753589630127, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6826575398445129, + "num_tokens": 114607306.0, + "step": 4456 + }, + { + "epoch": 0.4894575005490885, + "grad_norm": 2.1728224754333496, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6829453706741333, + "num_tokens": 114633020.0, + "step": 4457 + }, + { + "epoch": 0.4895673182517022, + "grad_norm": 2.1685776710510254, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7152350544929504, + "num_tokens": 114660943.0, + "step": 4458 + }, + { + "epoch": 0.48967713595431583, + "grad_norm": 2.406511068344116, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6844384670257568, + "num_tokens": 114682385.0, + "step": 4459 + }, + { + "epoch": 0.4897869536569295, + "grad_norm": 2.478809356689453, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7175836563110352, + "num_tokens": 114702737.0, + "step": 4460 + }, + { + "epoch": 0.4898967713595432, + "grad_norm": 2.3045461177825928, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6870883107185364, + "num_tokens": 114727842.0, + "step": 4461 + }, + { + "epoch": 0.4900065890621568, + "grad_norm": 2.0139453411102295, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7006455659866333, + "num_tokens": 114756932.0, + "step": 4462 + }, + { + "epoch": 0.4901164067647705, + "grad_norm": 2.062088966369629, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7087142467498779, + "num_tokens": 114784371.0, + "step": 4463 + }, + { + "epoch": 0.4902262244673841, + "grad_norm": 2.3912203311920166, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7109173536300659, + "num_tokens": 114806722.0, + "step": 4464 + }, + { + "epoch": 0.4903360421699978, + "grad_norm": 2.8842430114746094, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7225056886672974, + "num_tokens": 114822981.0, + "step": 4465 + }, + { + "epoch": 0.49044585987261147, + "grad_norm": 2.1128487586975098, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7192167639732361, + "num_tokens": 114847461.0, + "step": 4466 + }, + { + "epoch": 0.4905556775752251, + "grad_norm": 2.099363327026367, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7019298076629639, + "num_tokens": 114875545.0, + "step": 4467 + }, + { + "epoch": 0.4906654952778388, + "grad_norm": 1.7862143516540527, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6947580575942993, + "num_tokens": 114912105.0, + "step": 4468 + }, + { + "epoch": 0.49077531298045246, + "grad_norm": 2.14864182472229, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6970765590667725, + "num_tokens": 114939223.0, + "step": 4469 + }, + { + "epoch": 0.4908851306830661, + "grad_norm": 1.7811164855957031, + "learning_rate": 1e-06, + "loss": 1.0709, + "mean_token_accuracy": 0.6757047772407532, + "num_tokens": 114979144.0, + "step": 4470 + }, + { + "epoch": 0.49099494838567975, + "grad_norm": 2.302572011947632, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6954169273376465, + "num_tokens": 115002631.0, + "step": 4471 + }, + { + "epoch": 0.49110476608829345, + "grad_norm": 2.026442050933838, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6876695156097412, + "num_tokens": 115031985.0, + "step": 4472 + }, + { + "epoch": 0.4912145837909071, + "grad_norm": 1.9960793256759644, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6943652629852295, + "num_tokens": 115060450.0, + "step": 4473 + }, + { + "epoch": 0.49132440149352075, + "grad_norm": 2.782552719116211, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7190227508544922, + "num_tokens": 115077847.0, + "step": 4474 + }, + { + "epoch": 0.4914342191961344, + "grad_norm": 1.977810025215149, + "learning_rate": 1e-06, + "loss": 1.0817, + "mean_token_accuracy": 0.6695384979248047, + "num_tokens": 115109149.0, + "step": 4475 + }, + { + "epoch": 0.4915440368987481, + "grad_norm": 2.3502132892608643, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6947296261787415, + "num_tokens": 115130691.0, + "step": 4476 + }, + { + "epoch": 0.49165385460136174, + "grad_norm": 2.1205995082855225, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7117639780044556, + "num_tokens": 115157526.0, + "step": 4477 + }, + { + "epoch": 0.4917636723039754, + "grad_norm": 1.9978413581848145, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.690200924873352, + "num_tokens": 115189420.0, + "step": 4478 + }, + { + "epoch": 0.4918734900065891, + "grad_norm": 1.9902994632720947, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6983442306518555, + "num_tokens": 115219131.0, + "step": 4479 + }, + { + "epoch": 0.49198330770920273, + "grad_norm": 2.2789340019226074, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.709656834602356, + "num_tokens": 115243540.0, + "step": 4480 + }, + { + "epoch": 0.4920931254118164, + "grad_norm": 2.0490567684173584, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6977714896202087, + "num_tokens": 115271111.0, + "step": 4481 + }, + { + "epoch": 0.49220294311443, + "grad_norm": 2.06877064704895, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7210894823074341, + "num_tokens": 115300072.0, + "step": 4482 + }, + { + "epoch": 0.4923127608170437, + "grad_norm": 2.021449327468872, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.697547197341919, + "num_tokens": 115329831.0, + "step": 4483 + }, + { + "epoch": 0.4924225785196574, + "grad_norm": 2.2075796127319336, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6902323961257935, + "num_tokens": 115354644.0, + "step": 4484 + }, + { + "epoch": 0.492532396222271, + "grad_norm": 2.6084258556365967, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7117977142333984, + "num_tokens": 115374917.0, + "step": 4485 + }, + { + "epoch": 0.49264221392488466, + "grad_norm": 2.23618221282959, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6958339214324951, + "num_tokens": 115401753.0, + "step": 4486 + }, + { + "epoch": 0.49275203162749837, + "grad_norm": 2.037229537963867, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7062331438064575, + "num_tokens": 115431199.0, + "step": 4487 + }, + { + "epoch": 0.492861849330112, + "grad_norm": 2.3345794677734375, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7057106494903564, + "num_tokens": 115455345.0, + "step": 4488 + }, + { + "epoch": 0.49297166703272566, + "grad_norm": 2.2030820846557617, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.711122989654541, + "num_tokens": 115479973.0, + "step": 4489 + }, + { + "epoch": 0.49308148473533936, + "grad_norm": 2.245666742324829, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7035627365112305, + "num_tokens": 115504466.0, + "step": 4490 + }, + { + "epoch": 0.493191302437953, + "grad_norm": 2.2027220726013184, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7000547647476196, + "num_tokens": 115529888.0, + "step": 4491 + }, + { + "epoch": 0.49330112014056665, + "grad_norm": 2.065938949584961, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7018410563468933, + "num_tokens": 115557804.0, + "step": 4492 + }, + { + "epoch": 0.4934109378431803, + "grad_norm": 2.203298807144165, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.697468638420105, + "num_tokens": 115583760.0, + "step": 4493 + }, + { + "epoch": 0.493520755545794, + "grad_norm": 2.327672004699707, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7085705995559692, + "num_tokens": 115607224.0, + "step": 4494 + }, + { + "epoch": 0.49363057324840764, + "grad_norm": 2.2414896488189697, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6803122162818909, + "num_tokens": 115631659.0, + "step": 4495 + }, + { + "epoch": 0.4937403909510213, + "grad_norm": 1.9134291410446167, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6803021430969238, + "num_tokens": 115666821.0, + "step": 4496 + }, + { + "epoch": 0.493850208653635, + "grad_norm": 2.3799593448638916, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.7007792592048645, + "num_tokens": 115689961.0, + "step": 4497 + }, + { + "epoch": 0.49396002635624864, + "grad_norm": 2.423042058944702, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7241740822792053, + "num_tokens": 115711423.0, + "step": 4498 + }, + { + "epoch": 0.4940698440588623, + "grad_norm": 2.2878000736236572, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7103109955787659, + "num_tokens": 115735321.0, + "step": 4499 + }, + { + "epoch": 0.49417966176147593, + "grad_norm": 2.1987972259521484, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6878011226654053, + "num_tokens": 115762246.0, + "step": 4500 + }, + { + "epoch": 0.49428947946408963, + "grad_norm": 2.2666330337524414, + "learning_rate": 1e-06, + "loss": 1.0692, + "mean_token_accuracy": 0.6973093152046204, + "num_tokens": 115785636.0, + "step": 4501 + }, + { + "epoch": 0.4943992971667033, + "grad_norm": 2.053959608078003, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7001407146453857, + "num_tokens": 115814841.0, + "step": 4502 + }, + { + "epoch": 0.4945091148693169, + "grad_norm": 2.1829521656036377, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7084841728210449, + "num_tokens": 115840233.0, + "step": 4503 + }, + { + "epoch": 0.49461893257193057, + "grad_norm": 2.15177321434021, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.713081955909729, + "num_tokens": 115865182.0, + "step": 4504 + }, + { + "epoch": 0.49472875027454427, + "grad_norm": 2.272003412246704, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7354390025138855, + "num_tokens": 115887785.0, + "step": 4505 + }, + { + "epoch": 0.4948385679771579, + "grad_norm": 2.208724021911621, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7116985321044922, + "num_tokens": 115910097.0, + "step": 4506 + }, + { + "epoch": 0.49494838567977156, + "grad_norm": 2.2369613647460938, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6878879070281982, + "num_tokens": 115935108.0, + "step": 4507 + }, + { + "epoch": 0.49505820338238526, + "grad_norm": 2.2566685676574707, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.708973228931427, + "num_tokens": 115958592.0, + "step": 4508 + }, + { + "epoch": 0.4951680210849989, + "grad_norm": 2.0303995609283447, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7199554443359375, + "num_tokens": 115986729.0, + "step": 4509 + }, + { + "epoch": 0.49527783878761256, + "grad_norm": 2.2306952476501465, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.6814236640930176, + "num_tokens": 116014416.0, + "step": 4510 + }, + { + "epoch": 0.4953876564902262, + "grad_norm": 2.10727596282959, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7088647484779358, + "num_tokens": 116043101.0, + "step": 4511 + }, + { + "epoch": 0.4954974741928399, + "grad_norm": 2.419335126876831, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6962987780570984, + "num_tokens": 116066648.0, + "step": 4512 + }, + { + "epoch": 0.49560729189545355, + "grad_norm": 2.0872230529785156, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6791390180587769, + "num_tokens": 116095069.0, + "step": 4513 + }, + { + "epoch": 0.4957171095980672, + "grad_norm": 2.229520797729492, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7468787431716919, + "num_tokens": 116116795.0, + "step": 4514 + }, + { + "epoch": 0.4958269273006809, + "grad_norm": 2.456308364868164, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7140390872955322, + "num_tokens": 116136897.0, + "step": 4515 + }, + { + "epoch": 0.49593674500329454, + "grad_norm": 2.087423086166382, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.704649806022644, + "num_tokens": 116165415.0, + "step": 4516 + }, + { + "epoch": 0.4960465627059082, + "grad_norm": 2.33317494392395, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7008107900619507, + "num_tokens": 116186968.0, + "step": 4517 + }, + { + "epoch": 0.49615638040852184, + "grad_norm": 2.2714927196502686, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7041922807693481, + "num_tokens": 116211387.0, + "step": 4518 + }, + { + "epoch": 0.49626619811113554, + "grad_norm": 2.3209590911865234, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.702599823474884, + "num_tokens": 116233429.0, + "step": 4519 + }, + { + "epoch": 0.4963760158137492, + "grad_norm": 2.4524214267730713, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7226399183273315, + "num_tokens": 116255166.0, + "step": 4520 + }, + { + "epoch": 0.49648583351636283, + "grad_norm": 2.046679973602295, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7056626081466675, + "num_tokens": 116284210.0, + "step": 4521 + }, + { + "epoch": 0.4965956512189765, + "grad_norm": 2.114680767059326, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6926838755607605, + "num_tokens": 116314463.0, + "step": 4522 + }, + { + "epoch": 0.4967054689215902, + "grad_norm": 2.311267137527466, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.7001931667327881, + "num_tokens": 116338406.0, + "step": 4523 + }, + { + "epoch": 0.4968152866242038, + "grad_norm": 2.0593738555908203, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7173432111740112, + "num_tokens": 116367276.0, + "step": 4524 + }, + { + "epoch": 0.49692510432681747, + "grad_norm": 2.407085657119751, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7184813022613525, + "num_tokens": 116388276.0, + "step": 4525 + }, + { + "epoch": 0.49703492202943117, + "grad_norm": 2.521007776260376, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7164179682731628, + "num_tokens": 116410417.0, + "step": 4526 + }, + { + "epoch": 0.4971447397320448, + "grad_norm": 2.193300247192383, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7035750150680542, + "num_tokens": 116435316.0, + "step": 4527 + }, + { + "epoch": 0.49725455743465846, + "grad_norm": 2.4935848712921143, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7008756399154663, + "num_tokens": 116455519.0, + "step": 4528 + }, + { + "epoch": 0.4973643751372721, + "grad_norm": 2.225736379623413, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7086459398269653, + "num_tokens": 116479120.0, + "step": 4529 + }, + { + "epoch": 0.4974741928398858, + "grad_norm": 2.335904121398926, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7021437287330627, + "num_tokens": 116502669.0, + "step": 4530 + }, + { + "epoch": 0.49758401054249946, + "grad_norm": 2.223649740219116, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7073859572410583, + "num_tokens": 116526676.0, + "step": 4531 + }, + { + "epoch": 0.4976938282451131, + "grad_norm": 2.1994829177856445, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.708646297454834, + "num_tokens": 116552702.0, + "step": 4532 + }, + { + "epoch": 0.49780364594772675, + "grad_norm": 2.118964195251465, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7031866312026978, + "num_tokens": 116578459.0, + "step": 4533 + }, + { + "epoch": 0.49791346365034045, + "grad_norm": 2.285334348678589, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7184983491897583, + "num_tokens": 116601397.0, + "step": 4534 + }, + { + "epoch": 0.4980232813529541, + "grad_norm": 2.297874927520752, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6963986158370972, + "num_tokens": 116623472.0, + "step": 4535 + }, + { + "epoch": 0.49813309905556774, + "grad_norm": 2.109882116317749, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7236758470535278, + "num_tokens": 116650615.0, + "step": 4536 + }, + { + "epoch": 0.49824291675818144, + "grad_norm": 2.328139066696167, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7078226208686829, + "num_tokens": 116674611.0, + "step": 4537 + }, + { + "epoch": 0.4983527344607951, + "grad_norm": 2.1932578086853027, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.70168536901474, + "num_tokens": 116699658.0, + "step": 4538 + }, + { + "epoch": 0.49846255216340873, + "grad_norm": 2.118360996246338, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.715258002281189, + "num_tokens": 116727508.0, + "step": 4539 + }, + { + "epoch": 0.4985723698660224, + "grad_norm": 2.247804880142212, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6875702142715454, + "num_tokens": 116753110.0, + "step": 4540 + }, + { + "epoch": 0.4986821875686361, + "grad_norm": 2.189324378967285, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7065117359161377, + "num_tokens": 116777952.0, + "step": 4541 + }, + { + "epoch": 0.4987920052712497, + "grad_norm": 2.5613348484039307, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6994693279266357, + "num_tokens": 116797479.0, + "step": 4542 + }, + { + "epoch": 0.4989018229738634, + "grad_norm": 2.309708595275879, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7050525546073914, + "num_tokens": 116820195.0, + "step": 4543 + }, + { + "epoch": 0.4990116406764771, + "grad_norm": 2.023951292037964, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7129331827163696, + "num_tokens": 116846781.0, + "step": 4544 + }, + { + "epoch": 0.4991214583790907, + "grad_norm": 2.21524715423584, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6870144009590149, + "num_tokens": 116872889.0, + "step": 4545 + }, + { + "epoch": 0.49923127608170437, + "grad_norm": 2.1419806480407715, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6970856189727783, + "num_tokens": 116900107.0, + "step": 4546 + }, + { + "epoch": 0.499341093784318, + "grad_norm": 2.2537240982055664, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6871142387390137, + "num_tokens": 116927108.0, + "step": 4547 + }, + { + "epoch": 0.4994509114869317, + "grad_norm": 2.0746145248413086, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7027808427810669, + "num_tokens": 116954912.0, + "step": 4548 + }, + { + "epoch": 0.49956072918954536, + "grad_norm": 2.174344778060913, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6922240257263184, + "num_tokens": 116979266.0, + "step": 4549 + }, + { + "epoch": 0.499670546892159, + "grad_norm": 1.9903110265731812, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7050858736038208, + "num_tokens": 117007249.0, + "step": 4550 + }, + { + "epoch": 0.49978036459477265, + "grad_norm": 2.0016274452209473, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6817054748535156, + "num_tokens": 117036911.0, + "step": 4551 + }, + { + "epoch": 0.49989018229738635, + "grad_norm": 2.2950682640075684, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7065621018409729, + "num_tokens": 117059571.0, + "step": 4552 + }, + { + "epoch": 0.5, + "grad_norm": 2.068274736404419, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7072173357009888, + "num_tokens": 117086712.0, + "step": 4553 + }, + { + "epoch": 0.5001098177026136, + "grad_norm": 2.17350697517395, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.709439218044281, + "num_tokens": 117111460.0, + "step": 4554 + }, + { + "epoch": 0.5002196354052273, + "grad_norm": 2.100801706314087, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7114384770393372, + "num_tokens": 117137724.0, + "step": 4555 + }, + { + "epoch": 0.5003294531078409, + "grad_norm": 2.1225173473358154, + "learning_rate": 1e-06, + "loss": 1.0656, + "mean_token_accuracy": 0.6768450736999512, + "num_tokens": 117165101.0, + "step": 4556 + }, + { + "epoch": 0.5004392708104547, + "grad_norm": 2.3127939701080322, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6897966861724854, + "num_tokens": 117189553.0, + "step": 4557 + }, + { + "epoch": 0.5005490885130683, + "grad_norm": 1.9474589824676514, + "learning_rate": 1e-06, + "loss": 1.1062, + "mean_token_accuracy": 0.669274628162384, + "num_tokens": 117220789.0, + "step": 4558 + }, + { + "epoch": 0.500658906215682, + "grad_norm": 2.1732304096221924, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7031867504119873, + "num_tokens": 117246798.0, + "step": 4559 + }, + { + "epoch": 0.5007687239182956, + "grad_norm": 2.693328380584717, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7324314117431641, + "num_tokens": 117263070.0, + "step": 4560 + }, + { + "epoch": 0.5008785416209093, + "grad_norm": 2.2903592586517334, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6927302479743958, + "num_tokens": 117285998.0, + "step": 4561 + }, + { + "epoch": 0.5009883593235229, + "grad_norm": 2.475264549255371, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6975299119949341, + "num_tokens": 117305529.0, + "step": 4562 + }, + { + "epoch": 0.5010981770261366, + "grad_norm": 2.1126818656921387, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.713842511177063, + "num_tokens": 117330363.0, + "step": 4563 + }, + { + "epoch": 0.5012079947287503, + "grad_norm": 2.4013710021972656, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7073609828948975, + "num_tokens": 117351479.0, + "step": 4564 + }, + { + "epoch": 0.501317812431364, + "grad_norm": 2.181908130645752, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6975411176681519, + "num_tokens": 117376601.0, + "step": 4565 + }, + { + "epoch": 0.5014276301339776, + "grad_norm": 2.5750813484191895, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6864434480667114, + "num_tokens": 117396327.0, + "step": 4566 + }, + { + "epoch": 0.5015374478365913, + "grad_norm": 2.451775312423706, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6920942068099976, + "num_tokens": 117419258.0, + "step": 4567 + }, + { + "epoch": 0.5016472655392049, + "grad_norm": 2.289590358734131, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.705656886100769, + "num_tokens": 117444024.0, + "step": 4568 + }, + { + "epoch": 0.5017570832418186, + "grad_norm": 2.2621958255767822, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7182977795600891, + "num_tokens": 117467816.0, + "step": 4569 + }, + { + "epoch": 0.5018669009444322, + "grad_norm": 2.328352451324463, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6880806088447571, + "num_tokens": 117492102.0, + "step": 4570 + }, + { + "epoch": 0.5019767186470458, + "grad_norm": 2.2145633697509766, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.720272421836853, + "num_tokens": 117517105.0, + "step": 4571 + }, + { + "epoch": 0.5020865363496596, + "grad_norm": 2.1654200553894043, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6850133538246155, + "num_tokens": 117543888.0, + "step": 4572 + }, + { + "epoch": 0.5021963540522733, + "grad_norm": 2.1860127449035645, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6911953687667847, + "num_tokens": 117569296.0, + "step": 4573 + }, + { + "epoch": 0.5023061717548869, + "grad_norm": 2.380136489868164, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6923787593841553, + "num_tokens": 117591853.0, + "step": 4574 + }, + { + "epoch": 0.5024159894575005, + "grad_norm": 2.196518898010254, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6977379322052002, + "num_tokens": 117619038.0, + "step": 4575 + }, + { + "epoch": 0.5025258071601142, + "grad_norm": 2.4402358531951904, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7075629234313965, + "num_tokens": 117642094.0, + "step": 4576 + }, + { + "epoch": 0.5026356248627278, + "grad_norm": 2.231236457824707, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7031897306442261, + "num_tokens": 117665354.0, + "step": 4577 + }, + { + "epoch": 0.5027454425653415, + "grad_norm": 2.5297906398773193, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7291054725646973, + "num_tokens": 117685509.0, + "step": 4578 + }, + { + "epoch": 0.5028552602679552, + "grad_norm": 2.525881290435791, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7130625247955322, + "num_tokens": 117707161.0, + "step": 4579 + }, + { + "epoch": 0.5029650779705689, + "grad_norm": 2.2650034427642822, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.698219895362854, + "num_tokens": 117730250.0, + "step": 4580 + }, + { + "epoch": 0.5030748956731825, + "grad_norm": 2.0059900283813477, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7126609086990356, + "num_tokens": 117756983.0, + "step": 4581 + }, + { + "epoch": 0.5031847133757962, + "grad_norm": 2.2873880863189697, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7101726531982422, + "num_tokens": 117780731.0, + "step": 4582 + }, + { + "epoch": 0.5032945310784098, + "grad_norm": 2.237501859664917, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6919719576835632, + "num_tokens": 117807371.0, + "step": 4583 + }, + { + "epoch": 0.5034043487810235, + "grad_norm": 2.3614490032196045, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.710605263710022, + "num_tokens": 117829838.0, + "step": 4584 + }, + { + "epoch": 0.5035141664836371, + "grad_norm": 1.8828504085540771, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7141124606132507, + "num_tokens": 117861532.0, + "step": 4585 + }, + { + "epoch": 0.5036239841862509, + "grad_norm": 1.9780908823013306, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.6800349950790405, + "num_tokens": 117893495.0, + "step": 4586 + }, + { + "epoch": 0.5037338018888645, + "grad_norm": 1.8787140846252441, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6962205171585083, + "num_tokens": 117924693.0, + "step": 4587 + }, + { + "epoch": 0.5038436195914782, + "grad_norm": 2.211299419403076, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.703608512878418, + "num_tokens": 117949603.0, + "step": 4588 + }, + { + "epoch": 0.5039534372940918, + "grad_norm": 2.154953718185425, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.6950870752334595, + "num_tokens": 117975980.0, + "step": 4589 + }, + { + "epoch": 0.5040632549967055, + "grad_norm": 1.8342634439468384, + "learning_rate": 1e-06, + "loss": 1.1403, + "mean_token_accuracy": 0.6730165481567383, + "num_tokens": 118013799.0, + "step": 4590 + }, + { + "epoch": 0.5041730726993191, + "grad_norm": 2.015463352203369, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7022411823272705, + "num_tokens": 118044339.0, + "step": 4591 + }, + { + "epoch": 0.5042828904019327, + "grad_norm": 2.190992593765259, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7134606838226318, + "num_tokens": 118069737.0, + "step": 4592 + }, + { + "epoch": 0.5043927081045465, + "grad_norm": 2.1125330924987793, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7023409008979797, + "num_tokens": 118097639.0, + "step": 4593 + }, + { + "epoch": 0.5045025258071602, + "grad_norm": 2.490051507949829, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7120401859283447, + "num_tokens": 118120344.0, + "step": 4594 + }, + { + "epoch": 0.5046123435097738, + "grad_norm": 2.240267276763916, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6938736438751221, + "num_tokens": 118147343.0, + "step": 4595 + }, + { + "epoch": 0.5047221612123874, + "grad_norm": 2.1409103870391846, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7286362648010254, + "num_tokens": 118173669.0, + "step": 4596 + }, + { + "epoch": 0.5048319789150011, + "grad_norm": 2.199277400970459, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.7041290998458862, + "num_tokens": 118199268.0, + "step": 4597 + }, + { + "epoch": 0.5049417966176147, + "grad_norm": 2.273604393005371, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7021395564079285, + "num_tokens": 118222118.0, + "step": 4598 + }, + { + "epoch": 0.5050516143202284, + "grad_norm": 2.2161128520965576, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6982367634773254, + "num_tokens": 118248083.0, + "step": 4599 + }, + { + "epoch": 0.505161432022842, + "grad_norm": 2.134664535522461, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6984821557998657, + "num_tokens": 118275717.0, + "step": 4600 + }, + { + "epoch": 0.5052712497254558, + "grad_norm": 2.3474740982055664, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6943373680114746, + "num_tokens": 118298815.0, + "step": 4601 + }, + { + "epoch": 0.5053810674280694, + "grad_norm": 2.098308563232422, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6914705634117126, + "num_tokens": 118325462.0, + "step": 4602 + }, + { + "epoch": 0.5054908851306831, + "grad_norm": 2.166782855987549, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6907438635826111, + "num_tokens": 118351868.0, + "step": 4603 + }, + { + "epoch": 0.5056007028332967, + "grad_norm": 2.273130416870117, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7157315015792847, + "num_tokens": 118375038.0, + "step": 4604 + }, + { + "epoch": 0.5057105205359104, + "grad_norm": 2.2438454627990723, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7182265520095825, + "num_tokens": 118400304.0, + "step": 4605 + }, + { + "epoch": 0.505820338238524, + "grad_norm": 2.1758205890655518, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.690129280090332, + "num_tokens": 118427605.0, + "step": 4606 + }, + { + "epoch": 0.5059301559411377, + "grad_norm": 2.102113962173462, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7089011073112488, + "num_tokens": 118454518.0, + "step": 4607 + }, + { + "epoch": 0.5060399736437514, + "grad_norm": 2.1079797744750977, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6955687403678894, + "num_tokens": 118480993.0, + "step": 4608 + }, + { + "epoch": 0.5061497913463651, + "grad_norm": 2.2758829593658447, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6996133923530579, + "num_tokens": 118506389.0, + "step": 4609 + }, + { + "epoch": 0.5062596090489787, + "grad_norm": 2.0186288356781006, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6984291672706604, + "num_tokens": 118537949.0, + "step": 4610 + }, + { + "epoch": 0.5063694267515924, + "grad_norm": 2.2192704677581787, + "learning_rate": 1e-06, + "loss": 1.112, + "mean_token_accuracy": 0.6693540811538696, + "num_tokens": 118566450.0, + "step": 4611 + }, + { + "epoch": 0.506479244454206, + "grad_norm": 1.937700629234314, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6922997832298279, + "num_tokens": 118598962.0, + "step": 4612 + }, + { + "epoch": 0.5065890621568196, + "grad_norm": 2.0410001277923584, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7239617109298706, + "num_tokens": 118626528.0, + "step": 4613 + }, + { + "epoch": 0.5066988798594333, + "grad_norm": 2.208923816680908, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.70506352186203, + "num_tokens": 118652000.0, + "step": 4614 + }, + { + "epoch": 0.506808697562047, + "grad_norm": 2.1891379356384277, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6795159578323364, + "num_tokens": 118679112.0, + "step": 4615 + }, + { + "epoch": 0.5069185152646607, + "grad_norm": 2.2465803623199463, + "learning_rate": 1e-06, + "loss": 1.0642, + "mean_token_accuracy": 0.6839959621429443, + "num_tokens": 118705326.0, + "step": 4616 + }, + { + "epoch": 0.5070283329672743, + "grad_norm": 1.9737727642059326, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6937801837921143, + "num_tokens": 118734816.0, + "step": 4617 + }, + { + "epoch": 0.507138150669888, + "grad_norm": 2.111903429031372, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6888600587844849, + "num_tokens": 118762318.0, + "step": 4618 + }, + { + "epoch": 0.5072479683725016, + "grad_norm": 2.2401599884033203, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7140727043151855, + "num_tokens": 118787985.0, + "step": 4619 + }, + { + "epoch": 0.5073577860751153, + "grad_norm": 2.3183436393737793, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7254635095596313, + "num_tokens": 118808931.0, + "step": 4620 + }, + { + "epoch": 0.5074676037777289, + "grad_norm": 2.2807676792144775, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7020889520645142, + "num_tokens": 118832647.0, + "step": 4621 + }, + { + "epoch": 0.5075774214803427, + "grad_norm": 2.388000011444092, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6919065713882446, + "num_tokens": 118856173.0, + "step": 4622 + }, + { + "epoch": 0.5076872391829563, + "grad_norm": 2.040677309036255, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7081866264343262, + "num_tokens": 118883559.0, + "step": 4623 + }, + { + "epoch": 0.50779705688557, + "grad_norm": 2.175786256790161, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7064181566238403, + "num_tokens": 118908456.0, + "step": 4624 + }, + { + "epoch": 0.5079068745881836, + "grad_norm": 2.0478179454803467, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7038465738296509, + "num_tokens": 118937591.0, + "step": 4625 + }, + { + "epoch": 0.5080166922907973, + "grad_norm": 2.343069314956665, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6928838491439819, + "num_tokens": 118961491.0, + "step": 4626 + }, + { + "epoch": 0.5081265099934109, + "grad_norm": 2.1430063247680664, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6867654323577881, + "num_tokens": 118988866.0, + "step": 4627 + }, + { + "epoch": 0.5082363276960246, + "grad_norm": 2.016669273376465, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6784998178482056, + "num_tokens": 119018312.0, + "step": 4628 + }, + { + "epoch": 0.5083461453986382, + "grad_norm": 1.9659289121627808, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7094829082489014, + "num_tokens": 119048614.0, + "step": 4629 + }, + { + "epoch": 0.508455963101252, + "grad_norm": 2.13698148727417, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7052531242370605, + "num_tokens": 119075292.0, + "step": 4630 + }, + { + "epoch": 0.5085657808038656, + "grad_norm": 2.168701171875, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6878718137741089, + "num_tokens": 119103529.0, + "step": 4631 + }, + { + "epoch": 0.5086755985064793, + "grad_norm": 2.198768138885498, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6941016912460327, + "num_tokens": 119131222.0, + "step": 4632 + }, + { + "epoch": 0.5087854162090929, + "grad_norm": 2.24456787109375, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6972248554229736, + "num_tokens": 119155936.0, + "step": 4633 + }, + { + "epoch": 0.5088952339117065, + "grad_norm": 2.328533411026001, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7042698264122009, + "num_tokens": 119179878.0, + "step": 4634 + }, + { + "epoch": 0.5090050516143202, + "grad_norm": 2.2243711948394775, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7331516146659851, + "num_tokens": 119202898.0, + "step": 4635 + }, + { + "epoch": 0.5091148693169338, + "grad_norm": 2.2186014652252197, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7087116837501526, + "num_tokens": 119228213.0, + "step": 4636 + }, + { + "epoch": 0.5092246870195476, + "grad_norm": 2.223806381225586, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6872491836547852, + "num_tokens": 119254340.0, + "step": 4637 + }, + { + "epoch": 0.5093345047221612, + "grad_norm": 2.038724899291992, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7141425013542175, + "num_tokens": 119281411.0, + "step": 4638 + }, + { + "epoch": 0.5094443224247749, + "grad_norm": 1.9386509656906128, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6968622207641602, + "num_tokens": 119314401.0, + "step": 4639 + }, + { + "epoch": 0.5095541401273885, + "grad_norm": 2.2799220085144043, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6907188892364502, + "num_tokens": 119339499.0, + "step": 4640 + }, + { + "epoch": 0.5096639578300022, + "grad_norm": 2.268120527267456, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7039998769760132, + "num_tokens": 119364093.0, + "step": 4641 + }, + { + "epoch": 0.5097737755326158, + "grad_norm": 2.1397829055786133, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7009652853012085, + "num_tokens": 119390838.0, + "step": 4642 + }, + { + "epoch": 0.5098835932352295, + "grad_norm": 1.9337772130966187, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7161728143692017, + "num_tokens": 119421947.0, + "step": 4643 + }, + { + "epoch": 0.5099934109378432, + "grad_norm": 2.4342591762542725, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7125334739685059, + "num_tokens": 119443043.0, + "step": 4644 + }, + { + "epoch": 0.5101032286404569, + "grad_norm": 1.9552258253097534, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7032315731048584, + "num_tokens": 119474474.0, + "step": 4645 + }, + { + "epoch": 0.5102130463430705, + "grad_norm": 2.5206167697906494, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6831350326538086, + "num_tokens": 119495876.0, + "step": 4646 + }, + { + "epoch": 0.5103228640456842, + "grad_norm": 2.595109224319458, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6978527307510376, + "num_tokens": 119516622.0, + "step": 4647 + }, + { + "epoch": 0.5104326817482978, + "grad_norm": 2.0066895484924316, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6996488571166992, + "num_tokens": 119545130.0, + "step": 4648 + }, + { + "epoch": 0.5105424994509115, + "grad_norm": 2.2378365993499756, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6913827657699585, + "num_tokens": 119570676.0, + "step": 4649 + }, + { + "epoch": 0.5106523171535251, + "grad_norm": 2.080617666244507, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.704204797744751, + "num_tokens": 119597623.0, + "step": 4650 + }, + { + "epoch": 0.5107621348561389, + "grad_norm": 2.249877452850342, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7309962511062622, + "num_tokens": 119620343.0, + "step": 4651 + }, + { + "epoch": 0.5108719525587525, + "grad_norm": 2.4667396545410156, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6926778554916382, + "num_tokens": 119641350.0, + "step": 4652 + }, + { + "epoch": 0.5109817702613662, + "grad_norm": 2.215456247329712, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7020715475082397, + "num_tokens": 119666018.0, + "step": 4653 + }, + { + "epoch": 0.5110915879639798, + "grad_norm": 2.109304428100586, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7146087288856506, + "num_tokens": 119691869.0, + "step": 4654 + }, + { + "epoch": 0.5112014056665934, + "grad_norm": 2.081655740737915, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7228503227233887, + "num_tokens": 119717350.0, + "step": 4655 + }, + { + "epoch": 0.5113112233692071, + "grad_norm": 1.8657665252685547, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7276391983032227, + "num_tokens": 119747011.0, + "step": 4656 + }, + { + "epoch": 0.5114210410718207, + "grad_norm": 1.8926056623458862, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6984896063804626, + "num_tokens": 119779613.0, + "step": 4657 + }, + { + "epoch": 0.5115308587744345, + "grad_norm": 2.162431001663208, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6839358806610107, + "num_tokens": 119806965.0, + "step": 4658 + }, + { + "epoch": 0.5116406764770481, + "grad_norm": 2.6059513092041016, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7043710947036743, + "num_tokens": 119826252.0, + "step": 4659 + }, + { + "epoch": 0.5117504941796618, + "grad_norm": 2.06706166267395, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6931395530700684, + "num_tokens": 119856691.0, + "step": 4660 + }, + { + "epoch": 0.5118603118822754, + "grad_norm": 1.9741472005844116, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7106867432594299, + "num_tokens": 119885830.0, + "step": 4661 + }, + { + "epoch": 0.5119701295848891, + "grad_norm": 2.1069257259368896, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6934306621551514, + "num_tokens": 119914038.0, + "step": 4662 + }, + { + "epoch": 0.5120799472875027, + "grad_norm": 1.9867230653762817, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.699133038520813, + "num_tokens": 119944944.0, + "step": 4663 + }, + { + "epoch": 0.5121897649901164, + "grad_norm": 2.2797467708587646, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7169651389122009, + "num_tokens": 119967577.0, + "step": 4664 + }, + { + "epoch": 0.51229958269273, + "grad_norm": 2.3922336101531982, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7075703740119934, + "num_tokens": 119989705.0, + "step": 4665 + }, + { + "epoch": 0.5124094003953438, + "grad_norm": 1.981053352355957, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7019059658050537, + "num_tokens": 120020742.0, + "step": 4666 + }, + { + "epoch": 0.5125192180979574, + "grad_norm": 2.346975326538086, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.7101166248321533, + "num_tokens": 120044750.0, + "step": 4667 + }, + { + "epoch": 0.5126290358005711, + "grad_norm": 2.0080654621124268, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7163717746734619, + "num_tokens": 120072708.0, + "step": 4668 + }, + { + "epoch": 0.5127388535031847, + "grad_norm": 2.5654568672180176, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7346314191818237, + "num_tokens": 120090553.0, + "step": 4669 + }, + { + "epoch": 0.5128486712057984, + "grad_norm": 2.184265613555908, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7092263698577881, + "num_tokens": 120117005.0, + "step": 4670 + }, + { + "epoch": 0.512958488908412, + "grad_norm": 2.298544406890869, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.703655481338501, + "num_tokens": 120140027.0, + "step": 4671 + }, + { + "epoch": 0.5130683066110256, + "grad_norm": 2.5128591060638428, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6953622102737427, + "num_tokens": 120161699.0, + "step": 4672 + }, + { + "epoch": 0.5131781243136394, + "grad_norm": 2.335224151611328, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7068793773651123, + "num_tokens": 120184931.0, + "step": 4673 + }, + { + "epoch": 0.513287942016253, + "grad_norm": 2.1439945697784424, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6944867372512817, + "num_tokens": 120212596.0, + "step": 4674 + }, + { + "epoch": 0.5133977597188667, + "grad_norm": 2.465632677078247, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7165390253067017, + "num_tokens": 120233288.0, + "step": 4675 + }, + { + "epoch": 0.5135075774214803, + "grad_norm": 2.4325270652770996, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.709673285484314, + "num_tokens": 120254539.0, + "step": 4676 + }, + { + "epoch": 0.513617395124094, + "grad_norm": 1.9243838787078857, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7048974633216858, + "num_tokens": 120286058.0, + "step": 4677 + }, + { + "epoch": 0.5137272128267076, + "grad_norm": 2.3063061237335205, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7299578189849854, + "num_tokens": 120309490.0, + "step": 4678 + }, + { + "epoch": 0.5138370305293213, + "grad_norm": 2.6248505115509033, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7027453184127808, + "num_tokens": 120328505.0, + "step": 4679 + }, + { + "epoch": 0.513946848231935, + "grad_norm": 2.6838996410369873, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7122686505317688, + "num_tokens": 120347828.0, + "step": 4680 + }, + { + "epoch": 0.5140566659345487, + "grad_norm": 2.50028133392334, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7224422693252563, + "num_tokens": 120366991.0, + "step": 4681 + }, + { + "epoch": 0.5141664836371623, + "grad_norm": 2.4888763427734375, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7098913192749023, + "num_tokens": 120387956.0, + "step": 4682 + }, + { + "epoch": 0.514276301339776, + "grad_norm": 2.0916929244995117, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.7073708772659302, + "num_tokens": 120415270.0, + "step": 4683 + }, + { + "epoch": 0.5143861190423896, + "grad_norm": 2.159410238265991, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7170054912567139, + "num_tokens": 120441439.0, + "step": 4684 + }, + { + "epoch": 0.5144959367450033, + "grad_norm": 2.154498815536499, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7287805676460266, + "num_tokens": 120467548.0, + "step": 4685 + }, + { + "epoch": 0.5146057544476169, + "grad_norm": 2.037020444869995, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7079509496688843, + "num_tokens": 120496172.0, + "step": 4686 + }, + { + "epoch": 0.5147155721502307, + "grad_norm": 2.291696786880493, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6968364715576172, + "num_tokens": 120519759.0, + "step": 4687 + }, + { + "epoch": 0.5148253898528443, + "grad_norm": 2.5416946411132812, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6921477317810059, + "num_tokens": 120541317.0, + "step": 4688 + }, + { + "epoch": 0.514935207555458, + "grad_norm": 2.193254232406616, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6938461065292358, + "num_tokens": 120566600.0, + "step": 4689 + }, + { + "epoch": 0.5150450252580716, + "grad_norm": 2.101154088973999, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7056983709335327, + "num_tokens": 120593453.0, + "step": 4690 + }, + { + "epoch": 0.5151548429606853, + "grad_norm": 2.0536997318267822, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7111246585845947, + "num_tokens": 120620833.0, + "step": 4691 + }, + { + "epoch": 0.5152646606632989, + "grad_norm": 1.982894778251648, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6750668883323669, + "num_tokens": 120650925.0, + "step": 4692 + }, + { + "epoch": 0.5153744783659125, + "grad_norm": 2.3236916065216064, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.684604287147522, + "num_tokens": 120676173.0, + "step": 4693 + }, + { + "epoch": 0.5154842960685262, + "grad_norm": 2.511366128921509, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7216906547546387, + "num_tokens": 120695527.0, + "step": 4694 + }, + { + "epoch": 0.51559411377114, + "grad_norm": 2.3845150470733643, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6956652402877808, + "num_tokens": 120719240.0, + "step": 4695 + }, + { + "epoch": 0.5157039314737536, + "grad_norm": 2.251046657562256, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7274070978164673, + "num_tokens": 120742169.0, + "step": 4696 + }, + { + "epoch": 0.5158137491763672, + "grad_norm": 2.235018014907837, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7197413444519043, + "num_tokens": 120766914.0, + "step": 4697 + }, + { + "epoch": 0.5159235668789809, + "grad_norm": 2.2311911582946777, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7351044416427612, + "num_tokens": 120790906.0, + "step": 4698 + }, + { + "epoch": 0.5160333845815945, + "grad_norm": 1.9895247220993042, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6906698346138, + "num_tokens": 120823647.0, + "step": 4699 + }, + { + "epoch": 0.5161432022842082, + "grad_norm": 2.1227822303771973, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7202203273773193, + "num_tokens": 120849959.0, + "step": 4700 + }, + { + "epoch": 0.5162530199868218, + "grad_norm": 2.196284532546997, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7013857364654541, + "num_tokens": 120875690.0, + "step": 4701 + }, + { + "epoch": 0.5163628376894356, + "grad_norm": 2.107038974761963, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7045361995697021, + "num_tokens": 120903067.0, + "step": 4702 + }, + { + "epoch": 0.5164726553920492, + "grad_norm": 2.0738329887390137, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7129530906677246, + "num_tokens": 120931108.0, + "step": 4703 + }, + { + "epoch": 0.5165824730946629, + "grad_norm": 1.841375708580017, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6975014209747314, + "num_tokens": 120964773.0, + "step": 4704 + }, + { + "epoch": 0.5166922907972765, + "grad_norm": 2.1566317081451416, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7136987447738647, + "num_tokens": 120989156.0, + "step": 4705 + }, + { + "epoch": 0.5168021084998902, + "grad_norm": 2.1204617023468018, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7096506357192993, + "num_tokens": 121015329.0, + "step": 4706 + }, + { + "epoch": 0.5169119262025038, + "grad_norm": 2.3407111167907715, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7090927362442017, + "num_tokens": 121037428.0, + "step": 4707 + }, + { + "epoch": 0.5170217439051175, + "grad_norm": 2.641720771789551, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7162304520606995, + "num_tokens": 121058985.0, + "step": 4708 + }, + { + "epoch": 0.5171315616077312, + "grad_norm": 2.3230907917022705, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7213131785392761, + "num_tokens": 121080676.0, + "step": 4709 + }, + { + "epoch": 0.5172413793103449, + "grad_norm": 2.4424285888671875, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7271651029586792, + "num_tokens": 121100157.0, + "step": 4710 + }, + { + "epoch": 0.5173511970129585, + "grad_norm": 2.1057779788970947, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7233813405036926, + "num_tokens": 121127190.0, + "step": 4711 + }, + { + "epoch": 0.5174610147155722, + "grad_norm": 2.4105193614959717, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7013242244720459, + "num_tokens": 121148995.0, + "step": 4712 + }, + { + "epoch": 0.5175708324181858, + "grad_norm": 2.409451961517334, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7164634466171265, + "num_tokens": 121169389.0, + "step": 4713 + }, + { + "epoch": 0.5176806501207994, + "grad_norm": 2.1418721675872803, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6935795545578003, + "num_tokens": 121194497.0, + "step": 4714 + }, + { + "epoch": 0.5177904678234131, + "grad_norm": 2.176839828491211, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7022646069526672, + "num_tokens": 121219857.0, + "step": 4715 + }, + { + "epoch": 0.5179002855260268, + "grad_norm": 2.2166242599487305, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.698283314704895, + "num_tokens": 121245021.0, + "step": 4716 + }, + { + "epoch": 0.5180101032286405, + "grad_norm": 2.113128900527954, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6904550194740295, + "num_tokens": 121273638.0, + "step": 4717 + }, + { + "epoch": 0.5181199209312541, + "grad_norm": 2.525378942489624, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.708480954170227, + "num_tokens": 121294943.0, + "step": 4718 + }, + { + "epoch": 0.5182297386338678, + "grad_norm": 2.0076746940612793, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.702366292476654, + "num_tokens": 121325603.0, + "step": 4719 + }, + { + "epoch": 0.5183395563364814, + "grad_norm": 2.0679242610931396, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7046470642089844, + "num_tokens": 121352344.0, + "step": 4720 + }, + { + "epoch": 0.5184493740390951, + "grad_norm": 1.8204509019851685, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7049162983894348, + "num_tokens": 121384131.0, + "step": 4721 + }, + { + "epoch": 0.5185591917417087, + "grad_norm": 2.0212724208831787, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7175507545471191, + "num_tokens": 121413936.0, + "step": 4722 + }, + { + "epoch": 0.5186690094443224, + "grad_norm": 2.1237235069274902, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7009716629981995, + "num_tokens": 121440675.0, + "step": 4723 + }, + { + "epoch": 0.5187788271469361, + "grad_norm": 2.1461451053619385, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6960006952285767, + "num_tokens": 121466730.0, + "step": 4724 + }, + { + "epoch": 0.5188886448495498, + "grad_norm": 2.0367207527160645, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6907585859298706, + "num_tokens": 121496380.0, + "step": 4725 + }, + { + "epoch": 0.5189984625521634, + "grad_norm": 2.3489527702331543, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7026304006576538, + "num_tokens": 121518982.0, + "step": 4726 + }, + { + "epoch": 0.5191082802547771, + "grad_norm": 2.436589002609253, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7199822664260864, + "num_tokens": 121539577.0, + "step": 4727 + }, + { + "epoch": 0.5192180979573907, + "grad_norm": 2.095663070678711, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6979538202285767, + "num_tokens": 121568462.0, + "step": 4728 + }, + { + "epoch": 0.5193279156600044, + "grad_norm": 2.1364171504974365, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6923274993896484, + "num_tokens": 121596465.0, + "step": 4729 + }, + { + "epoch": 0.519437733362618, + "grad_norm": 2.405886173248291, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7177491188049316, + "num_tokens": 121617531.0, + "step": 4730 + }, + { + "epoch": 0.5195475510652318, + "grad_norm": 2.1700611114501953, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7296091318130493, + "num_tokens": 121641934.0, + "step": 4731 + }, + { + "epoch": 0.5196573687678454, + "grad_norm": 2.1738641262054443, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7155551314353943, + "num_tokens": 121668813.0, + "step": 4732 + }, + { + "epoch": 0.519767186470459, + "grad_norm": 2.0590741634368896, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6949211359024048, + "num_tokens": 121695048.0, + "step": 4733 + }, + { + "epoch": 0.5198770041730727, + "grad_norm": 2.1694118976593018, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7262163758277893, + "num_tokens": 121721490.0, + "step": 4734 + }, + { + "epoch": 0.5199868218756863, + "grad_norm": 2.32214093208313, + "learning_rate": 1e-06, + "loss": 1.1044, + "mean_token_accuracy": 0.6692203283309937, + "num_tokens": 121744479.0, + "step": 4735 + }, + { + "epoch": 0.5200966395783, + "grad_norm": 2.097174644470215, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6895222663879395, + "num_tokens": 121773219.0, + "step": 4736 + }, + { + "epoch": 0.5202064572809136, + "grad_norm": 2.233436346054077, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7165261507034302, + "num_tokens": 121796881.0, + "step": 4737 + }, + { + "epoch": 0.5203162749835274, + "grad_norm": 2.129940986633301, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.715582013130188, + "num_tokens": 121822950.0, + "step": 4738 + }, + { + "epoch": 0.520426092686141, + "grad_norm": 2.148635149002075, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6998776793479919, + "num_tokens": 121849736.0, + "step": 4739 + }, + { + "epoch": 0.5205359103887547, + "grad_norm": 2.3069093227386475, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7032210826873779, + "num_tokens": 121874383.0, + "step": 4740 + }, + { + "epoch": 0.5206457280913683, + "grad_norm": 2.1164963245391846, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7195645570755005, + "num_tokens": 121899821.0, + "step": 4741 + }, + { + "epoch": 0.520755545793982, + "grad_norm": 2.2684876918792725, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7127252817153931, + "num_tokens": 121922411.0, + "step": 4742 + }, + { + "epoch": 0.5208653634965956, + "grad_norm": 2.0283610820770264, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7183001041412354, + "num_tokens": 121952965.0, + "step": 4743 + }, + { + "epoch": 0.5209751811992093, + "grad_norm": 2.3243885040283203, + "learning_rate": 1e-06, + "loss": 1.0889, + "mean_token_accuracy": 0.6763420104980469, + "num_tokens": 121977611.0, + "step": 4744 + }, + { + "epoch": 0.521084998901823, + "grad_norm": 2.1522579193115234, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7123069763183594, + "num_tokens": 122003045.0, + "step": 4745 + }, + { + "epoch": 0.5211948166044367, + "grad_norm": 2.2567455768585205, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6881580352783203, + "num_tokens": 122027556.0, + "step": 4746 + }, + { + "epoch": 0.5213046343070503, + "grad_norm": 2.2818703651428223, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7170660495758057, + "num_tokens": 122050656.0, + "step": 4747 + }, + { + "epoch": 0.521414452009664, + "grad_norm": 2.338930130004883, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7002530097961426, + "num_tokens": 122073344.0, + "step": 4748 + }, + { + "epoch": 0.5215242697122776, + "grad_norm": 1.8688386678695679, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6904385685920715, + "num_tokens": 122108037.0, + "step": 4749 + }, + { + "epoch": 0.5216340874148913, + "grad_norm": 2.460756540298462, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7054430246353149, + "num_tokens": 122127974.0, + "step": 4750 + }, + { + "epoch": 0.5217439051175049, + "grad_norm": 2.278402090072632, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7103720903396606, + "num_tokens": 122150917.0, + "step": 4751 + }, + { + "epoch": 0.5218537228201185, + "grad_norm": 2.3103394508361816, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7190074920654297, + "num_tokens": 122174892.0, + "step": 4752 + }, + { + "epoch": 0.5219635405227323, + "grad_norm": 2.1079514026641846, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6886475682258606, + "num_tokens": 122203112.0, + "step": 4753 + }, + { + "epoch": 0.522073358225346, + "grad_norm": 2.039801836013794, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7039089202880859, + "num_tokens": 122233330.0, + "step": 4754 + }, + { + "epoch": 0.5221831759279596, + "grad_norm": 1.9628974199295044, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6893752813339233, + "num_tokens": 122265944.0, + "step": 4755 + }, + { + "epoch": 0.5222929936305732, + "grad_norm": 2.232712745666504, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7012832760810852, + "num_tokens": 122290735.0, + "step": 4756 + }, + { + "epoch": 0.5224028113331869, + "grad_norm": 2.077813148498535, + "learning_rate": 1e-06, + "loss": 1.0827, + "mean_token_accuracy": 0.6731084585189819, + "num_tokens": 122320706.0, + "step": 4757 + }, + { + "epoch": 0.5225126290358005, + "grad_norm": 2.0644073486328125, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6915060877799988, + "num_tokens": 122347584.0, + "step": 4758 + }, + { + "epoch": 0.5226224467384142, + "grad_norm": 1.9564412832260132, + "learning_rate": 1e-06, + "loss": 1.0723, + "mean_token_accuracy": 0.6816083192825317, + "num_tokens": 122382019.0, + "step": 4759 + }, + { + "epoch": 0.5227322644410279, + "grad_norm": 2.0935938358306885, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7274987101554871, + "num_tokens": 122407822.0, + "step": 4760 + }, + { + "epoch": 0.5228420821436416, + "grad_norm": 1.9641156196594238, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7118269205093384, + "num_tokens": 122437089.0, + "step": 4761 + }, + { + "epoch": 0.5229518998462552, + "grad_norm": 2.2189745903015137, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7196179032325745, + "num_tokens": 122461413.0, + "step": 4762 + }, + { + "epoch": 0.5230617175488689, + "grad_norm": 2.1767659187316895, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7318903803825378, + "num_tokens": 122484509.0, + "step": 4763 + }, + { + "epoch": 0.5231715352514825, + "grad_norm": 2.361412286758423, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7158091068267822, + "num_tokens": 122506461.0, + "step": 4764 + }, + { + "epoch": 0.5232813529540962, + "grad_norm": 2.3685319423675537, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7111274600028992, + "num_tokens": 122527048.0, + "step": 4765 + }, + { + "epoch": 0.5233911706567098, + "grad_norm": 2.2085695266723633, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.69841468334198, + "num_tokens": 122553230.0, + "step": 4766 + }, + { + "epoch": 0.5235009883593236, + "grad_norm": 2.0585474967956543, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6993733644485474, + "num_tokens": 122581090.0, + "step": 4767 + }, + { + "epoch": 0.5236108060619372, + "grad_norm": 2.5160837173461914, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7018285989761353, + "num_tokens": 122601114.0, + "step": 4768 + }, + { + "epoch": 0.5237206237645509, + "grad_norm": 2.226212978363037, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7057321071624756, + "num_tokens": 122628027.0, + "step": 4769 + }, + { + "epoch": 0.5238304414671645, + "grad_norm": 1.9729686975479126, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7107633948326111, + "num_tokens": 122657702.0, + "step": 4770 + }, + { + "epoch": 0.5239402591697782, + "grad_norm": 2.2347004413604736, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.734960675239563, + "num_tokens": 122680751.0, + "step": 4771 + }, + { + "epoch": 0.5240500768723918, + "grad_norm": 1.8915880918502808, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7274014949798584, + "num_tokens": 122711633.0, + "step": 4772 + }, + { + "epoch": 0.5241598945750054, + "grad_norm": 2.45428729057312, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7219011783599854, + "num_tokens": 122733627.0, + "step": 4773 + }, + { + "epoch": 0.5242697122776192, + "grad_norm": 2.159493923187256, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.708928108215332, + "num_tokens": 122758770.0, + "step": 4774 + }, + { + "epoch": 0.5243795299802329, + "grad_norm": 2.092893123626709, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7001991271972656, + "num_tokens": 122786693.0, + "step": 4775 + }, + { + "epoch": 0.5244893476828465, + "grad_norm": 1.9916274547576904, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6937923431396484, + "num_tokens": 122814902.0, + "step": 4776 + }, + { + "epoch": 0.5245991653854601, + "grad_norm": 2.3296403884887695, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7342233061790466, + "num_tokens": 122836427.0, + "step": 4777 + }, + { + "epoch": 0.5247089830880738, + "grad_norm": 2.1831140518188477, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.739476203918457, + "num_tokens": 122860470.0, + "step": 4778 + }, + { + "epoch": 0.5248188007906874, + "grad_norm": 2.744288682937622, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6917585730552673, + "num_tokens": 122879547.0, + "step": 4779 + }, + { + "epoch": 0.5249286184933011, + "grad_norm": 2.1349809169769287, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7070150375366211, + "num_tokens": 122907128.0, + "step": 4780 + }, + { + "epoch": 0.5250384361959147, + "grad_norm": 2.148042917251587, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6947607398033142, + "num_tokens": 122934941.0, + "step": 4781 + }, + { + "epoch": 0.5251482538985285, + "grad_norm": 2.253063440322876, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6999281644821167, + "num_tokens": 122959460.0, + "step": 4782 + }, + { + "epoch": 0.5252580716011421, + "grad_norm": 2.1935641765594482, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6960477828979492, + "num_tokens": 122987183.0, + "step": 4783 + }, + { + "epoch": 0.5253678893037558, + "grad_norm": 2.067495822906494, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6810585260391235, + "num_tokens": 123019190.0, + "step": 4784 + }, + { + "epoch": 0.5254777070063694, + "grad_norm": 2.384490966796875, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7213844060897827, + "num_tokens": 123038917.0, + "step": 4785 + }, + { + "epoch": 0.5255875247089831, + "grad_norm": 2.329737424850464, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6960840225219727, + "num_tokens": 123062583.0, + "step": 4786 + }, + { + "epoch": 0.5256973424115967, + "grad_norm": 2.28560209274292, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6946196556091309, + "num_tokens": 123085078.0, + "step": 4787 + }, + { + "epoch": 0.5258071601142104, + "grad_norm": 2.2323198318481445, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7124404311180115, + "num_tokens": 123109340.0, + "step": 4788 + }, + { + "epoch": 0.5259169778168241, + "grad_norm": 2.2423672676086426, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7249107956886292, + "num_tokens": 123132050.0, + "step": 4789 + }, + { + "epoch": 0.5260267955194378, + "grad_norm": 2.3788678646087646, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7033588290214539, + "num_tokens": 123153809.0, + "step": 4790 + }, + { + "epoch": 0.5261366132220514, + "grad_norm": 2.3221442699432373, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7036315202713013, + "num_tokens": 123178779.0, + "step": 4791 + }, + { + "epoch": 0.526246430924665, + "grad_norm": 2.395409345626831, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6961669325828552, + "num_tokens": 123203457.0, + "step": 4792 + }, + { + "epoch": 0.5263562486272787, + "grad_norm": 2.3436272144317627, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7090678811073303, + "num_tokens": 123225850.0, + "step": 4793 + }, + { + "epoch": 0.5264660663298923, + "grad_norm": 2.156493663787842, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6967581510543823, + "num_tokens": 123252073.0, + "step": 4794 + }, + { + "epoch": 0.526575884032506, + "grad_norm": 2.0178394317626953, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.699121356010437, + "num_tokens": 123279738.0, + "step": 4795 + }, + { + "epoch": 0.5266857017351197, + "grad_norm": 2.176616668701172, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6783192157745361, + "num_tokens": 123305230.0, + "step": 4796 + }, + { + "epoch": 0.5267955194377334, + "grad_norm": 2.173278331756592, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7169209122657776, + "num_tokens": 123328840.0, + "step": 4797 + }, + { + "epoch": 0.526905337140347, + "grad_norm": 2.2577638626098633, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7180270552635193, + "num_tokens": 123351195.0, + "step": 4798 + }, + { + "epoch": 0.5270151548429607, + "grad_norm": 2.225027561187744, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6988977193832397, + "num_tokens": 123378200.0, + "step": 4799 + }, + { + "epoch": 0.5271249725455743, + "grad_norm": 2.4062886238098145, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7001438736915588, + "num_tokens": 123400650.0, + "step": 4800 + }, + { + "epoch": 0.527234790248188, + "grad_norm": 2.096313953399658, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6938276886940002, + "num_tokens": 123425763.0, + "step": 4801 + }, + { + "epoch": 0.5273446079508016, + "grad_norm": 2.401563882827759, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.7011635303497314, + "num_tokens": 123448927.0, + "step": 4802 + }, + { + "epoch": 0.5274544256534154, + "grad_norm": 2.354079484939575, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6955999135971069, + "num_tokens": 123472569.0, + "step": 4803 + }, + { + "epoch": 0.527564243356029, + "grad_norm": 2.1217849254608154, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6767702102661133, + "num_tokens": 123499546.0, + "step": 4804 + }, + { + "epoch": 0.5276740610586427, + "grad_norm": 2.144582509994507, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7166901230812073, + "num_tokens": 123525353.0, + "step": 4805 + }, + { + "epoch": 0.5277838787612563, + "grad_norm": 2.2402138710021973, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7191126942634583, + "num_tokens": 123549429.0, + "step": 4806 + }, + { + "epoch": 0.52789369646387, + "grad_norm": 2.161928653717041, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7102868556976318, + "num_tokens": 123579687.0, + "step": 4807 + }, + { + "epoch": 0.5280035141664836, + "grad_norm": 2.380481004714966, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7007944583892822, + "num_tokens": 123601655.0, + "step": 4808 + }, + { + "epoch": 0.5281133318690973, + "grad_norm": 1.972224473953247, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6985522508621216, + "num_tokens": 123630829.0, + "step": 4809 + }, + { + "epoch": 0.528223149571711, + "grad_norm": 2.0883219242095947, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7075958847999573, + "num_tokens": 123659190.0, + "step": 4810 + }, + { + "epoch": 0.5283329672743247, + "grad_norm": 2.1868038177490234, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6944596767425537, + "num_tokens": 123686550.0, + "step": 4811 + }, + { + "epoch": 0.5284427849769383, + "grad_norm": 2.274589776992798, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7127332091331482, + "num_tokens": 123712204.0, + "step": 4812 + }, + { + "epoch": 0.528552602679552, + "grad_norm": 2.1739447116851807, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7032759785652161, + "num_tokens": 123739398.0, + "step": 4813 + }, + { + "epoch": 0.5286624203821656, + "grad_norm": 2.455821990966797, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7075955867767334, + "num_tokens": 123760507.0, + "step": 4814 + }, + { + "epoch": 0.5287722380847792, + "grad_norm": 1.9968221187591553, + "learning_rate": 1e-06, + "loss": 1.0832, + "mean_token_accuracy": 0.6813317537307739, + "num_tokens": 123789596.0, + "step": 4815 + }, + { + "epoch": 0.5288820557873929, + "grad_norm": 2.1753504276275635, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6941523551940918, + "num_tokens": 123816056.0, + "step": 4816 + }, + { + "epoch": 0.5289918734900065, + "grad_norm": 2.276287794113159, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7072340846061707, + "num_tokens": 123838355.0, + "step": 4817 + }, + { + "epoch": 0.5291016911926203, + "grad_norm": 1.8951985836029053, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.677487850189209, + "num_tokens": 123870848.0, + "step": 4818 + }, + { + "epoch": 0.5292115088952339, + "grad_norm": 2.2174277305603027, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6959319114685059, + "num_tokens": 123898402.0, + "step": 4819 + }, + { + "epoch": 0.5293213265978476, + "grad_norm": 2.2840828895568848, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7022230625152588, + "num_tokens": 123924094.0, + "step": 4820 + }, + { + "epoch": 0.5294311443004612, + "grad_norm": 2.154139280319214, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6936050653457642, + "num_tokens": 123951683.0, + "step": 4821 + }, + { + "epoch": 0.5295409620030749, + "grad_norm": 1.9601097106933594, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6983746290206909, + "num_tokens": 123981801.0, + "step": 4822 + }, + { + "epoch": 0.5296507797056885, + "grad_norm": 2.0941643714904785, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7027648687362671, + "num_tokens": 124010255.0, + "step": 4823 + }, + { + "epoch": 0.5297605974083022, + "grad_norm": 2.1550650596618652, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.713692307472229, + "num_tokens": 124036743.0, + "step": 4824 + }, + { + "epoch": 0.5298704151109159, + "grad_norm": 2.1829440593719482, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7045586109161377, + "num_tokens": 124063134.0, + "step": 4825 + }, + { + "epoch": 0.5299802328135296, + "grad_norm": 2.2423722743988037, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7129726409912109, + "num_tokens": 124087105.0, + "step": 4826 + }, + { + "epoch": 0.5300900505161432, + "grad_norm": 2.386375904083252, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7168750762939453, + "num_tokens": 124110580.0, + "step": 4827 + }, + { + "epoch": 0.5301998682187569, + "grad_norm": 2.150111198425293, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.701736569404602, + "num_tokens": 124135617.0, + "step": 4828 + }, + { + "epoch": 0.5303096859213705, + "grad_norm": 2.1894898414611816, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7115974426269531, + "num_tokens": 124159874.0, + "step": 4829 + }, + { + "epoch": 0.5304195036239842, + "grad_norm": 2.052239418029785, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7008599042892456, + "num_tokens": 124190494.0, + "step": 4830 + }, + { + "epoch": 0.5305293213265978, + "grad_norm": 2.083803653717041, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7255398631095886, + "num_tokens": 124217445.0, + "step": 4831 + }, + { + "epoch": 0.5306391390292116, + "grad_norm": 2.192991018295288, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6969252824783325, + "num_tokens": 124242328.0, + "step": 4832 + }, + { + "epoch": 0.5307489567318252, + "grad_norm": 2.5969796180725098, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6978243589401245, + "num_tokens": 124262952.0, + "step": 4833 + }, + { + "epoch": 0.5308587744344389, + "grad_norm": 2.3074357509613037, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7376149892807007, + "num_tokens": 124285673.0, + "step": 4834 + }, + { + "epoch": 0.5309685921370525, + "grad_norm": 2.3234221935272217, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6978071928024292, + "num_tokens": 124311436.0, + "step": 4835 + }, + { + "epoch": 0.5310784098396661, + "grad_norm": 2.304530143737793, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6936305165290833, + "num_tokens": 124334989.0, + "step": 4836 + }, + { + "epoch": 0.5311882275422798, + "grad_norm": 2.2168750762939453, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7196065187454224, + "num_tokens": 124359850.0, + "step": 4837 + }, + { + "epoch": 0.5312980452448934, + "grad_norm": 2.162623643875122, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7139052152633667, + "num_tokens": 124385476.0, + "step": 4838 + }, + { + "epoch": 0.5314078629475072, + "grad_norm": 2.193032741546631, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6922831535339355, + "num_tokens": 124411623.0, + "step": 4839 + }, + { + "epoch": 0.5315176806501208, + "grad_norm": 2.4022252559661865, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7146679162979126, + "num_tokens": 124431526.0, + "step": 4840 + }, + { + "epoch": 0.5316274983527345, + "grad_norm": 1.9788087606430054, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6826649308204651, + "num_tokens": 124461816.0, + "step": 4841 + }, + { + "epoch": 0.5317373160553481, + "grad_norm": 2.3152835369110107, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7007125616073608, + "num_tokens": 124484824.0, + "step": 4842 + }, + { + "epoch": 0.5318471337579618, + "grad_norm": 2.108513832092285, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7223964333534241, + "num_tokens": 124508749.0, + "step": 4843 + }, + { + "epoch": 0.5319569514605754, + "grad_norm": 2.4501068592071533, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.7098141312599182, + "num_tokens": 124528744.0, + "step": 4844 + }, + { + "epoch": 0.5320667691631891, + "grad_norm": 2.254202127456665, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6796542406082153, + "num_tokens": 124554383.0, + "step": 4845 + }, + { + "epoch": 0.5321765868658027, + "grad_norm": 2.292537212371826, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6934899091720581, + "num_tokens": 124577904.0, + "step": 4846 + }, + { + "epoch": 0.5322864045684165, + "grad_norm": 2.174788236618042, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7082611918449402, + "num_tokens": 124601042.0, + "step": 4847 + }, + { + "epoch": 0.5323962222710301, + "grad_norm": 1.9006061553955078, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6998289823532104, + "num_tokens": 124631073.0, + "step": 4848 + }, + { + "epoch": 0.5325060399736438, + "grad_norm": 2.2306437492370605, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7212100028991699, + "num_tokens": 124655964.0, + "step": 4849 + }, + { + "epoch": 0.5326158576762574, + "grad_norm": 2.110743284225464, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6818217039108276, + "num_tokens": 124683703.0, + "step": 4850 + }, + { + "epoch": 0.532725675378871, + "grad_norm": 2.137766122817993, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7095640897750854, + "num_tokens": 124711148.0, + "step": 4851 + }, + { + "epoch": 0.5328354930814847, + "grad_norm": 2.3258438110351562, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7032289505004883, + "num_tokens": 124732989.0, + "step": 4852 + }, + { + "epoch": 0.5329453107840983, + "grad_norm": 2.139786720275879, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6817291975021362, + "num_tokens": 124761967.0, + "step": 4853 + }, + { + "epoch": 0.5330551284867121, + "grad_norm": 2.6938278675079346, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7026695013046265, + "num_tokens": 124781275.0, + "step": 4854 + }, + { + "epoch": 0.5331649461893258, + "grad_norm": 2.093219041824341, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6992360353469849, + "num_tokens": 124809930.0, + "step": 4855 + }, + { + "epoch": 0.5332747638919394, + "grad_norm": 2.03828501701355, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7058523297309875, + "num_tokens": 124836901.0, + "step": 4856 + }, + { + "epoch": 0.533384581594553, + "grad_norm": 2.0896317958831787, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7066631317138672, + "num_tokens": 124863147.0, + "step": 4857 + }, + { + "epoch": 0.5334943992971667, + "grad_norm": 2.2497518062591553, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7204093933105469, + "num_tokens": 124887965.0, + "step": 4858 + }, + { + "epoch": 0.5336042169997803, + "grad_norm": 1.999935507774353, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7017983198165894, + "num_tokens": 124915718.0, + "step": 4859 + }, + { + "epoch": 0.533714034702394, + "grad_norm": 2.0062546730041504, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.7000680565834045, + "num_tokens": 124946447.0, + "step": 4860 + }, + { + "epoch": 0.5338238524050077, + "grad_norm": 2.069478988647461, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6966359615325928, + "num_tokens": 124975459.0, + "step": 4861 + }, + { + "epoch": 0.5339336701076214, + "grad_norm": 2.2072560787200928, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7054792642593384, + "num_tokens": 125001038.0, + "step": 4862 + }, + { + "epoch": 0.534043487810235, + "grad_norm": 1.807915210723877, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6950483322143555, + "num_tokens": 125038066.0, + "step": 4863 + }, + { + "epoch": 0.5341533055128487, + "grad_norm": 2.0032498836517334, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7021604180335999, + "num_tokens": 125065630.0, + "step": 4864 + }, + { + "epoch": 0.5342631232154623, + "grad_norm": 2.1057610511779785, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6957969069480896, + "num_tokens": 125093093.0, + "step": 4865 + }, + { + "epoch": 0.534372940918076, + "grad_norm": 2.545679807662964, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7070873379707336, + "num_tokens": 125113986.0, + "step": 4866 + }, + { + "epoch": 0.5344827586206896, + "grad_norm": 2.493842124938965, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7048137187957764, + "num_tokens": 125137577.0, + "step": 4867 + }, + { + "epoch": 0.5345925763233034, + "grad_norm": 2.483147382736206, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7095141410827637, + "num_tokens": 125158940.0, + "step": 4868 + }, + { + "epoch": 0.534702394025917, + "grad_norm": 2.0872106552124023, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7208960652351379, + "num_tokens": 125184239.0, + "step": 4869 + }, + { + "epoch": 0.5348122117285307, + "grad_norm": 2.211097240447998, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7050604224205017, + "num_tokens": 125209579.0, + "step": 4870 + }, + { + "epoch": 0.5349220294311443, + "grad_norm": 2.2024359703063965, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6927379965782166, + "num_tokens": 125237134.0, + "step": 4871 + }, + { + "epoch": 0.535031847133758, + "grad_norm": 2.193765640258789, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7212413549423218, + "num_tokens": 125259367.0, + "step": 4872 + }, + { + "epoch": 0.5351416648363716, + "grad_norm": 2.1046998500823975, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.6766879558563232, + "num_tokens": 125290141.0, + "step": 4873 + }, + { + "epoch": 0.5352514825389852, + "grad_norm": 2.190908193588257, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7073953151702881, + "num_tokens": 125315401.0, + "step": 4874 + }, + { + "epoch": 0.5353613002415989, + "grad_norm": 2.698622703552246, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7040361166000366, + "num_tokens": 125334844.0, + "step": 4875 + }, + { + "epoch": 0.5354711179442126, + "grad_norm": 2.7229530811309814, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7164677381515503, + "num_tokens": 125353625.0, + "step": 4876 + }, + { + "epoch": 0.5355809356468263, + "grad_norm": 2.274322748184204, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6873568296432495, + "num_tokens": 125378165.0, + "step": 4877 + }, + { + "epoch": 0.5356907533494399, + "grad_norm": 2.076202630996704, + "learning_rate": 1e-06, + "loss": 1.0893, + "mean_token_accuracy": 0.6745927333831787, + "num_tokens": 125410338.0, + "step": 4878 + }, + { + "epoch": 0.5358005710520536, + "grad_norm": 2.0487077236175537, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.695603609085083, + "num_tokens": 125438186.0, + "step": 4879 + }, + { + "epoch": 0.5359103887546672, + "grad_norm": 2.195544719696045, + "learning_rate": 1e-06, + "loss": 1.0866, + "mean_token_accuracy": 0.6812240481376648, + "num_tokens": 125463970.0, + "step": 4880 + }, + { + "epoch": 0.5360202064572809, + "grad_norm": 2.0706286430358887, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7089599370956421, + "num_tokens": 125490153.0, + "step": 4881 + }, + { + "epoch": 0.5361300241598945, + "grad_norm": 2.3978943824768066, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.704696536064148, + "num_tokens": 125511238.0, + "step": 4882 + }, + { + "epoch": 0.5362398418625083, + "grad_norm": 2.329932689666748, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7250229716300964, + "num_tokens": 125534052.0, + "step": 4883 + }, + { + "epoch": 0.5363496595651219, + "grad_norm": 2.3811559677124023, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7198328971862793, + "num_tokens": 125555221.0, + "step": 4884 + }, + { + "epoch": 0.5364594772677356, + "grad_norm": 2.410343885421753, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7072004079818726, + "num_tokens": 125578139.0, + "step": 4885 + }, + { + "epoch": 0.5365692949703492, + "grad_norm": 2.05726957321167, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7097232937812805, + "num_tokens": 125604273.0, + "step": 4886 + }, + { + "epoch": 0.5366791126729629, + "grad_norm": 2.332381010055542, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.7038865089416504, + "num_tokens": 125627950.0, + "step": 4887 + }, + { + "epoch": 0.5367889303755765, + "grad_norm": 2.0413386821746826, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.710175096988678, + "num_tokens": 125658949.0, + "step": 4888 + }, + { + "epoch": 0.5368987480781902, + "grad_norm": 2.355386734008789, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6831779479980469, + "num_tokens": 125684424.0, + "step": 4889 + }, + { + "epoch": 0.5370085657808039, + "grad_norm": 1.9317845106124878, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6888065338134766, + "num_tokens": 125716871.0, + "step": 4890 + }, + { + "epoch": 0.5371183834834176, + "grad_norm": 2.450191020965576, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7035821080207825, + "num_tokens": 125740120.0, + "step": 4891 + }, + { + "epoch": 0.5372282011860312, + "grad_norm": 2.410815954208374, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7168950438499451, + "num_tokens": 125763461.0, + "step": 4892 + }, + { + "epoch": 0.5373380188886449, + "grad_norm": 1.9032847881317139, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6846332550048828, + "num_tokens": 125796128.0, + "step": 4893 + }, + { + "epoch": 0.5374478365912585, + "grad_norm": 2.152639627456665, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7097086906433105, + "num_tokens": 125821913.0, + "step": 4894 + }, + { + "epoch": 0.5375576542938721, + "grad_norm": 2.1240270137786865, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6902908086776733, + "num_tokens": 125848916.0, + "step": 4895 + }, + { + "epoch": 0.5376674719964858, + "grad_norm": 2.3565824031829834, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.713714599609375, + "num_tokens": 125870917.0, + "step": 4896 + }, + { + "epoch": 0.5377772896990995, + "grad_norm": 2.191251277923584, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6987320780754089, + "num_tokens": 125900073.0, + "step": 4897 + }, + { + "epoch": 0.5378871074017132, + "grad_norm": 2.1483185291290283, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6883344650268555, + "num_tokens": 125927638.0, + "step": 4898 + }, + { + "epoch": 0.5379969251043268, + "grad_norm": 2.5056402683258057, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7005377411842346, + "num_tokens": 125949254.0, + "step": 4899 + }, + { + "epoch": 0.5381067428069405, + "grad_norm": 2.2639098167419434, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7080065011978149, + "num_tokens": 125974076.0, + "step": 4900 + }, + { + "epoch": 0.5382165605095541, + "grad_norm": 2.265652894973755, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6928606033325195, + "num_tokens": 126000630.0, + "step": 4901 + }, + { + "epoch": 0.5383263782121678, + "grad_norm": 2.1723198890686035, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6924052238464355, + "num_tokens": 126025908.0, + "step": 4902 + }, + { + "epoch": 0.5384361959147814, + "grad_norm": 2.0186097621917725, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6926959753036499, + "num_tokens": 126056725.0, + "step": 4903 + }, + { + "epoch": 0.5385460136173951, + "grad_norm": 2.127429962158203, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7031136751174927, + "num_tokens": 126085830.0, + "step": 4904 + }, + { + "epoch": 0.5386558313200088, + "grad_norm": 2.259202241897583, + "learning_rate": 1e-06, + "loss": 1.0765, + "mean_token_accuracy": 0.6793620586395264, + "num_tokens": 126111412.0, + "step": 4905 + }, + { + "epoch": 0.5387656490226225, + "grad_norm": 2.4354324340820312, + "learning_rate": 1e-06, + "loss": 1.0916, + "mean_token_accuracy": 0.6800158023834229, + "num_tokens": 126134171.0, + "step": 4906 + }, + { + "epoch": 0.5388754667252361, + "grad_norm": 2.3869380950927734, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7220322489738464, + "num_tokens": 126155627.0, + "step": 4907 + }, + { + "epoch": 0.5389852844278498, + "grad_norm": 2.193312406539917, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7259804010391235, + "num_tokens": 126179925.0, + "step": 4908 + }, + { + "epoch": 0.5390951021304634, + "grad_norm": 1.9683302640914917, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7115875482559204, + "num_tokens": 126209684.0, + "step": 4909 + }, + { + "epoch": 0.5392049198330771, + "grad_norm": 2.213670492172241, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7190364599227905, + "num_tokens": 126236666.0, + "step": 4910 + }, + { + "epoch": 0.5393147375356907, + "grad_norm": 2.205718517303467, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7080983519554138, + "num_tokens": 126260728.0, + "step": 4911 + }, + { + "epoch": 0.5394245552383045, + "grad_norm": 2.461700201034546, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6936249732971191, + "num_tokens": 126282142.0, + "step": 4912 + }, + { + "epoch": 0.5395343729409181, + "grad_norm": 2.203376054763794, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.698033332824707, + "num_tokens": 126307847.0, + "step": 4913 + }, + { + "epoch": 0.5396441906435318, + "grad_norm": 2.161139726638794, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.70073401927948, + "num_tokens": 126334394.0, + "step": 4914 + }, + { + "epoch": 0.5397540083461454, + "grad_norm": 2.0843167304992676, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7031818628311157, + "num_tokens": 126360449.0, + "step": 4915 + }, + { + "epoch": 0.539863826048759, + "grad_norm": 2.0840580463409424, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7144699692726135, + "num_tokens": 126388055.0, + "step": 4916 + }, + { + "epoch": 0.5399736437513727, + "grad_norm": 2.1630735397338867, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6920385360717773, + "num_tokens": 126414187.0, + "step": 4917 + }, + { + "epoch": 0.5400834614539863, + "grad_norm": 2.2867496013641357, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7118891477584839, + "num_tokens": 126436738.0, + "step": 4918 + }, + { + "epoch": 0.5401932791566001, + "grad_norm": 2.4006354808807373, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6728420257568359, + "num_tokens": 126459882.0, + "step": 4919 + }, + { + "epoch": 0.5403030968592137, + "grad_norm": 2.7124288082122803, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6946511268615723, + "num_tokens": 126478399.0, + "step": 4920 + }, + { + "epoch": 0.5404129145618274, + "grad_norm": 2.0268383026123047, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7029893398284912, + "num_tokens": 126508732.0, + "step": 4921 + }, + { + "epoch": 0.540522732264441, + "grad_norm": 1.982254981994629, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6873132586479187, + "num_tokens": 126538532.0, + "step": 4922 + }, + { + "epoch": 0.5406325499670547, + "grad_norm": 2.3129589557647705, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6958497762680054, + "num_tokens": 126563547.0, + "step": 4923 + }, + { + "epoch": 0.5407423676696683, + "grad_norm": 2.5890519618988037, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6975305080413818, + "num_tokens": 126583548.0, + "step": 4924 + }, + { + "epoch": 0.540852185372282, + "grad_norm": 2.0829360485076904, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6857324242591858, + "num_tokens": 126613748.0, + "step": 4925 + }, + { + "epoch": 0.5409620030748957, + "grad_norm": 2.2700278759002686, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7224459648132324, + "num_tokens": 126636650.0, + "step": 4926 + }, + { + "epoch": 0.5410718207775094, + "grad_norm": 2.1411197185516357, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6818053722381592, + "num_tokens": 126662793.0, + "step": 4927 + }, + { + "epoch": 0.541181638480123, + "grad_norm": 2.1371490955352783, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7224153876304626, + "num_tokens": 126688503.0, + "step": 4928 + }, + { + "epoch": 0.5412914561827367, + "grad_norm": 2.1259713172912598, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6876498460769653, + "num_tokens": 126716373.0, + "step": 4929 + }, + { + "epoch": 0.5414012738853503, + "grad_norm": 2.591907262802124, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7109994888305664, + "num_tokens": 126735602.0, + "step": 4930 + }, + { + "epoch": 0.541511091587964, + "grad_norm": 2.3603713512420654, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7142283916473389, + "num_tokens": 126756727.0, + "step": 4931 + }, + { + "epoch": 0.5416209092905776, + "grad_norm": 2.255167007446289, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6990100741386414, + "num_tokens": 126780652.0, + "step": 4932 + }, + { + "epoch": 0.5417307269931912, + "grad_norm": 2.0622506141662598, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7054064273834229, + "num_tokens": 126807457.0, + "step": 4933 + }, + { + "epoch": 0.541840544695805, + "grad_norm": 2.123854160308838, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6897092461585999, + "num_tokens": 126835633.0, + "step": 4934 + }, + { + "epoch": 0.5419503623984187, + "grad_norm": 2.510114908218384, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7136662006378174, + "num_tokens": 126855459.0, + "step": 4935 + }, + { + "epoch": 0.5420601801010323, + "grad_norm": 2.0002593994140625, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7273504734039307, + "num_tokens": 126883905.0, + "step": 4936 + }, + { + "epoch": 0.5421699978036459, + "grad_norm": 2.2558183670043945, + "learning_rate": 1e-06, + "loss": 1.0941, + "mean_token_accuracy": 0.6753093004226685, + "num_tokens": 126908021.0, + "step": 4937 + }, + { + "epoch": 0.5422798155062596, + "grad_norm": 2.0996360778808594, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6970080137252808, + "num_tokens": 126937141.0, + "step": 4938 + }, + { + "epoch": 0.5423896332088732, + "grad_norm": 2.1386494636535645, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.7008036375045776, + "num_tokens": 126964376.0, + "step": 4939 + }, + { + "epoch": 0.5424994509114869, + "grad_norm": 2.2043254375457764, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7036443948745728, + "num_tokens": 126989726.0, + "step": 4940 + }, + { + "epoch": 0.5426092686141006, + "grad_norm": 2.382139205932617, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6830493807792664, + "num_tokens": 127014204.0, + "step": 4941 + }, + { + "epoch": 0.5427190863167143, + "grad_norm": 2.2458009719848633, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7251807451248169, + "num_tokens": 127037691.0, + "step": 4942 + }, + { + "epoch": 0.5428289040193279, + "grad_norm": 2.010195732116699, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6947869062423706, + "num_tokens": 127067547.0, + "step": 4943 + }, + { + "epoch": 0.5429387217219416, + "grad_norm": 2.2434630393981934, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7219541072845459, + "num_tokens": 127091505.0, + "step": 4944 + }, + { + "epoch": 0.5430485394245552, + "grad_norm": 2.2003417015075684, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7169642448425293, + "num_tokens": 127115982.0, + "step": 4945 + }, + { + "epoch": 0.5431583571271689, + "grad_norm": 1.9956926107406616, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6929621696472168, + "num_tokens": 127147021.0, + "step": 4946 + }, + { + "epoch": 0.5432681748297825, + "grad_norm": 2.3566887378692627, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7029747366905212, + "num_tokens": 127169394.0, + "step": 4947 + }, + { + "epoch": 0.5433779925323963, + "grad_norm": 2.297647476196289, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7077793478965759, + "num_tokens": 127192382.0, + "step": 4948 + }, + { + "epoch": 0.5434878102350099, + "grad_norm": 2.196808099746704, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6901161074638367, + "num_tokens": 127217673.0, + "step": 4949 + }, + { + "epoch": 0.5435976279376236, + "grad_norm": 2.043092966079712, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7020739316940308, + "num_tokens": 127246114.0, + "step": 4950 + }, + { + "epoch": 0.5437074456402372, + "grad_norm": 2.264139413833618, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.701004147529602, + "num_tokens": 127269981.0, + "step": 4951 + }, + { + "epoch": 0.5438172633428509, + "grad_norm": 2.542263984680176, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7033485174179077, + "num_tokens": 127290330.0, + "step": 4952 + }, + { + "epoch": 0.5439270810454645, + "grad_norm": 2.126509666442871, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7101442217826843, + "num_tokens": 127317417.0, + "step": 4953 + }, + { + "epoch": 0.5440368987480781, + "grad_norm": 1.9904900789260864, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6768242120742798, + "num_tokens": 127348942.0, + "step": 4954 + }, + { + "epoch": 0.5441467164506919, + "grad_norm": 1.9367142915725708, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.701298713684082, + "num_tokens": 127379082.0, + "step": 4955 + }, + { + "epoch": 0.5442565341533055, + "grad_norm": 2.2111117839813232, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7058649659156799, + "num_tokens": 127405093.0, + "step": 4956 + }, + { + "epoch": 0.5443663518559192, + "grad_norm": 2.3658716678619385, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.716048538684845, + "num_tokens": 127426921.0, + "step": 4957 + }, + { + "epoch": 0.5444761695585328, + "grad_norm": 2.2069339752197266, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6865717768669128, + "num_tokens": 127453203.0, + "step": 4958 + }, + { + "epoch": 0.5445859872611465, + "grad_norm": 2.451197624206543, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7198854684829712, + "num_tokens": 127473928.0, + "step": 4959 + }, + { + "epoch": 0.5446958049637601, + "grad_norm": 2.2313265800476074, + "learning_rate": 1e-06, + "loss": 1.0957, + "mean_token_accuracy": 0.6731864213943481, + "num_tokens": 127500677.0, + "step": 4960 + }, + { + "epoch": 0.5448056226663738, + "grad_norm": 2.2548904418945312, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6813055276870728, + "num_tokens": 127525861.0, + "step": 4961 + }, + { + "epoch": 0.5449154403689875, + "grad_norm": 2.527601957321167, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.715973973274231, + "num_tokens": 127546338.0, + "step": 4962 + }, + { + "epoch": 0.5450252580716012, + "grad_norm": 2.3379969596862793, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7123806476593018, + "num_tokens": 127569303.0, + "step": 4963 + }, + { + "epoch": 0.5451350757742148, + "grad_norm": 2.271028518676758, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.6997389793395996, + "num_tokens": 127593685.0, + "step": 4964 + }, + { + "epoch": 0.5452448934768285, + "grad_norm": 2.5595901012420654, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7160775661468506, + "num_tokens": 127614753.0, + "step": 4965 + }, + { + "epoch": 0.5453547111794421, + "grad_norm": 2.22308611869812, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7027826309204102, + "num_tokens": 127639477.0, + "step": 4966 + }, + { + "epoch": 0.5454645288820558, + "grad_norm": 2.0083866119384766, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7062568068504333, + "num_tokens": 127670577.0, + "step": 4967 + }, + { + "epoch": 0.5455743465846694, + "grad_norm": 2.4328393936157227, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6780939698219299, + "num_tokens": 127692460.0, + "step": 4968 + }, + { + "epoch": 0.5456841642872831, + "grad_norm": 2.0859220027923584, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.709653377532959, + "num_tokens": 127721354.0, + "step": 4969 + }, + { + "epoch": 0.5457939819898968, + "grad_norm": 2.1867282390594482, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7137935161590576, + "num_tokens": 127746849.0, + "step": 4970 + }, + { + "epoch": 0.5459037996925105, + "grad_norm": 2.577239513397217, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.721185564994812, + "num_tokens": 127766600.0, + "step": 4971 + }, + { + "epoch": 0.5460136173951241, + "grad_norm": 2.0449047088623047, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7041385173797607, + "num_tokens": 127797747.0, + "step": 4972 + }, + { + "epoch": 0.5461234350977378, + "grad_norm": 2.4234232902526855, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.70037841796875, + "num_tokens": 127820026.0, + "step": 4973 + }, + { + "epoch": 0.5462332528003514, + "grad_norm": 2.122119665145874, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7021879553794861, + "num_tokens": 127846690.0, + "step": 4974 + }, + { + "epoch": 0.546343070502965, + "grad_norm": 2.1059744358062744, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7068241834640503, + "num_tokens": 127872139.0, + "step": 4975 + }, + { + "epoch": 0.5464528882055787, + "grad_norm": 2.109964609146118, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7017973065376282, + "num_tokens": 127899797.0, + "step": 4976 + }, + { + "epoch": 0.5465627059081924, + "grad_norm": 2.2130284309387207, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6898151636123657, + "num_tokens": 127925777.0, + "step": 4977 + }, + { + "epoch": 0.5466725236108061, + "grad_norm": 2.0724287033081055, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6919413805007935, + "num_tokens": 127954845.0, + "step": 4978 + }, + { + "epoch": 0.5467823413134197, + "grad_norm": 2.0717618465423584, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6880332231521606, + "num_tokens": 127982773.0, + "step": 4979 + }, + { + "epoch": 0.5468921590160334, + "grad_norm": 2.063821792602539, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7169356942176819, + "num_tokens": 128007198.0, + "step": 4980 + }, + { + "epoch": 0.547001976718647, + "grad_norm": 2.465029716491699, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6959865093231201, + "num_tokens": 128028495.0, + "step": 4981 + }, + { + "epoch": 0.5471117944212607, + "grad_norm": 2.3578219413757324, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7186382412910461, + "num_tokens": 128049897.0, + "step": 4982 + }, + { + "epoch": 0.5472216121238743, + "grad_norm": 2.3674991130828857, + "learning_rate": 1e-06, + "loss": 1.0534, + "mean_token_accuracy": 0.6870965957641602, + "num_tokens": 128071679.0, + "step": 4983 + }, + { + "epoch": 0.5473314298264881, + "grad_norm": 2.0715980529785156, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.720501720905304, + "num_tokens": 128099467.0, + "step": 4984 + }, + { + "epoch": 0.5474412475291017, + "grad_norm": 2.5868546962738037, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7116110324859619, + "num_tokens": 128120018.0, + "step": 4985 + }, + { + "epoch": 0.5475510652317154, + "grad_norm": 2.1338257789611816, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7246711254119873, + "num_tokens": 128146266.0, + "step": 4986 + }, + { + "epoch": 0.547660882934329, + "grad_norm": 2.3803505897521973, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7113392353057861, + "num_tokens": 128169295.0, + "step": 4987 + }, + { + "epoch": 0.5477707006369427, + "grad_norm": 2.1240150928497314, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6931410431861877, + "num_tokens": 128196702.0, + "step": 4988 + }, + { + "epoch": 0.5478805183395563, + "grad_norm": 2.126537561416626, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7139027118682861, + "num_tokens": 128223648.0, + "step": 4989 + }, + { + "epoch": 0.54799033604217, + "grad_norm": 2.2454190254211426, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7178680896759033, + "num_tokens": 128249382.0, + "step": 4990 + }, + { + "epoch": 0.5481001537447837, + "grad_norm": 2.541574001312256, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6990015506744385, + "num_tokens": 128269476.0, + "step": 4991 + }, + { + "epoch": 0.5482099714473974, + "grad_norm": 2.549309730529785, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7095069885253906, + "num_tokens": 128292942.0, + "step": 4992 + }, + { + "epoch": 0.548319789150011, + "grad_norm": 2.087761402130127, + "learning_rate": 1e-06, + "loss": 1.1002, + "mean_token_accuracy": 0.6725457906723022, + "num_tokens": 128321976.0, + "step": 4993 + }, + { + "epoch": 0.5484296068526247, + "grad_norm": 2.2075717449188232, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7253177165985107, + "num_tokens": 128345733.0, + "step": 4994 + }, + { + "epoch": 0.5485394245552383, + "grad_norm": 1.892425537109375, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7074769139289856, + "num_tokens": 128377441.0, + "step": 4995 + }, + { + "epoch": 0.5486492422578519, + "grad_norm": 2.090343952178955, + "learning_rate": 1e-06, + "loss": 1.1003, + "mean_token_accuracy": 0.6690922379493713, + "num_tokens": 128407098.0, + "step": 4996 + }, + { + "epoch": 0.5487590599604656, + "grad_norm": 2.190281629562378, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7068883180618286, + "num_tokens": 128432594.0, + "step": 4997 + }, + { + "epoch": 0.5488688776630792, + "grad_norm": 2.341564416885376, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7126774191856384, + "num_tokens": 128456727.0, + "step": 4998 + }, + { + "epoch": 0.548978695365693, + "grad_norm": 2.0631296634674072, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6925544738769531, + "num_tokens": 128483979.0, + "step": 4999 + }, + { + "epoch": 0.5490885130683066, + "grad_norm": 2.630161762237549, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6930363178253174, + "num_tokens": 128502936.0, + "step": 5000 + }, + { + "epoch": 0.5491983307709203, + "grad_norm": 2.0616161823272705, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7130630016326904, + "num_tokens": 128531106.0, + "step": 5001 + }, + { + "epoch": 0.5493081484735339, + "grad_norm": 2.1046290397644043, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7014117240905762, + "num_tokens": 128558389.0, + "step": 5002 + }, + { + "epoch": 0.5494179661761476, + "grad_norm": 2.284291982650757, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7221410274505615, + "num_tokens": 128579615.0, + "step": 5003 + }, + { + "epoch": 0.5495277838787612, + "grad_norm": 1.9978376626968384, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.722392201423645, + "num_tokens": 128609739.0, + "step": 5004 + }, + { + "epoch": 0.5496376015813749, + "grad_norm": 2.199394702911377, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7295786738395691, + "num_tokens": 128634077.0, + "step": 5005 + }, + { + "epoch": 0.5497474192839886, + "grad_norm": 2.132504940032959, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7196046113967896, + "num_tokens": 128658516.0, + "step": 5006 + }, + { + "epoch": 0.5498572369866023, + "grad_norm": 2.2066140174865723, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7143200039863586, + "num_tokens": 128683075.0, + "step": 5007 + }, + { + "epoch": 0.5499670546892159, + "grad_norm": 2.400007486343384, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6952840685844421, + "num_tokens": 128706023.0, + "step": 5008 + }, + { + "epoch": 0.5500768723918296, + "grad_norm": 2.2858474254608154, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7089657783508301, + "num_tokens": 128728839.0, + "step": 5009 + }, + { + "epoch": 0.5501866900944432, + "grad_norm": 2.455716848373413, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.708027184009552, + "num_tokens": 128748939.0, + "step": 5010 + }, + { + "epoch": 0.5502965077970569, + "grad_norm": 2.337724447250366, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.725128173828125, + "num_tokens": 128769493.0, + "step": 5011 + }, + { + "epoch": 0.5504063254996705, + "grad_norm": 2.129066228866577, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6897258162498474, + "num_tokens": 128798244.0, + "step": 5012 + }, + { + "epoch": 0.5505161432022843, + "grad_norm": 2.1680097579956055, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6898858547210693, + "num_tokens": 128822884.0, + "step": 5013 + }, + { + "epoch": 0.5506259609048979, + "grad_norm": 2.1358020305633545, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7335861921310425, + "num_tokens": 128847709.0, + "step": 5014 + }, + { + "epoch": 0.5507357786075116, + "grad_norm": 2.196373224258423, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7203911542892456, + "num_tokens": 128872900.0, + "step": 5015 + }, + { + "epoch": 0.5508455963101252, + "grad_norm": 2.128199338912964, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6848273277282715, + "num_tokens": 128899382.0, + "step": 5016 + }, + { + "epoch": 0.5509554140127388, + "grad_norm": 2.4342966079711914, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7156779766082764, + "num_tokens": 128921158.0, + "step": 5017 + }, + { + "epoch": 0.5510652317153525, + "grad_norm": 2.379957675933838, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7124937176704407, + "num_tokens": 128944165.0, + "step": 5018 + }, + { + "epoch": 0.5511750494179661, + "grad_norm": 2.0796101093292236, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7038329839706421, + "num_tokens": 128971169.0, + "step": 5019 + }, + { + "epoch": 0.5512848671205799, + "grad_norm": 2.29343318939209, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7184077501296997, + "num_tokens": 128996062.0, + "step": 5020 + }, + { + "epoch": 0.5513946848231935, + "grad_norm": 2.097254991531372, + "learning_rate": 1e-06, + "loss": 1.1035, + "mean_token_accuracy": 0.6777302026748657, + "num_tokens": 129025427.0, + "step": 5021 + }, + { + "epoch": 0.5515045025258072, + "grad_norm": 2.3711581230163574, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7136902809143066, + "num_tokens": 129046667.0, + "step": 5022 + }, + { + "epoch": 0.5516143202284208, + "grad_norm": 2.427852153778076, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.731447160243988, + "num_tokens": 129067872.0, + "step": 5023 + }, + { + "epoch": 0.5517241379310345, + "grad_norm": 1.9934966564178467, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6937054395675659, + "num_tokens": 129097738.0, + "step": 5024 + }, + { + "epoch": 0.5518339556336481, + "grad_norm": 2.108654022216797, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6962918043136597, + "num_tokens": 129125199.0, + "step": 5025 + }, + { + "epoch": 0.5519437733362618, + "grad_norm": 2.3928439617156982, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7127094864845276, + "num_tokens": 129147532.0, + "step": 5026 + }, + { + "epoch": 0.5520535910388754, + "grad_norm": 2.112846851348877, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7135181427001953, + "num_tokens": 129174641.0, + "step": 5027 + }, + { + "epoch": 0.5521634087414892, + "grad_norm": 2.3470396995544434, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7123416066169739, + "num_tokens": 129199352.0, + "step": 5028 + }, + { + "epoch": 0.5522732264441028, + "grad_norm": 2.5920867919921875, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6776496171951294, + "num_tokens": 129219534.0, + "step": 5029 + }, + { + "epoch": 0.5523830441467165, + "grad_norm": 2.3914735317230225, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7002441883087158, + "num_tokens": 129241199.0, + "step": 5030 + }, + { + "epoch": 0.5524928618493301, + "grad_norm": 2.4261412620544434, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6954354643821716, + "num_tokens": 129264532.0, + "step": 5031 + }, + { + "epoch": 0.5526026795519438, + "grad_norm": 1.9750221967697144, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7109276652336121, + "num_tokens": 129293748.0, + "step": 5032 + }, + { + "epoch": 0.5527124972545574, + "grad_norm": 2.178295850753784, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6966757774353027, + "num_tokens": 129321876.0, + "step": 5033 + }, + { + "epoch": 0.552822314957171, + "grad_norm": 2.038254499435425, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7048351168632507, + "num_tokens": 129348661.0, + "step": 5034 + }, + { + "epoch": 0.5529321326597848, + "grad_norm": 2.362093448638916, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7106018662452698, + "num_tokens": 129369613.0, + "step": 5035 + }, + { + "epoch": 0.5530419503623984, + "grad_norm": 2.0959417819976807, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7089307308197021, + "num_tokens": 129398232.0, + "step": 5036 + }, + { + "epoch": 0.5531517680650121, + "grad_norm": 2.557466506958008, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7282448410987854, + "num_tokens": 129416279.0, + "step": 5037 + }, + { + "epoch": 0.5532615857676257, + "grad_norm": 2.0953707695007324, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6865032911300659, + "num_tokens": 129445105.0, + "step": 5038 + }, + { + "epoch": 0.5533714034702394, + "grad_norm": 2.057809829711914, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6989873647689819, + "num_tokens": 129474584.0, + "step": 5039 + }, + { + "epoch": 0.553481221172853, + "grad_norm": 2.1214451789855957, + "learning_rate": 1e-06, + "loss": 1.1349, + "mean_token_accuracy": 0.6630257368087769, + "num_tokens": 129503092.0, + "step": 5040 + }, + { + "epoch": 0.5535910388754667, + "grad_norm": 2.0537264347076416, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6917294263839722, + "num_tokens": 129533412.0, + "step": 5041 + }, + { + "epoch": 0.5537008565780804, + "grad_norm": 2.417354106903076, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.716950535774231, + "num_tokens": 129555193.0, + "step": 5042 + }, + { + "epoch": 0.5538106742806941, + "grad_norm": 2.190192699432373, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6773170232772827, + "num_tokens": 129580693.0, + "step": 5043 + }, + { + "epoch": 0.5539204919833077, + "grad_norm": 2.1602396965026855, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7051620483398438, + "num_tokens": 129606508.0, + "step": 5044 + }, + { + "epoch": 0.5540303096859214, + "grad_norm": 1.9474852085113525, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7031059265136719, + "num_tokens": 129637188.0, + "step": 5045 + }, + { + "epoch": 0.554140127388535, + "grad_norm": 2.3289883136749268, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7096806168556213, + "num_tokens": 129660725.0, + "step": 5046 + }, + { + "epoch": 0.5542499450911487, + "grad_norm": 2.2074670791625977, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7090752124786377, + "num_tokens": 129688484.0, + "step": 5047 + }, + { + "epoch": 0.5543597627937623, + "grad_norm": 2.1378276348114014, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7165283560752869, + "num_tokens": 129714813.0, + "step": 5048 + }, + { + "epoch": 0.5544695804963761, + "grad_norm": 2.217695713043213, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.714802622795105, + "num_tokens": 129738564.0, + "step": 5049 + }, + { + "epoch": 0.5545793981989897, + "grad_norm": 2.465590000152588, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6999146938323975, + "num_tokens": 129760916.0, + "step": 5050 + }, + { + "epoch": 0.5546892159016034, + "grad_norm": 2.596301794052124, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7044232487678528, + "num_tokens": 129779707.0, + "step": 5051 + }, + { + "epoch": 0.554799033604217, + "grad_norm": 1.958426594734192, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7195185422897339, + "num_tokens": 129809525.0, + "step": 5052 + }, + { + "epoch": 0.5549088513068307, + "grad_norm": 2.153614044189453, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.698706328868866, + "num_tokens": 129836214.0, + "step": 5053 + }, + { + "epoch": 0.5550186690094443, + "grad_norm": 2.1389899253845215, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7104092836380005, + "num_tokens": 129862221.0, + "step": 5054 + }, + { + "epoch": 0.555128486712058, + "grad_norm": 2.3816282749176025, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6949827671051025, + "num_tokens": 129887331.0, + "step": 5055 + }, + { + "epoch": 0.5552383044146716, + "grad_norm": 2.1466236114501953, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6969699859619141, + "num_tokens": 129913850.0, + "step": 5056 + }, + { + "epoch": 0.5553481221172853, + "grad_norm": 1.9896081686019897, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6973623037338257, + "num_tokens": 129945578.0, + "step": 5057 + }, + { + "epoch": 0.555457939819899, + "grad_norm": 2.2665960788726807, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6988925337791443, + "num_tokens": 129969274.0, + "step": 5058 + }, + { + "epoch": 0.5555677575225126, + "grad_norm": 2.3873510360717773, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6896436214447021, + "num_tokens": 129991900.0, + "step": 5059 + }, + { + "epoch": 0.5556775752251263, + "grad_norm": 2.092508316040039, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.687255322933197, + "num_tokens": 130019833.0, + "step": 5060 + }, + { + "epoch": 0.5557873929277399, + "grad_norm": 2.380878448486328, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6913605332374573, + "num_tokens": 130042193.0, + "step": 5061 + }, + { + "epoch": 0.5558972106303536, + "grad_norm": 2.0540661811828613, + "learning_rate": 1e-06, + "loss": 1.0743, + "mean_token_accuracy": 0.68683260679245, + "num_tokens": 130072794.0, + "step": 5062 + }, + { + "epoch": 0.5560070283329672, + "grad_norm": 2.2109789848327637, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.680006206035614, + "num_tokens": 130099857.0, + "step": 5063 + }, + { + "epoch": 0.556116846035581, + "grad_norm": 2.3111636638641357, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.703048825263977, + "num_tokens": 130124298.0, + "step": 5064 + }, + { + "epoch": 0.5562266637381946, + "grad_norm": 1.973067045211792, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6886427402496338, + "num_tokens": 130156692.0, + "step": 5065 + }, + { + "epoch": 0.5563364814408083, + "grad_norm": 2.382096767425537, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6967364549636841, + "num_tokens": 130180045.0, + "step": 5066 + }, + { + "epoch": 0.5564462991434219, + "grad_norm": 2.302626132965088, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7191257476806641, + "num_tokens": 130204354.0, + "step": 5067 + }, + { + "epoch": 0.5565561168460356, + "grad_norm": 2.1275360584259033, + "learning_rate": 1e-06, + "loss": 1.0868, + "mean_token_accuracy": 0.6803761720657349, + "num_tokens": 130235616.0, + "step": 5068 + }, + { + "epoch": 0.5566659345486492, + "grad_norm": 2.2110936641693115, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6941945552825928, + "num_tokens": 130258740.0, + "step": 5069 + }, + { + "epoch": 0.5567757522512629, + "grad_norm": 2.15899658203125, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.697685182094574, + "num_tokens": 130284811.0, + "step": 5070 + }, + { + "epoch": 0.5568855699538766, + "grad_norm": 2.508359432220459, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.736005425453186, + "num_tokens": 130303349.0, + "step": 5071 + }, + { + "epoch": 0.5569953876564903, + "grad_norm": 2.1621382236480713, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6841017007827759, + "num_tokens": 130327872.0, + "step": 5072 + }, + { + "epoch": 0.5571052053591039, + "grad_norm": 2.085568904876709, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6769427061080933, + "num_tokens": 130354651.0, + "step": 5073 + }, + { + "epoch": 0.5572150230617176, + "grad_norm": 2.0728511810302734, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6959452629089355, + "num_tokens": 130382914.0, + "step": 5074 + }, + { + "epoch": 0.5573248407643312, + "grad_norm": 2.3270516395568848, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6932598352432251, + "num_tokens": 130406481.0, + "step": 5075 + }, + { + "epoch": 0.5574346584669448, + "grad_norm": 2.3954405784606934, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.716958224773407, + "num_tokens": 130428888.0, + "step": 5076 + }, + { + "epoch": 0.5575444761695585, + "grad_norm": 2.2192223072052, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6843512654304504, + "num_tokens": 130453737.0, + "step": 5077 + }, + { + "epoch": 0.5576542938721722, + "grad_norm": 2.1478471755981445, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7190223932266235, + "num_tokens": 130480173.0, + "step": 5078 + }, + { + "epoch": 0.5577641115747859, + "grad_norm": 2.187509059906006, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6995128393173218, + "num_tokens": 130505118.0, + "step": 5079 + }, + { + "epoch": 0.5578739292773995, + "grad_norm": 2.2129361629486084, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7031711339950562, + "num_tokens": 130529889.0, + "step": 5080 + }, + { + "epoch": 0.5579837469800132, + "grad_norm": 2.200474500656128, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7136358618736267, + "num_tokens": 130555002.0, + "step": 5081 + }, + { + "epoch": 0.5580935646826268, + "grad_norm": 2.191281318664551, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.720060408115387, + "num_tokens": 130578651.0, + "step": 5082 + }, + { + "epoch": 0.5582033823852405, + "grad_norm": 2.109873056411743, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7011653780937195, + "num_tokens": 130604557.0, + "step": 5083 + }, + { + "epoch": 0.5583132000878541, + "grad_norm": 2.171510696411133, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.698642373085022, + "num_tokens": 130630090.0, + "step": 5084 + }, + { + "epoch": 0.5584230177904678, + "grad_norm": 2.1248300075531006, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7039076685905457, + "num_tokens": 130656601.0, + "step": 5085 + }, + { + "epoch": 0.5585328354930815, + "grad_norm": 2.3289082050323486, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6890652775764465, + "num_tokens": 130679736.0, + "step": 5086 + }, + { + "epoch": 0.5586426531956952, + "grad_norm": 2.0262510776519775, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7064217329025269, + "num_tokens": 130708521.0, + "step": 5087 + }, + { + "epoch": 0.5587524708983088, + "grad_norm": 2.386507749557495, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6882444620132446, + "num_tokens": 130731293.0, + "step": 5088 + }, + { + "epoch": 0.5588622886009225, + "grad_norm": 2.457118272781372, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6938411593437195, + "num_tokens": 130751944.0, + "step": 5089 + }, + { + "epoch": 0.5589721063035361, + "grad_norm": 2.1655163764953613, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6885294914245605, + "num_tokens": 130777327.0, + "step": 5090 + }, + { + "epoch": 0.5590819240061498, + "grad_norm": 2.0121076107025146, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.701030969619751, + "num_tokens": 130804040.0, + "step": 5091 + }, + { + "epoch": 0.5591917417087634, + "grad_norm": 2.2475087642669678, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.6907795667648315, + "num_tokens": 130832209.0, + "step": 5092 + }, + { + "epoch": 0.5593015594113772, + "grad_norm": 1.9989818334579468, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7100183963775635, + "num_tokens": 130859911.0, + "step": 5093 + }, + { + "epoch": 0.5594113771139908, + "grad_norm": 2.0672764778137207, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.692541778087616, + "num_tokens": 130889639.0, + "step": 5094 + }, + { + "epoch": 0.5595211948166045, + "grad_norm": 2.101682662963867, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6919589638710022, + "num_tokens": 130918072.0, + "step": 5095 + }, + { + "epoch": 0.5596310125192181, + "grad_norm": 2.0472075939178467, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7010622024536133, + "num_tokens": 130947836.0, + "step": 5096 + }, + { + "epoch": 0.5597408302218317, + "grad_norm": 1.9519234895706177, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6827530264854431, + "num_tokens": 130979090.0, + "step": 5097 + }, + { + "epoch": 0.5598506479244454, + "grad_norm": 2.293947458267212, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6923753023147583, + "num_tokens": 131003802.0, + "step": 5098 + }, + { + "epoch": 0.559960465627059, + "grad_norm": 2.1743075847625732, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7045447826385498, + "num_tokens": 131033853.0, + "step": 5099 + }, + { + "epoch": 0.5600702833296728, + "grad_norm": 2.2104430198669434, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6983579993247986, + "num_tokens": 131061349.0, + "step": 5100 + }, + { + "epoch": 0.5601801010322864, + "grad_norm": 2.4000604152679443, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7041099071502686, + "num_tokens": 131083158.0, + "step": 5101 + }, + { + "epoch": 0.5602899187349001, + "grad_norm": 2.263636827468872, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6930937767028809, + "num_tokens": 131106907.0, + "step": 5102 + }, + { + "epoch": 0.5603997364375137, + "grad_norm": 2.1766040325164795, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6974824666976929, + "num_tokens": 131134141.0, + "step": 5103 + }, + { + "epoch": 0.5605095541401274, + "grad_norm": 2.325330972671509, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6914310455322266, + "num_tokens": 131156525.0, + "step": 5104 + }, + { + "epoch": 0.560619371842741, + "grad_norm": 2.057107448577881, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6873633861541748, + "num_tokens": 131187521.0, + "step": 5105 + }, + { + "epoch": 0.5607291895453547, + "grad_norm": 2.1837387084960938, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6965164542198181, + "num_tokens": 131212653.0, + "step": 5106 + }, + { + "epoch": 0.5608390072479684, + "grad_norm": 2.2267725467681885, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.732495903968811, + "num_tokens": 131239538.0, + "step": 5107 + }, + { + "epoch": 0.5609488249505821, + "grad_norm": 2.3879411220550537, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7042213678359985, + "num_tokens": 131261556.0, + "step": 5108 + }, + { + "epoch": 0.5610586426531957, + "grad_norm": 2.2588300704956055, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6755304336547852, + "num_tokens": 131286988.0, + "step": 5109 + }, + { + "epoch": 0.5611684603558094, + "grad_norm": 2.4670324325561523, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7036945223808289, + "num_tokens": 131307391.0, + "step": 5110 + }, + { + "epoch": 0.561278278058423, + "grad_norm": 1.9707368612289429, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6940761804580688, + "num_tokens": 131335957.0, + "step": 5111 + }, + { + "epoch": 0.5613880957610367, + "grad_norm": 2.1830055713653564, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6951761245727539, + "num_tokens": 131361911.0, + "step": 5112 + }, + { + "epoch": 0.5614979134636503, + "grad_norm": 2.405911684036255, + "learning_rate": 1e-06, + "loss": 1.0892, + "mean_token_accuracy": 0.6791901588439941, + "num_tokens": 131385370.0, + "step": 5113 + }, + { + "epoch": 0.5616077311662641, + "grad_norm": 2.3658623695373535, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.706546425819397, + "num_tokens": 131406968.0, + "step": 5114 + }, + { + "epoch": 0.5617175488688777, + "grad_norm": 2.1361169815063477, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6958252191543579, + "num_tokens": 131434063.0, + "step": 5115 + }, + { + "epoch": 0.5618273665714913, + "grad_norm": 2.074162721633911, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7296476364135742, + "num_tokens": 131460696.0, + "step": 5116 + }, + { + "epoch": 0.561937184274105, + "grad_norm": 2.3740241527557373, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7275298833847046, + "num_tokens": 131484621.0, + "step": 5117 + }, + { + "epoch": 0.5620470019767186, + "grad_norm": 1.9106547832489014, + "learning_rate": 1e-06, + "loss": 1.0878, + "mean_token_accuracy": 0.6731743812561035, + "num_tokens": 131518901.0, + "step": 5118 + }, + { + "epoch": 0.5621568196793323, + "grad_norm": 2.288923978805542, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7027868032455444, + "num_tokens": 131543469.0, + "step": 5119 + }, + { + "epoch": 0.5622666373819459, + "grad_norm": 2.1007080078125, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.70296311378479, + "num_tokens": 131572484.0, + "step": 5120 + }, + { + "epoch": 0.5623764550845596, + "grad_norm": 2.0201478004455566, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.691849410533905, + "num_tokens": 131602653.0, + "step": 5121 + }, + { + "epoch": 0.5624862727871733, + "grad_norm": 2.132814407348633, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.721038818359375, + "num_tokens": 131627360.0, + "step": 5122 + }, + { + "epoch": 0.562596090489787, + "grad_norm": 1.980485200881958, + "learning_rate": 1e-06, + "loss": 1.0871, + "mean_token_accuracy": 0.6732211112976074, + "num_tokens": 131661017.0, + "step": 5123 + }, + { + "epoch": 0.5627059081924006, + "grad_norm": 2.20162034034729, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7202046513557434, + "num_tokens": 131686111.0, + "step": 5124 + }, + { + "epoch": 0.5628157258950143, + "grad_norm": 2.392451047897339, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.692865788936615, + "num_tokens": 131710796.0, + "step": 5125 + }, + { + "epoch": 0.5629255435976279, + "grad_norm": 2.086134195327759, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7012739181518555, + "num_tokens": 131738574.0, + "step": 5126 + }, + { + "epoch": 0.5630353613002416, + "grad_norm": 2.3343005180358887, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7101050615310669, + "num_tokens": 131760819.0, + "step": 5127 + }, + { + "epoch": 0.5631451790028552, + "grad_norm": 2.4826338291168213, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7169110774993896, + "num_tokens": 131781107.0, + "step": 5128 + }, + { + "epoch": 0.563254996705469, + "grad_norm": 2.1077005863189697, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7103210687637329, + "num_tokens": 131809897.0, + "step": 5129 + }, + { + "epoch": 0.5633648144080826, + "grad_norm": 2.0758473873138428, + "learning_rate": 1e-06, + "loss": 1.1111, + "mean_token_accuracy": 0.6729882955551147, + "num_tokens": 131841143.0, + "step": 5130 + }, + { + "epoch": 0.5634746321106963, + "grad_norm": 1.8596980571746826, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7225631475448608, + "num_tokens": 131874447.0, + "step": 5131 + }, + { + "epoch": 0.5635844498133099, + "grad_norm": 2.2146973609924316, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6925643682479858, + "num_tokens": 131901572.0, + "step": 5132 + }, + { + "epoch": 0.5636942675159236, + "grad_norm": 2.2931952476501465, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.705267608165741, + "num_tokens": 131924147.0, + "step": 5133 + }, + { + "epoch": 0.5638040852185372, + "grad_norm": 1.9897394180297852, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.720174252986908, + "num_tokens": 131954521.0, + "step": 5134 + }, + { + "epoch": 0.5639139029211508, + "grad_norm": 2.0672812461853027, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7049252986907959, + "num_tokens": 131982973.0, + "step": 5135 + }, + { + "epoch": 0.5640237206237646, + "grad_norm": 2.2341842651367188, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7014585137367249, + "num_tokens": 132006681.0, + "step": 5136 + }, + { + "epoch": 0.5641335383263782, + "grad_norm": 2.0908851623535156, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7072287797927856, + "num_tokens": 132034372.0, + "step": 5137 + }, + { + "epoch": 0.5642433560289919, + "grad_norm": 2.710702657699585, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7037481665611267, + "num_tokens": 132058234.0, + "step": 5138 + }, + { + "epoch": 0.5643531737316055, + "grad_norm": 2.0671169757843018, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7172513008117676, + "num_tokens": 132084532.0, + "step": 5139 + }, + { + "epoch": 0.5644629914342192, + "grad_norm": 2.1647047996520996, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6896806955337524, + "num_tokens": 132112682.0, + "step": 5140 + }, + { + "epoch": 0.5645728091368328, + "grad_norm": 2.0808300971984863, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7292141318321228, + "num_tokens": 132142170.0, + "step": 5141 + }, + { + "epoch": 0.5646826268394465, + "grad_norm": 2.2751107215881348, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6852437853813171, + "num_tokens": 132167466.0, + "step": 5142 + }, + { + "epoch": 0.5647924445420602, + "grad_norm": 2.0487143993377686, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7080599069595337, + "num_tokens": 132198228.0, + "step": 5143 + }, + { + "epoch": 0.5649022622446739, + "grad_norm": 2.351112127304077, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7212228775024414, + "num_tokens": 132219059.0, + "step": 5144 + }, + { + "epoch": 0.5650120799472875, + "grad_norm": 2.1958401203155518, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7071781158447266, + "num_tokens": 132243475.0, + "step": 5145 + }, + { + "epoch": 0.5651218976499012, + "grad_norm": 2.1374294757843018, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7033498287200928, + "num_tokens": 132271242.0, + "step": 5146 + }, + { + "epoch": 0.5652317153525148, + "grad_norm": 2.248547077178955, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6830397844314575, + "num_tokens": 132296088.0, + "step": 5147 + }, + { + "epoch": 0.5653415330551285, + "grad_norm": 2.1282289028167725, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7072065472602844, + "num_tokens": 132322437.0, + "step": 5148 + }, + { + "epoch": 0.5654513507577421, + "grad_norm": 2.067138195037842, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7194541692733765, + "num_tokens": 132350652.0, + "step": 5149 + }, + { + "epoch": 0.5655611684603558, + "grad_norm": 2.2625741958618164, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7072553634643555, + "num_tokens": 132373919.0, + "step": 5150 + }, + { + "epoch": 0.5656709861629695, + "grad_norm": 2.331528663635254, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7075763940811157, + "num_tokens": 132397767.0, + "step": 5151 + }, + { + "epoch": 0.5657808038655832, + "grad_norm": 2.1875603199005127, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7140313386917114, + "num_tokens": 132421835.0, + "step": 5152 + }, + { + "epoch": 0.5658906215681968, + "grad_norm": 2.1747233867645264, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6973423957824707, + "num_tokens": 132448966.0, + "step": 5153 + }, + { + "epoch": 0.5660004392708105, + "grad_norm": 2.0035274028778076, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7132130861282349, + "num_tokens": 132477972.0, + "step": 5154 + }, + { + "epoch": 0.5661102569734241, + "grad_norm": 2.1372599601745605, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7094945907592773, + "num_tokens": 132506163.0, + "step": 5155 + }, + { + "epoch": 0.5662200746760377, + "grad_norm": 2.1994471549987793, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6872334480285645, + "num_tokens": 132533043.0, + "step": 5156 + }, + { + "epoch": 0.5663298923786514, + "grad_norm": 2.239696979522705, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7194223403930664, + "num_tokens": 132559382.0, + "step": 5157 + }, + { + "epoch": 0.5664397100812651, + "grad_norm": 2.130737543106079, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7096954584121704, + "num_tokens": 132586232.0, + "step": 5158 + }, + { + "epoch": 0.5665495277838788, + "grad_norm": 2.680041790008545, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6998695135116577, + "num_tokens": 132604225.0, + "step": 5159 + }, + { + "epoch": 0.5666593454864924, + "grad_norm": 2.2478678226470947, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7282475829124451, + "num_tokens": 132628506.0, + "step": 5160 + }, + { + "epoch": 0.5667691631891061, + "grad_norm": 2.6995904445648193, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7363274097442627, + "num_tokens": 132644485.0, + "step": 5161 + }, + { + "epoch": 0.5668789808917197, + "grad_norm": 2.257373571395874, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7199069261550903, + "num_tokens": 132667125.0, + "step": 5162 + }, + { + "epoch": 0.5669887985943334, + "grad_norm": 1.9408442974090576, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6998510360717773, + "num_tokens": 132698209.0, + "step": 5163 + }, + { + "epoch": 0.567098616296947, + "grad_norm": 1.9748826026916504, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7145432233810425, + "num_tokens": 132728553.0, + "step": 5164 + }, + { + "epoch": 0.5672084339995608, + "grad_norm": 2.3617186546325684, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6998817324638367, + "num_tokens": 132750440.0, + "step": 5165 + }, + { + "epoch": 0.5673182517021744, + "grad_norm": 2.068161964416504, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7000489234924316, + "num_tokens": 132781509.0, + "step": 5166 + }, + { + "epoch": 0.5674280694047881, + "grad_norm": 2.0876991748809814, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6976301670074463, + "num_tokens": 132810461.0, + "step": 5167 + }, + { + "epoch": 0.5675378871074017, + "grad_norm": 2.4925742149353027, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7182432413101196, + "num_tokens": 132829904.0, + "step": 5168 + }, + { + "epoch": 0.5676477048100154, + "grad_norm": 1.9095498323440552, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6927222013473511, + "num_tokens": 132861080.0, + "step": 5169 + }, + { + "epoch": 0.567757522512629, + "grad_norm": 2.520329236984253, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7095850706100464, + "num_tokens": 132881429.0, + "step": 5170 + }, + { + "epoch": 0.5678673402152427, + "grad_norm": 2.1554620265960693, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7165473699569702, + "num_tokens": 132907366.0, + "step": 5171 + }, + { + "epoch": 0.5679771579178564, + "grad_norm": 2.476844072341919, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6980487704277039, + "num_tokens": 132929167.0, + "step": 5172 + }, + { + "epoch": 0.5680869756204701, + "grad_norm": 2.0899178981781006, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7038092613220215, + "num_tokens": 132956936.0, + "step": 5173 + }, + { + "epoch": 0.5681967933230837, + "grad_norm": 2.108130931854248, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7001914978027344, + "num_tokens": 132986306.0, + "step": 5174 + }, + { + "epoch": 0.5683066110256974, + "grad_norm": 2.447765350341797, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7234519720077515, + "num_tokens": 133007309.0, + "step": 5175 + }, + { + "epoch": 0.568416428728311, + "grad_norm": 2.11769437789917, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6951937675476074, + "num_tokens": 133035917.0, + "step": 5176 + }, + { + "epoch": 0.5685262464309246, + "grad_norm": 2.510410785675049, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7094571590423584, + "num_tokens": 133059245.0, + "step": 5177 + }, + { + "epoch": 0.5686360641335383, + "grad_norm": 1.8482015132904053, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6873193383216858, + "num_tokens": 133093973.0, + "step": 5178 + }, + { + "epoch": 0.5687458818361519, + "grad_norm": 2.1621580123901367, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.7008355855941772, + "num_tokens": 133120850.0, + "step": 5179 + }, + { + "epoch": 0.5688556995387657, + "grad_norm": 2.0430850982666016, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6939610838890076, + "num_tokens": 133149954.0, + "step": 5180 + }, + { + "epoch": 0.5689655172413793, + "grad_norm": 2.137065887451172, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6918742656707764, + "num_tokens": 133177497.0, + "step": 5181 + }, + { + "epoch": 0.569075334943993, + "grad_norm": 2.1338460445404053, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.685551643371582, + "num_tokens": 133205927.0, + "step": 5182 + }, + { + "epoch": 0.5691851526466066, + "grad_norm": 1.8987641334533691, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7032314538955688, + "num_tokens": 133238545.0, + "step": 5183 + }, + { + "epoch": 0.5692949703492203, + "grad_norm": 2.3593497276306152, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.700136125087738, + "num_tokens": 133263117.0, + "step": 5184 + }, + { + "epoch": 0.5694047880518339, + "grad_norm": 2.0087890625, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7009465098381042, + "num_tokens": 133293216.0, + "step": 5185 + }, + { + "epoch": 0.5695146057544476, + "grad_norm": 2.075082540512085, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7286490201950073, + "num_tokens": 133320024.0, + "step": 5186 + }, + { + "epoch": 0.5696244234570613, + "grad_norm": 2.3118958473205566, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6944650411605835, + "num_tokens": 133344434.0, + "step": 5187 + }, + { + "epoch": 0.569734241159675, + "grad_norm": 2.154782295227051, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7300060391426086, + "num_tokens": 133370575.0, + "step": 5188 + }, + { + "epoch": 0.5698440588622886, + "grad_norm": 2.5395655632019043, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7218373417854309, + "num_tokens": 133390815.0, + "step": 5189 + }, + { + "epoch": 0.5699538765649023, + "grad_norm": 2.2072582244873047, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.71045982837677, + "num_tokens": 133415442.0, + "step": 5190 + }, + { + "epoch": 0.5700636942675159, + "grad_norm": 2.0742745399475098, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6894148588180542, + "num_tokens": 133443829.0, + "step": 5191 + }, + { + "epoch": 0.5701735119701296, + "grad_norm": 2.017717123031616, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6959995031356812, + "num_tokens": 133471796.0, + "step": 5192 + }, + { + "epoch": 0.5702833296727432, + "grad_norm": 2.295865535736084, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6999038457870483, + "num_tokens": 133495298.0, + "step": 5193 + }, + { + "epoch": 0.570393147375357, + "grad_norm": 2.4429376125335693, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7269101738929749, + "num_tokens": 133514715.0, + "step": 5194 + }, + { + "epoch": 0.5705029650779706, + "grad_norm": 1.8077480792999268, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6885137557983398, + "num_tokens": 133553011.0, + "step": 5195 + }, + { + "epoch": 0.5706127827805842, + "grad_norm": 2.069812536239624, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7069491744041443, + "num_tokens": 133579686.0, + "step": 5196 + }, + { + "epoch": 0.5707226004831979, + "grad_norm": 2.1555542945861816, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6969851851463318, + "num_tokens": 133605138.0, + "step": 5197 + }, + { + "epoch": 0.5708324181858115, + "grad_norm": 2.3321304321289062, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7004455327987671, + "num_tokens": 133626880.0, + "step": 5198 + }, + { + "epoch": 0.5709422358884252, + "grad_norm": 2.229848861694336, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6992029547691345, + "num_tokens": 133651898.0, + "step": 5199 + }, + { + "epoch": 0.5710520535910388, + "grad_norm": 2.2074649333953857, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7182215452194214, + "num_tokens": 133676363.0, + "step": 5200 + }, + { + "epoch": 0.5711618712936526, + "grad_norm": 2.192105531692505, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.69132000207901, + "num_tokens": 133704454.0, + "step": 5201 + }, + { + "epoch": 0.5712716889962662, + "grad_norm": 1.973426103591919, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6995817422866821, + "num_tokens": 133736813.0, + "step": 5202 + }, + { + "epoch": 0.5713815066988799, + "grad_norm": 2.1289708614349365, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.709185004234314, + "num_tokens": 133761863.0, + "step": 5203 + }, + { + "epoch": 0.5714913244014935, + "grad_norm": 2.104933738708496, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7020571231842041, + "num_tokens": 133788391.0, + "step": 5204 + }, + { + "epoch": 0.5716011421041072, + "grad_norm": 2.595059394836426, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7128853797912598, + "num_tokens": 133807925.0, + "step": 5205 + }, + { + "epoch": 0.5717109598067208, + "grad_norm": 2.226252555847168, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6914331316947937, + "num_tokens": 133834453.0, + "step": 5206 + }, + { + "epoch": 0.5718207775093345, + "grad_norm": 2.0656979084014893, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7319349050521851, + "num_tokens": 133861189.0, + "step": 5207 + }, + { + "epoch": 0.5719305952119481, + "grad_norm": 2.3241994380950928, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7009363174438477, + "num_tokens": 133884402.0, + "step": 5208 + }, + { + "epoch": 0.5720404129145619, + "grad_norm": 2.197946786880493, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.6855249404907227, + "num_tokens": 133910617.0, + "step": 5209 + }, + { + "epoch": 0.5721502306171755, + "grad_norm": 2.154982566833496, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6968898773193359, + "num_tokens": 133936888.0, + "step": 5210 + }, + { + "epoch": 0.5722600483197892, + "grad_norm": 2.463852882385254, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6992605924606323, + "num_tokens": 133958699.0, + "step": 5211 + }, + { + "epoch": 0.5723698660224028, + "grad_norm": 2.4039487838745117, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7112432718276978, + "num_tokens": 133979998.0, + "step": 5212 + }, + { + "epoch": 0.5724796837250165, + "grad_norm": 2.054752826690674, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6988747119903564, + "num_tokens": 134010826.0, + "step": 5213 + }, + { + "epoch": 0.5725895014276301, + "grad_norm": 2.0626907348632812, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7149761915206909, + "num_tokens": 134037442.0, + "step": 5214 + }, + { + "epoch": 0.5726993191302437, + "grad_norm": 2.2888734340667725, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.693210780620575, + "num_tokens": 134062049.0, + "step": 5215 + }, + { + "epoch": 0.5728091368328575, + "grad_norm": 2.0921103954315186, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6973788142204285, + "num_tokens": 134089345.0, + "step": 5216 + }, + { + "epoch": 0.5729189545354711, + "grad_norm": 2.2307989597320557, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.68678879737854, + "num_tokens": 134115729.0, + "step": 5217 + }, + { + "epoch": 0.5730287722380848, + "grad_norm": 2.076413154602051, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.690830647945404, + "num_tokens": 134146208.0, + "step": 5218 + }, + { + "epoch": 0.5731385899406984, + "grad_norm": 2.317570447921753, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.7026517391204834, + "num_tokens": 134169805.0, + "step": 5219 + }, + { + "epoch": 0.5732484076433121, + "grad_norm": 2.4506072998046875, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7214797735214233, + "num_tokens": 134190274.0, + "step": 5220 + }, + { + "epoch": 0.5733582253459257, + "grad_norm": 2.2045414447784424, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6994750499725342, + "num_tokens": 134215954.0, + "step": 5221 + }, + { + "epoch": 0.5734680430485394, + "grad_norm": 2.2079708576202393, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6972591876983643, + "num_tokens": 134241861.0, + "step": 5222 + }, + { + "epoch": 0.5735778607511531, + "grad_norm": 2.4633920192718506, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7140121459960938, + "num_tokens": 134263313.0, + "step": 5223 + }, + { + "epoch": 0.5736876784537668, + "grad_norm": 2.1436610221862793, + "learning_rate": 1e-06, + "loss": 1.0572, + "mean_token_accuracy": 0.6891424655914307, + "num_tokens": 134292594.0, + "step": 5224 + }, + { + "epoch": 0.5737974961563804, + "grad_norm": 2.0182242393493652, + "learning_rate": 1e-06, + "loss": 1.0784, + "mean_token_accuracy": 0.67633056640625, + "num_tokens": 134323229.0, + "step": 5225 + }, + { + "epoch": 0.5739073138589941, + "grad_norm": 2.1409554481506348, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7044958472251892, + "num_tokens": 134351196.0, + "step": 5226 + }, + { + "epoch": 0.5740171315616077, + "grad_norm": 2.3698184490203857, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7148762941360474, + "num_tokens": 134373215.0, + "step": 5227 + }, + { + "epoch": 0.5741269492642214, + "grad_norm": 2.0933728218078613, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6922804713249207, + "num_tokens": 134402084.0, + "step": 5228 + }, + { + "epoch": 0.574236766966835, + "grad_norm": 2.4826245307922363, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7059068083763123, + "num_tokens": 134422741.0, + "step": 5229 + }, + { + "epoch": 0.5743465846694488, + "grad_norm": 2.0105578899383545, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.6866059899330139, + "num_tokens": 134455277.0, + "step": 5230 + }, + { + "epoch": 0.5744564023720624, + "grad_norm": 2.2475318908691406, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7077566981315613, + "num_tokens": 134480613.0, + "step": 5231 + }, + { + "epoch": 0.5745662200746761, + "grad_norm": 2.40163254737854, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7147831916809082, + "num_tokens": 134501167.0, + "step": 5232 + }, + { + "epoch": 0.5746760377772897, + "grad_norm": 2.296408176422119, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6834633350372314, + "num_tokens": 134525182.0, + "step": 5233 + }, + { + "epoch": 0.5747858554799034, + "grad_norm": 2.0118181705474854, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7056477069854736, + "num_tokens": 134555827.0, + "step": 5234 + }, + { + "epoch": 0.574895673182517, + "grad_norm": 2.26782488822937, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7119182348251343, + "num_tokens": 134579792.0, + "step": 5235 + }, + { + "epoch": 0.5750054908851306, + "grad_norm": 2.006471872329712, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7148004770278931, + "num_tokens": 134608334.0, + "step": 5236 + }, + { + "epoch": 0.5751153085877443, + "grad_norm": 2.50675368309021, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7135794162750244, + "num_tokens": 134627768.0, + "step": 5237 + }, + { + "epoch": 0.575225126290358, + "grad_norm": 1.9869672060012817, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7072102427482605, + "num_tokens": 134657609.0, + "step": 5238 + }, + { + "epoch": 0.5753349439929717, + "grad_norm": 2.2029430866241455, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6936423778533936, + "num_tokens": 134683272.0, + "step": 5239 + }, + { + "epoch": 0.5754447616955853, + "grad_norm": 2.0060853958129883, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.698698878288269, + "num_tokens": 134714598.0, + "step": 5240 + }, + { + "epoch": 0.575554579398199, + "grad_norm": 2.4371378421783447, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7071545720100403, + "num_tokens": 134736019.0, + "step": 5241 + }, + { + "epoch": 0.5756643971008126, + "grad_norm": 1.8792057037353516, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6872319579124451, + "num_tokens": 134770197.0, + "step": 5242 + }, + { + "epoch": 0.5757742148034263, + "grad_norm": 2.395735025405884, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.7002271413803101, + "num_tokens": 134794888.0, + "step": 5243 + }, + { + "epoch": 0.5758840325060399, + "grad_norm": 2.475877523422241, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7239090204238892, + "num_tokens": 134814477.0, + "step": 5244 + }, + { + "epoch": 0.5759938502086537, + "grad_norm": 2.3638243675231934, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7102169990539551, + "num_tokens": 134836908.0, + "step": 5245 + }, + { + "epoch": 0.5761036679112673, + "grad_norm": 2.234325647354126, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7254450917243958, + "num_tokens": 134860352.0, + "step": 5246 + }, + { + "epoch": 0.576213485613881, + "grad_norm": 2.2669901847839355, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7046946287155151, + "num_tokens": 134884554.0, + "step": 5247 + }, + { + "epoch": 0.5763233033164946, + "grad_norm": 1.9594695568084717, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.700471043586731, + "num_tokens": 134913928.0, + "step": 5248 + }, + { + "epoch": 0.5764331210191083, + "grad_norm": 1.9895128011703491, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6841222643852234, + "num_tokens": 134947678.0, + "step": 5249 + }, + { + "epoch": 0.5765429387217219, + "grad_norm": 2.382134199142456, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7039667367935181, + "num_tokens": 134971550.0, + "step": 5250 + }, + { + "epoch": 0.5766527564243356, + "grad_norm": 2.172654628753662, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7015098929405212, + "num_tokens": 134996348.0, + "step": 5251 + }, + { + "epoch": 0.5767625741269493, + "grad_norm": 2.2129058837890625, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6994506120681763, + "num_tokens": 135019912.0, + "step": 5252 + }, + { + "epoch": 0.576872391829563, + "grad_norm": 2.098403215408325, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6954472064971924, + "num_tokens": 135047760.0, + "step": 5253 + }, + { + "epoch": 0.5769822095321766, + "grad_norm": 2.3864314556121826, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.7044543027877808, + "num_tokens": 135072112.0, + "step": 5254 + }, + { + "epoch": 0.5770920272347903, + "grad_norm": 2.3338510990142822, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6889567971229553, + "num_tokens": 135095441.0, + "step": 5255 + }, + { + "epoch": 0.5772018449374039, + "grad_norm": 2.1673262119293213, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.706446647644043, + "num_tokens": 135121603.0, + "step": 5256 + }, + { + "epoch": 0.5773116626400175, + "grad_norm": 2.5389506816864014, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7202059030532837, + "num_tokens": 135143035.0, + "step": 5257 + }, + { + "epoch": 0.5774214803426312, + "grad_norm": 2.2867088317871094, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6964610815048218, + "num_tokens": 135168579.0, + "step": 5258 + }, + { + "epoch": 0.577531298045245, + "grad_norm": 2.1515274047851562, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7225208282470703, + "num_tokens": 135194655.0, + "step": 5259 + }, + { + "epoch": 0.5776411157478586, + "grad_norm": 1.9623805284500122, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7163307666778564, + "num_tokens": 135224624.0, + "step": 5260 + }, + { + "epoch": 0.5777509334504722, + "grad_norm": 1.9875128269195557, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6959626078605652, + "num_tokens": 135254940.0, + "step": 5261 + }, + { + "epoch": 0.5778607511530859, + "grad_norm": 2.544114589691162, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7152799367904663, + "num_tokens": 135273896.0, + "step": 5262 + }, + { + "epoch": 0.5779705688556995, + "grad_norm": 2.1340346336364746, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.715596616268158, + "num_tokens": 135300371.0, + "step": 5263 + }, + { + "epoch": 0.5780803865583132, + "grad_norm": 2.3474698066711426, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7041874527931213, + "num_tokens": 135322950.0, + "step": 5264 + }, + { + "epoch": 0.5781902042609268, + "grad_norm": 2.6270976066589355, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7051270008087158, + "num_tokens": 135342271.0, + "step": 5265 + }, + { + "epoch": 0.5783000219635406, + "grad_norm": 2.1277666091918945, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7033761739730835, + "num_tokens": 135370970.0, + "step": 5266 + }, + { + "epoch": 0.5784098396661542, + "grad_norm": 2.294760227203369, + "learning_rate": 1e-06, + "loss": 1.1205, + "mean_token_accuracy": 0.6755287647247314, + "num_tokens": 135396509.0, + "step": 5267 + }, + { + "epoch": 0.5785196573687679, + "grad_norm": 1.9560877084732056, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7037493586540222, + "num_tokens": 135428585.0, + "step": 5268 + }, + { + "epoch": 0.5786294750713815, + "grad_norm": 2.6101019382476807, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6864479780197144, + "num_tokens": 135449623.0, + "step": 5269 + }, + { + "epoch": 0.5787392927739952, + "grad_norm": 2.323688507080078, + "learning_rate": 1e-06, + "loss": 1.0695, + "mean_token_accuracy": 0.6827727556228638, + "num_tokens": 135473325.0, + "step": 5270 + }, + { + "epoch": 0.5788491104766088, + "grad_norm": 2.1470682621002197, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7018536329269409, + "num_tokens": 135500240.0, + "step": 5271 + }, + { + "epoch": 0.5789589281792225, + "grad_norm": 2.2614822387695312, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7093607187271118, + "num_tokens": 135525231.0, + "step": 5272 + }, + { + "epoch": 0.5790687458818361, + "grad_norm": 1.9888848066329956, + "learning_rate": 1e-06, + "loss": 1.0617, + "mean_token_accuracy": 0.6835367679595947, + "num_tokens": 135555264.0, + "step": 5273 + }, + { + "epoch": 0.5791785635844499, + "grad_norm": 2.3707382678985596, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7109453678131104, + "num_tokens": 135577654.0, + "step": 5274 + }, + { + "epoch": 0.5792883812870635, + "grad_norm": 2.374068260192871, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7309660911560059, + "num_tokens": 135598956.0, + "step": 5275 + }, + { + "epoch": 0.5793981989896771, + "grad_norm": 2.398035764694214, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7175350785255432, + "num_tokens": 135619697.0, + "step": 5276 + }, + { + "epoch": 0.5795080166922908, + "grad_norm": 2.3246259689331055, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6941491961479187, + "num_tokens": 135642928.0, + "step": 5277 + }, + { + "epoch": 0.5796178343949044, + "grad_norm": 2.3583056926727295, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7165483236312866, + "num_tokens": 135664870.0, + "step": 5278 + }, + { + "epoch": 0.5797276520975181, + "grad_norm": 2.063040018081665, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6939167976379395, + "num_tokens": 135693156.0, + "step": 5279 + }, + { + "epoch": 0.5798374698001317, + "grad_norm": 2.425467014312744, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.697638750076294, + "num_tokens": 135715621.0, + "step": 5280 + }, + { + "epoch": 0.5799472875027455, + "grad_norm": 2.138761520385742, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7082366347312927, + "num_tokens": 135741688.0, + "step": 5281 + }, + { + "epoch": 0.5800571052053591, + "grad_norm": 2.254181146621704, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.727988600730896, + "num_tokens": 135764018.0, + "step": 5282 + }, + { + "epoch": 0.5801669229079728, + "grad_norm": 2.5606753826141357, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.710067629814148, + "num_tokens": 135784411.0, + "step": 5283 + }, + { + "epoch": 0.5802767406105864, + "grad_norm": 2.262909412384033, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7063171863555908, + "num_tokens": 135808650.0, + "step": 5284 + }, + { + "epoch": 0.5803865583132001, + "grad_norm": 2.143934488296509, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.692641019821167, + "num_tokens": 135835859.0, + "step": 5285 + }, + { + "epoch": 0.5804963760158137, + "grad_norm": 2.1408166885375977, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7113150954246521, + "num_tokens": 135861495.0, + "step": 5286 + }, + { + "epoch": 0.5806061937184274, + "grad_norm": 2.2960894107818604, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6928730010986328, + "num_tokens": 135886885.0, + "step": 5287 + }, + { + "epoch": 0.5807160114210411, + "grad_norm": 2.115446090698242, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7020163536071777, + "num_tokens": 135913985.0, + "step": 5288 + }, + { + "epoch": 0.5808258291236548, + "grad_norm": 2.3548381328582764, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7283948659896851, + "num_tokens": 135938292.0, + "step": 5289 + }, + { + "epoch": 0.5809356468262684, + "grad_norm": 2.0563950538635254, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6728092432022095, + "num_tokens": 135966918.0, + "step": 5290 + }, + { + "epoch": 0.5810454645288821, + "grad_norm": 2.1101725101470947, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.696753740310669, + "num_tokens": 135994094.0, + "step": 5291 + }, + { + "epoch": 0.5811552822314957, + "grad_norm": 1.9386707544326782, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.731895923614502, + "num_tokens": 136022974.0, + "step": 5292 + }, + { + "epoch": 0.5812650999341094, + "grad_norm": 2.3617258071899414, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.709067702293396, + "num_tokens": 136045902.0, + "step": 5293 + }, + { + "epoch": 0.581374917636723, + "grad_norm": 2.0911242961883545, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7230409979820251, + "num_tokens": 136072585.0, + "step": 5294 + }, + { + "epoch": 0.5814847353393368, + "grad_norm": 2.140777826309204, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6981576681137085, + "num_tokens": 136101011.0, + "step": 5295 + }, + { + "epoch": 0.5815945530419504, + "grad_norm": 2.069535970687866, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7042256593704224, + "num_tokens": 136127242.0, + "step": 5296 + }, + { + "epoch": 0.581704370744564, + "grad_norm": 2.069054126739502, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7125552892684937, + "num_tokens": 136155409.0, + "step": 5297 + }, + { + "epoch": 0.5818141884471777, + "grad_norm": 2.3974874019622803, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7156614065170288, + "num_tokens": 136177298.0, + "step": 5298 + }, + { + "epoch": 0.5819240061497913, + "grad_norm": 2.2698724269866943, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6949012279510498, + "num_tokens": 136202915.0, + "step": 5299 + }, + { + "epoch": 0.582033823852405, + "grad_norm": 2.1067955493927, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6930897235870361, + "num_tokens": 136231187.0, + "step": 5300 + }, + { + "epoch": 0.5821436415550186, + "grad_norm": 1.9334203004837036, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7311375141143799, + "num_tokens": 136260292.0, + "step": 5301 + }, + { + "epoch": 0.5822534592576323, + "grad_norm": 2.0019161701202393, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6962857842445374, + "num_tokens": 136289356.0, + "step": 5302 + }, + { + "epoch": 0.582363276960246, + "grad_norm": 2.155020236968994, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6969937086105347, + "num_tokens": 136315197.0, + "step": 5303 + }, + { + "epoch": 0.5824730946628597, + "grad_norm": 2.103433609008789, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.729040265083313, + "num_tokens": 136340907.0, + "step": 5304 + }, + { + "epoch": 0.5825829123654733, + "grad_norm": 2.140845537185669, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6861826777458191, + "num_tokens": 136369243.0, + "step": 5305 + }, + { + "epoch": 0.582692730068087, + "grad_norm": 2.205428123474121, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6980424523353577, + "num_tokens": 136393918.0, + "step": 5306 + }, + { + "epoch": 0.5828025477707006, + "grad_norm": 2.1216013431549072, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6788161993026733, + "num_tokens": 136420386.0, + "step": 5307 + }, + { + "epoch": 0.5829123654733143, + "grad_norm": 2.0326340198516846, + "learning_rate": 1e-06, + "loss": 1.0849, + "mean_token_accuracy": 0.6759843826293945, + "num_tokens": 136450758.0, + "step": 5308 + }, + { + "epoch": 0.5830221831759279, + "grad_norm": 2.112748384475708, + "learning_rate": 1e-06, + "loss": 1.0833, + "mean_token_accuracy": 0.6789644956588745, + "num_tokens": 136479248.0, + "step": 5309 + }, + { + "epoch": 0.5831320008785417, + "grad_norm": 2.409289598464966, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7157273888587952, + "num_tokens": 136499598.0, + "step": 5310 + }, + { + "epoch": 0.5832418185811553, + "grad_norm": 2.1124191284179688, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.701486349105835, + "num_tokens": 136525947.0, + "step": 5311 + }, + { + "epoch": 0.583351636283769, + "grad_norm": 2.0433828830718994, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7049245834350586, + "num_tokens": 136554956.0, + "step": 5312 + }, + { + "epoch": 0.5834614539863826, + "grad_norm": 2.362241506576538, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.707420289516449, + "num_tokens": 136577694.0, + "step": 5313 + }, + { + "epoch": 0.5835712716889963, + "grad_norm": 2.1756529808044434, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.699539303779602, + "num_tokens": 136603395.0, + "step": 5314 + }, + { + "epoch": 0.5836810893916099, + "grad_norm": 2.03406023979187, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6881468296051025, + "num_tokens": 136631344.0, + "step": 5315 + }, + { + "epoch": 0.5837909070942235, + "grad_norm": 2.4572336673736572, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6821285486221313, + "num_tokens": 136652620.0, + "step": 5316 + }, + { + "epoch": 0.5839007247968373, + "grad_norm": 2.2609784603118896, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7085700035095215, + "num_tokens": 136675397.0, + "step": 5317 + }, + { + "epoch": 0.584010542499451, + "grad_norm": 2.205965757369995, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7046908736228943, + "num_tokens": 136700458.0, + "step": 5318 + }, + { + "epoch": 0.5841203602020646, + "grad_norm": 2.2089834213256836, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7229129076004028, + "num_tokens": 136726392.0, + "step": 5319 + }, + { + "epoch": 0.5842301779046782, + "grad_norm": 2.2288105487823486, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7017898559570312, + "num_tokens": 136749730.0, + "step": 5320 + }, + { + "epoch": 0.5843399956072919, + "grad_norm": 2.0057172775268555, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6966770887374878, + "num_tokens": 136779075.0, + "step": 5321 + }, + { + "epoch": 0.5844498133099055, + "grad_norm": 2.3217670917510986, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.692941427230835, + "num_tokens": 136804919.0, + "step": 5322 + }, + { + "epoch": 0.5845596310125192, + "grad_norm": 1.8150385618209839, + "learning_rate": 1e-06, + "loss": 1.0743, + "mean_token_accuracy": 0.6805499196052551, + "num_tokens": 136840353.0, + "step": 5323 + }, + { + "epoch": 0.5846694487151329, + "grad_norm": 1.9291263818740845, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6867613792419434, + "num_tokens": 136872526.0, + "step": 5324 + }, + { + "epoch": 0.5847792664177466, + "grad_norm": 1.9578661918640137, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6920098662376404, + "num_tokens": 136904409.0, + "step": 5325 + }, + { + "epoch": 0.5848890841203602, + "grad_norm": 2.3334081172943115, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7150534987449646, + "num_tokens": 136926768.0, + "step": 5326 + }, + { + "epoch": 0.5849989018229739, + "grad_norm": 2.3350777626037598, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7100425958633423, + "num_tokens": 136950237.0, + "step": 5327 + }, + { + "epoch": 0.5851087195255875, + "grad_norm": 2.2675423622131348, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7102535367012024, + "num_tokens": 136973967.0, + "step": 5328 + }, + { + "epoch": 0.5852185372282012, + "grad_norm": 2.3983914852142334, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7262950539588928, + "num_tokens": 136994510.0, + "step": 5329 + }, + { + "epoch": 0.5853283549308148, + "grad_norm": 2.421330690383911, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6873228549957275, + "num_tokens": 137017363.0, + "step": 5330 + }, + { + "epoch": 0.5854381726334285, + "grad_norm": 2.241267204284668, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7268282175064087, + "num_tokens": 137040244.0, + "step": 5331 + }, + { + "epoch": 0.5855479903360422, + "grad_norm": 2.2815170288085938, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7381749749183655, + "num_tokens": 137063710.0, + "step": 5332 + }, + { + "epoch": 0.5856578080386559, + "grad_norm": 1.9815633296966553, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6982458233833313, + "num_tokens": 137093659.0, + "step": 5333 + }, + { + "epoch": 0.5857676257412695, + "grad_norm": 2.355634927749634, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7183383703231812, + "num_tokens": 137114516.0, + "step": 5334 + }, + { + "epoch": 0.5858774434438832, + "grad_norm": 2.2258002758026123, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6937353610992432, + "num_tokens": 137142021.0, + "step": 5335 + }, + { + "epoch": 0.5859872611464968, + "grad_norm": 2.245605945587158, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6844848394393921, + "num_tokens": 137168555.0, + "step": 5336 + }, + { + "epoch": 0.5860970788491104, + "grad_norm": 2.005878448486328, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6787208318710327, + "num_tokens": 137199045.0, + "step": 5337 + }, + { + "epoch": 0.5862068965517241, + "grad_norm": 2.3082339763641357, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7331153154373169, + "num_tokens": 137222382.0, + "step": 5338 + }, + { + "epoch": 0.5863167142543378, + "grad_norm": 2.1754038333892822, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6970849633216858, + "num_tokens": 137249451.0, + "step": 5339 + }, + { + "epoch": 0.5864265319569515, + "grad_norm": 1.9243309497833252, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7075361609458923, + "num_tokens": 137278684.0, + "step": 5340 + }, + { + "epoch": 0.5865363496595651, + "grad_norm": 2.1986083984375, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7037973403930664, + "num_tokens": 137304446.0, + "step": 5341 + }, + { + "epoch": 0.5866461673621788, + "grad_norm": 2.3072288036346436, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6897525191307068, + "num_tokens": 137326969.0, + "step": 5342 + }, + { + "epoch": 0.5867559850647924, + "grad_norm": 2.1315555572509766, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6998993158340454, + "num_tokens": 137355837.0, + "step": 5343 + }, + { + "epoch": 0.5868658027674061, + "grad_norm": 2.044905424118042, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6937857866287231, + "num_tokens": 137387094.0, + "step": 5344 + }, + { + "epoch": 0.5869756204700197, + "grad_norm": 2.1549489498138428, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7321670055389404, + "num_tokens": 137412295.0, + "step": 5345 + }, + { + "epoch": 0.5870854381726335, + "grad_norm": 2.3377294540405273, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7232632637023926, + "num_tokens": 137434048.0, + "step": 5346 + }, + { + "epoch": 0.5871952558752471, + "grad_norm": 2.4387402534484863, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7063745260238647, + "num_tokens": 137456077.0, + "step": 5347 + }, + { + "epoch": 0.5873050735778608, + "grad_norm": 2.0932230949401855, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6784524917602539, + "num_tokens": 137486784.0, + "step": 5348 + }, + { + "epoch": 0.5874148912804744, + "grad_norm": 2.1814849376678467, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6970131993293762, + "num_tokens": 137515881.0, + "step": 5349 + }, + { + "epoch": 0.5875247089830881, + "grad_norm": 1.9897851943969727, + "learning_rate": 1e-06, + "loss": 1.1094, + "mean_token_accuracy": 0.667238175868988, + "num_tokens": 137546155.0, + "step": 5350 + }, + { + "epoch": 0.5876345266857017, + "grad_norm": 2.0395357608795166, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.712958812713623, + "num_tokens": 137574477.0, + "step": 5351 + }, + { + "epoch": 0.5877443443883154, + "grad_norm": 2.3210649490356445, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7180792689323425, + "num_tokens": 137595882.0, + "step": 5352 + }, + { + "epoch": 0.5878541620909291, + "grad_norm": 2.1408488750457764, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6933932304382324, + "num_tokens": 137621774.0, + "step": 5353 + }, + { + "epoch": 0.5879639797935428, + "grad_norm": 2.0797832012176514, + "learning_rate": 1e-06, + "loss": 1.0702, + "mean_token_accuracy": 0.6765720844268799, + "num_tokens": 137649842.0, + "step": 5354 + }, + { + "epoch": 0.5880737974961564, + "grad_norm": 2.1403539180755615, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7038723826408386, + "num_tokens": 137675135.0, + "step": 5355 + }, + { + "epoch": 0.58818361519877, + "grad_norm": 2.2010068893432617, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6966724991798401, + "num_tokens": 137699788.0, + "step": 5356 + }, + { + "epoch": 0.5882934329013837, + "grad_norm": 2.6637473106384277, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7204597592353821, + "num_tokens": 137719064.0, + "step": 5357 + }, + { + "epoch": 0.5884032506039973, + "grad_norm": 2.324985980987549, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.7048858404159546, + "num_tokens": 137743260.0, + "step": 5358 + }, + { + "epoch": 0.588513068306611, + "grad_norm": 2.604179620742798, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7039430141448975, + "num_tokens": 137763609.0, + "step": 5359 + }, + { + "epoch": 0.5886228860092246, + "grad_norm": 2.0601718425750732, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6963117122650146, + "num_tokens": 137792442.0, + "step": 5360 + }, + { + "epoch": 0.5887327037118384, + "grad_norm": 1.9461276531219482, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.747054934501648, + "num_tokens": 137821766.0, + "step": 5361 + }, + { + "epoch": 0.588842521414452, + "grad_norm": 2.351423978805542, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.701274037361145, + "num_tokens": 137843860.0, + "step": 5362 + }, + { + "epoch": 0.5889523391170657, + "grad_norm": 2.0585391521453857, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7160354256629944, + "num_tokens": 137871639.0, + "step": 5363 + }, + { + "epoch": 0.5890621568196793, + "grad_norm": 2.2599191665649414, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7229516506195068, + "num_tokens": 137893060.0, + "step": 5364 + }, + { + "epoch": 0.589171974522293, + "grad_norm": 2.371478796005249, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.727806568145752, + "num_tokens": 137914238.0, + "step": 5365 + }, + { + "epoch": 0.5892817922249066, + "grad_norm": 2.4206552505493164, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.736646294593811, + "num_tokens": 137934749.0, + "step": 5366 + }, + { + "epoch": 0.5893916099275203, + "grad_norm": 2.48006272315979, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7155489921569824, + "num_tokens": 137954831.0, + "step": 5367 + }, + { + "epoch": 0.589501427630134, + "grad_norm": 2.185887098312378, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.712716817855835, + "num_tokens": 137980087.0, + "step": 5368 + }, + { + "epoch": 0.5896112453327477, + "grad_norm": 2.196981191635132, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6910357475280762, + "num_tokens": 138005144.0, + "step": 5369 + }, + { + "epoch": 0.5897210630353613, + "grad_norm": 2.476431131362915, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.710483968257904, + "num_tokens": 138027141.0, + "step": 5370 + }, + { + "epoch": 0.589830880737975, + "grad_norm": 1.9815500974655151, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6984086036682129, + "num_tokens": 138057692.0, + "step": 5371 + }, + { + "epoch": 0.5899406984405886, + "grad_norm": 2.937075138092041, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.723111093044281, + "num_tokens": 138072459.0, + "step": 5372 + }, + { + "epoch": 0.5900505161432023, + "grad_norm": 2.0834968090057373, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6965515613555908, + "num_tokens": 138102315.0, + "step": 5373 + }, + { + "epoch": 0.5901603338458159, + "grad_norm": 2.0714287757873535, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7247340083122253, + "num_tokens": 138128408.0, + "step": 5374 + }, + { + "epoch": 0.5902701515484297, + "grad_norm": 2.147362232208252, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6950970888137817, + "num_tokens": 138156451.0, + "step": 5375 + }, + { + "epoch": 0.5903799692510433, + "grad_norm": 2.2951529026031494, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6973922252655029, + "num_tokens": 138182244.0, + "step": 5376 + }, + { + "epoch": 0.590489786953657, + "grad_norm": 2.038731575012207, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7077888250350952, + "num_tokens": 138209646.0, + "step": 5377 + }, + { + "epoch": 0.5905996046562706, + "grad_norm": 2.0806632041931152, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7044461965560913, + "num_tokens": 138236966.0, + "step": 5378 + }, + { + "epoch": 0.5907094223588842, + "grad_norm": 2.123260021209717, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.7017633318901062, + "num_tokens": 138264702.0, + "step": 5379 + }, + { + "epoch": 0.5908192400614979, + "grad_norm": 2.0621912479400635, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7160627841949463, + "num_tokens": 138293803.0, + "step": 5380 + }, + { + "epoch": 0.5909290577641115, + "grad_norm": 1.8709001541137695, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7023258209228516, + "num_tokens": 138325642.0, + "step": 5381 + }, + { + "epoch": 0.5910388754667253, + "grad_norm": 2.1325976848602295, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7014679908752441, + "num_tokens": 138353163.0, + "step": 5382 + }, + { + "epoch": 0.5911486931693389, + "grad_norm": 2.1805760860443115, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6944977641105652, + "num_tokens": 138380486.0, + "step": 5383 + }, + { + "epoch": 0.5912585108719526, + "grad_norm": 2.4069936275482178, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7104204893112183, + "num_tokens": 138401533.0, + "step": 5384 + }, + { + "epoch": 0.5913683285745662, + "grad_norm": 2.1065592765808105, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6863195896148682, + "num_tokens": 138430505.0, + "step": 5385 + }, + { + "epoch": 0.5914781462771799, + "grad_norm": 2.1853556632995605, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6962088346481323, + "num_tokens": 138457299.0, + "step": 5386 + }, + { + "epoch": 0.5915879639797935, + "grad_norm": 2.2244880199432373, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.67902672290802, + "num_tokens": 138481154.0, + "step": 5387 + }, + { + "epoch": 0.5916977816824072, + "grad_norm": 2.020141363143921, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6947239637374878, + "num_tokens": 138511051.0, + "step": 5388 + }, + { + "epoch": 0.5918075993850208, + "grad_norm": 2.3106634616851807, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6971309185028076, + "num_tokens": 138534872.0, + "step": 5389 + }, + { + "epoch": 0.5919174170876346, + "grad_norm": 2.3236916065216064, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.707014799118042, + "num_tokens": 138558643.0, + "step": 5390 + }, + { + "epoch": 0.5920272347902482, + "grad_norm": 2.180922269821167, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7160863280296326, + "num_tokens": 138583597.0, + "step": 5391 + }, + { + "epoch": 0.5921370524928619, + "grad_norm": 2.141937017440796, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7104902267456055, + "num_tokens": 138610411.0, + "step": 5392 + }, + { + "epoch": 0.5922468701954755, + "grad_norm": 1.9134900569915771, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6870174407958984, + "num_tokens": 138642852.0, + "step": 5393 + }, + { + "epoch": 0.5923566878980892, + "grad_norm": 2.1285605430603027, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7009148001670837, + "num_tokens": 138668354.0, + "step": 5394 + }, + { + "epoch": 0.5924665056007028, + "grad_norm": 2.0850462913513184, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7078366279602051, + "num_tokens": 138696577.0, + "step": 5395 + }, + { + "epoch": 0.5925763233033164, + "grad_norm": 2.4650933742523193, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6896854043006897, + "num_tokens": 138718865.0, + "step": 5396 + }, + { + "epoch": 0.5926861410059302, + "grad_norm": 2.36438250541687, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7168998718261719, + "num_tokens": 138742087.0, + "step": 5397 + }, + { + "epoch": 0.5927959587085438, + "grad_norm": 2.186347246170044, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.687597930431366, + "num_tokens": 138767557.0, + "step": 5398 + }, + { + "epoch": 0.5929057764111575, + "grad_norm": 2.140641212463379, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7120116949081421, + "num_tokens": 138794506.0, + "step": 5399 + }, + { + "epoch": 0.5930155941137711, + "grad_norm": 2.4298956394195557, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6907517910003662, + "num_tokens": 138815443.0, + "step": 5400 + }, + { + "epoch": 0.5931254118163848, + "grad_norm": 2.373502254486084, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7187256813049316, + "num_tokens": 138838377.0, + "step": 5401 + }, + { + "epoch": 0.5932352295189984, + "grad_norm": 2.094252824783325, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.707005500793457, + "num_tokens": 138865357.0, + "step": 5402 + }, + { + "epoch": 0.5933450472216121, + "grad_norm": 2.3181450366973877, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.711625337600708, + "num_tokens": 138888004.0, + "step": 5403 + }, + { + "epoch": 0.5934548649242258, + "grad_norm": 2.017714023590088, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7208662033081055, + "num_tokens": 138917718.0, + "step": 5404 + }, + { + "epoch": 0.5935646826268395, + "grad_norm": 2.1849989891052246, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6995200514793396, + "num_tokens": 138944656.0, + "step": 5405 + }, + { + "epoch": 0.5936745003294531, + "grad_norm": 2.4149510860443115, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7053846120834351, + "num_tokens": 138967675.0, + "step": 5406 + }, + { + "epoch": 0.5937843180320668, + "grad_norm": 2.332141637802124, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7121076583862305, + "num_tokens": 138990457.0, + "step": 5407 + }, + { + "epoch": 0.5938941357346804, + "grad_norm": 2.2371010780334473, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.70747971534729, + "num_tokens": 139014339.0, + "step": 5408 + }, + { + "epoch": 0.5940039534372941, + "grad_norm": 2.096461772918701, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7012494206428528, + "num_tokens": 139041467.0, + "step": 5409 + }, + { + "epoch": 0.5941137711399077, + "grad_norm": 2.2485344409942627, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.681007981300354, + "num_tokens": 139065605.0, + "step": 5410 + }, + { + "epoch": 0.5942235888425215, + "grad_norm": 2.3908345699310303, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7235717177391052, + "num_tokens": 139084715.0, + "step": 5411 + }, + { + "epoch": 0.5943334065451351, + "grad_norm": 1.9917978048324585, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7081716060638428, + "num_tokens": 139111867.0, + "step": 5412 + }, + { + "epoch": 0.5944432242477488, + "grad_norm": 2.0880603790283203, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.68796706199646, + "num_tokens": 139139387.0, + "step": 5413 + }, + { + "epoch": 0.5945530419503624, + "grad_norm": 1.9640090465545654, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6909255981445312, + "num_tokens": 139169902.0, + "step": 5414 + }, + { + "epoch": 0.594662859652976, + "grad_norm": 2.3342511653900146, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7141933441162109, + "num_tokens": 139193016.0, + "step": 5415 + }, + { + "epoch": 0.5947726773555897, + "grad_norm": 2.3632864952087402, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6936349868774414, + "num_tokens": 139217466.0, + "step": 5416 + }, + { + "epoch": 0.5948824950582033, + "grad_norm": 2.0848686695098877, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7192718386650085, + "num_tokens": 139245640.0, + "step": 5417 + }, + { + "epoch": 0.5949923127608171, + "grad_norm": 2.2540504932403564, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7082841992378235, + "num_tokens": 139268933.0, + "step": 5418 + }, + { + "epoch": 0.5951021304634307, + "grad_norm": 1.9681541919708252, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7404582500457764, + "num_tokens": 139298172.0, + "step": 5419 + }, + { + "epoch": 0.5952119481660444, + "grad_norm": 2.1062517166137695, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7096816897392273, + "num_tokens": 139324091.0, + "step": 5420 + }, + { + "epoch": 0.595321765868658, + "grad_norm": 1.9961376190185547, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7014771103858948, + "num_tokens": 139353643.0, + "step": 5421 + }, + { + "epoch": 0.5954315835712717, + "grad_norm": 2.1981143951416016, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6843946576118469, + "num_tokens": 139379755.0, + "step": 5422 + }, + { + "epoch": 0.5955414012738853, + "grad_norm": 2.468801975250244, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7033119797706604, + "num_tokens": 139400337.0, + "step": 5423 + }, + { + "epoch": 0.595651218976499, + "grad_norm": 2.181009531021118, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7152690887451172, + "num_tokens": 139425486.0, + "step": 5424 + }, + { + "epoch": 0.5957610366791126, + "grad_norm": 2.4621472358703613, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6936902403831482, + "num_tokens": 139446950.0, + "step": 5425 + }, + { + "epoch": 0.5958708543817264, + "grad_norm": 2.0256662368774414, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6928912401199341, + "num_tokens": 139476577.0, + "step": 5426 + }, + { + "epoch": 0.59598067208434, + "grad_norm": 2.226336717605591, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7117931842803955, + "num_tokens": 139500747.0, + "step": 5427 + }, + { + "epoch": 0.5960904897869537, + "grad_norm": 2.1365222930908203, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7055666446685791, + "num_tokens": 139529853.0, + "step": 5428 + }, + { + "epoch": 0.5962003074895673, + "grad_norm": 2.0860345363616943, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6911855936050415, + "num_tokens": 139558676.0, + "step": 5429 + }, + { + "epoch": 0.596310125192181, + "grad_norm": 2.4348466396331787, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7128386497497559, + "num_tokens": 139581424.0, + "step": 5430 + }, + { + "epoch": 0.5964199428947946, + "grad_norm": 2.3741648197174072, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7045263648033142, + "num_tokens": 139603521.0, + "step": 5431 + }, + { + "epoch": 0.5965297605974083, + "grad_norm": 2.119128704071045, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7200088500976562, + "num_tokens": 139631697.0, + "step": 5432 + }, + { + "epoch": 0.596639578300022, + "grad_norm": 2.2494845390319824, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7010312676429749, + "num_tokens": 139656610.0, + "step": 5433 + }, + { + "epoch": 0.5967493960026357, + "grad_norm": 2.1399149894714355, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7160303592681885, + "num_tokens": 139683185.0, + "step": 5434 + }, + { + "epoch": 0.5968592137052493, + "grad_norm": 1.9848086833953857, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7231196165084839, + "num_tokens": 139709702.0, + "step": 5435 + }, + { + "epoch": 0.596969031407863, + "grad_norm": 2.2931785583496094, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6854028701782227, + "num_tokens": 139735124.0, + "step": 5436 + }, + { + "epoch": 0.5970788491104766, + "grad_norm": 2.2162845134735107, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7101888656616211, + "num_tokens": 139759540.0, + "step": 5437 + }, + { + "epoch": 0.5971886668130902, + "grad_norm": 2.202516794204712, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6916484236717224, + "num_tokens": 139784442.0, + "step": 5438 + }, + { + "epoch": 0.5972984845157039, + "grad_norm": 2.211745023727417, + "learning_rate": 1e-06, + "loss": 1.1152, + "mean_token_accuracy": 0.6654353141784668, + "num_tokens": 139812009.0, + "step": 5439 + }, + { + "epoch": 0.5974083022183176, + "grad_norm": 2.228311777114868, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6982769966125488, + "num_tokens": 139837419.0, + "step": 5440 + }, + { + "epoch": 0.5975181199209313, + "grad_norm": 2.1457302570343018, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7079676389694214, + "num_tokens": 139863197.0, + "step": 5441 + }, + { + "epoch": 0.5976279376235449, + "grad_norm": 2.276466131210327, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6888034343719482, + "num_tokens": 139887142.0, + "step": 5442 + }, + { + "epoch": 0.5977377553261586, + "grad_norm": 2.058471918106079, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6963831186294556, + "num_tokens": 139913477.0, + "step": 5443 + }, + { + "epoch": 0.5978475730287722, + "grad_norm": 2.3551130294799805, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7021584510803223, + "num_tokens": 139935691.0, + "step": 5444 + }, + { + "epoch": 0.5979573907313859, + "grad_norm": 2.1977717876434326, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6850259900093079, + "num_tokens": 139961299.0, + "step": 5445 + }, + { + "epoch": 0.5980672084339995, + "grad_norm": 1.9126077890396118, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6908738613128662, + "num_tokens": 139995176.0, + "step": 5446 + }, + { + "epoch": 0.5981770261366133, + "grad_norm": 2.229517936706543, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6776119470596313, + "num_tokens": 140020450.0, + "step": 5447 + }, + { + "epoch": 0.5982868438392269, + "grad_norm": 2.183072328567505, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7007552981376648, + "num_tokens": 140044944.0, + "step": 5448 + }, + { + "epoch": 0.5983966615418406, + "grad_norm": 2.1026065349578857, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6926307082176208, + "num_tokens": 140073575.0, + "step": 5449 + }, + { + "epoch": 0.5985064792444542, + "grad_norm": 2.456876754760742, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7155168652534485, + "num_tokens": 140094092.0, + "step": 5450 + }, + { + "epoch": 0.5986162969470679, + "grad_norm": 2.383129119873047, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6958855986595154, + "num_tokens": 140116084.0, + "step": 5451 + }, + { + "epoch": 0.5987261146496815, + "grad_norm": 2.406665802001953, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.719186544418335, + "num_tokens": 140136567.0, + "step": 5452 + }, + { + "epoch": 0.5988359323522952, + "grad_norm": 2.0709261894226074, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6999096274375916, + "num_tokens": 140168132.0, + "step": 5453 + }, + { + "epoch": 0.5989457500549088, + "grad_norm": 1.9202848672866821, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6980710029602051, + "num_tokens": 140198669.0, + "step": 5454 + }, + { + "epoch": 0.5990555677575226, + "grad_norm": 2.1261353492736816, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6880075931549072, + "num_tokens": 140225949.0, + "step": 5455 + }, + { + "epoch": 0.5991653854601362, + "grad_norm": 2.0664865970611572, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6802306175231934, + "num_tokens": 140253256.0, + "step": 5456 + }, + { + "epoch": 0.5992752031627498, + "grad_norm": 1.9054205417633057, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7045590877532959, + "num_tokens": 140284466.0, + "step": 5457 + }, + { + "epoch": 0.5993850208653635, + "grad_norm": 2.356621503829956, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7075228691101074, + "num_tokens": 140307834.0, + "step": 5458 + }, + { + "epoch": 0.5994948385679771, + "grad_norm": 1.9711475372314453, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.7024310231208801, + "num_tokens": 140337178.0, + "step": 5459 + }, + { + "epoch": 0.5996046562705908, + "grad_norm": 2.1475603580474854, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7255541086196899, + "num_tokens": 140361430.0, + "step": 5460 + }, + { + "epoch": 0.5997144739732044, + "grad_norm": 1.875718593597412, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6977382898330688, + "num_tokens": 140392607.0, + "step": 5461 + }, + { + "epoch": 0.5998242916758182, + "grad_norm": 2.066621780395508, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6861151456832886, + "num_tokens": 140423825.0, + "step": 5462 + }, + { + "epoch": 0.5999341093784318, + "grad_norm": 2.396608591079712, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7097049951553345, + "num_tokens": 140446368.0, + "step": 5463 + }, + { + "epoch": 0.6000439270810455, + "grad_norm": 2.2434206008911133, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6946283578872681, + "num_tokens": 140469367.0, + "step": 5464 + }, + { + "epoch": 0.6001537447836591, + "grad_norm": 2.2905561923980713, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6941594481468201, + "num_tokens": 140495370.0, + "step": 5465 + }, + { + "epoch": 0.6002635624862728, + "grad_norm": 2.0308690071105957, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7141187787055969, + "num_tokens": 140524705.0, + "step": 5466 + }, + { + "epoch": 0.6003733801888864, + "grad_norm": 1.9454256296157837, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.722710907459259, + "num_tokens": 140557112.0, + "step": 5467 + }, + { + "epoch": 0.6004831978915001, + "grad_norm": 2.5861217975616455, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7210636138916016, + "num_tokens": 140576112.0, + "step": 5468 + }, + { + "epoch": 0.6005930155941138, + "grad_norm": 2.3386497497558594, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6870842576026917, + "num_tokens": 140601203.0, + "step": 5469 + }, + { + "epoch": 0.6007028332967275, + "grad_norm": 2.099313974380493, + "learning_rate": 1e-06, + "loss": 1.0928, + "mean_token_accuracy": 0.6731894016265869, + "num_tokens": 140628377.0, + "step": 5470 + }, + { + "epoch": 0.6008126509993411, + "grad_norm": 2.3114140033721924, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.7014085650444031, + "num_tokens": 140653167.0, + "step": 5471 + }, + { + "epoch": 0.6009224687019548, + "grad_norm": 2.107814311981201, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.711725115776062, + "num_tokens": 140679090.0, + "step": 5472 + }, + { + "epoch": 0.6010322864045684, + "grad_norm": 2.304657220840454, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.6824843883514404, + "num_tokens": 140705956.0, + "step": 5473 + }, + { + "epoch": 0.601142104107182, + "grad_norm": 2.117769956588745, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7107015252113342, + "num_tokens": 140731131.0, + "step": 5474 + }, + { + "epoch": 0.6012519218097957, + "grad_norm": 2.446197986602783, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6952701807022095, + "num_tokens": 140753102.0, + "step": 5475 + }, + { + "epoch": 0.6013617395124095, + "grad_norm": 2.4127700328826904, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7158493995666504, + "num_tokens": 140775075.0, + "step": 5476 + }, + { + "epoch": 0.6014715572150231, + "grad_norm": 2.32002592086792, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7072487473487854, + "num_tokens": 140798249.0, + "step": 5477 + }, + { + "epoch": 0.6015813749176367, + "grad_norm": 2.286893844604492, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6826028227806091, + "num_tokens": 140821129.0, + "step": 5478 + }, + { + "epoch": 0.6016911926202504, + "grad_norm": 2.06308913230896, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6934020519256592, + "num_tokens": 140850073.0, + "step": 5479 + }, + { + "epoch": 0.601801010322864, + "grad_norm": 2.391002655029297, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.691844642162323, + "num_tokens": 140871110.0, + "step": 5480 + }, + { + "epoch": 0.6019108280254777, + "grad_norm": 2.1651155948638916, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7106612920761108, + "num_tokens": 140895761.0, + "step": 5481 + }, + { + "epoch": 0.6020206457280913, + "grad_norm": 2.304011344909668, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7189159393310547, + "num_tokens": 140917745.0, + "step": 5482 + }, + { + "epoch": 0.602130463430705, + "grad_norm": 2.4645819664001465, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7145907878875732, + "num_tokens": 140939297.0, + "step": 5483 + }, + { + "epoch": 0.6022402811333187, + "grad_norm": 2.6361677646636963, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7178187370300293, + "num_tokens": 140958269.0, + "step": 5484 + }, + { + "epoch": 0.6023500988359324, + "grad_norm": 2.260178565979004, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6955029368400574, + "num_tokens": 140985215.0, + "step": 5485 + }, + { + "epoch": 0.602459916538546, + "grad_norm": 2.2080330848693848, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7005147933959961, + "num_tokens": 141009517.0, + "step": 5486 + }, + { + "epoch": 0.6025697342411597, + "grad_norm": 2.3443427085876465, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7147800922393799, + "num_tokens": 141031582.0, + "step": 5487 + }, + { + "epoch": 0.6026795519437733, + "grad_norm": 2.7166688442230225, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7376746535301208, + "num_tokens": 141048758.0, + "step": 5488 + }, + { + "epoch": 0.602789369646387, + "grad_norm": 2.1431057453155518, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6980816125869751, + "num_tokens": 141075433.0, + "step": 5489 + }, + { + "epoch": 0.6028991873490006, + "grad_norm": 2.3157293796539307, + "learning_rate": 1e-06, + "loss": 1.0748, + "mean_token_accuracy": 0.6829709410667419, + "num_tokens": 141103014.0, + "step": 5490 + }, + { + "epoch": 0.6030090050516144, + "grad_norm": 2.3144290447235107, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.728473961353302, + "num_tokens": 141125321.0, + "step": 5491 + }, + { + "epoch": 0.603118822754228, + "grad_norm": 2.180633068084717, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7054498195648193, + "num_tokens": 141151404.0, + "step": 5492 + }, + { + "epoch": 0.6032286404568417, + "grad_norm": 2.2584853172302246, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7074918746948242, + "num_tokens": 141175147.0, + "step": 5493 + }, + { + "epoch": 0.6033384581594553, + "grad_norm": 2.0611491203308105, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7174807190895081, + "num_tokens": 141202909.0, + "step": 5494 + }, + { + "epoch": 0.603448275862069, + "grad_norm": 1.9947402477264404, + "learning_rate": 1e-06, + "loss": 1.079, + "mean_token_accuracy": 0.6849308013916016, + "num_tokens": 141235609.0, + "step": 5495 + }, + { + "epoch": 0.6035580935646826, + "grad_norm": 2.1651499271392822, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6852794885635376, + "num_tokens": 141260975.0, + "step": 5496 + }, + { + "epoch": 0.6036679112672962, + "grad_norm": 2.2966601848602295, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7151191234588623, + "num_tokens": 141283482.0, + "step": 5497 + }, + { + "epoch": 0.60377772896991, + "grad_norm": 2.393665075302124, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7244116067886353, + "num_tokens": 141304357.0, + "step": 5498 + }, + { + "epoch": 0.6038875466725236, + "grad_norm": 2.1813623905181885, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6997638940811157, + "num_tokens": 141330510.0, + "step": 5499 + }, + { + "epoch": 0.6039973643751373, + "grad_norm": 2.134042263031006, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6859869360923767, + "num_tokens": 141356258.0, + "step": 5500 + }, + { + "epoch": 0.6041071820777509, + "grad_norm": 2.039605140686035, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7109394669532776, + "num_tokens": 141383110.0, + "step": 5501 + }, + { + "epoch": 0.6042169997803646, + "grad_norm": 1.9555245637893677, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7132601737976074, + "num_tokens": 141412378.0, + "step": 5502 + }, + { + "epoch": 0.6043268174829782, + "grad_norm": 2.19282603263855, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7170628905296326, + "num_tokens": 141436235.0, + "step": 5503 + }, + { + "epoch": 0.6044366351855919, + "grad_norm": 2.181380271911621, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7074630856513977, + "num_tokens": 141460855.0, + "step": 5504 + }, + { + "epoch": 0.6045464528882056, + "grad_norm": 1.9632890224456787, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6904031038284302, + "num_tokens": 141494824.0, + "step": 5505 + }, + { + "epoch": 0.6046562705908193, + "grad_norm": 2.163379669189453, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6929322481155396, + "num_tokens": 141523508.0, + "step": 5506 + }, + { + "epoch": 0.6047660882934329, + "grad_norm": 2.064279556274414, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7061172723770142, + "num_tokens": 141551125.0, + "step": 5507 + }, + { + "epoch": 0.6048759059960466, + "grad_norm": 2.217867374420166, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7185337543487549, + "num_tokens": 141576129.0, + "step": 5508 + }, + { + "epoch": 0.6049857236986602, + "grad_norm": 2.2564804553985596, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7164782881736755, + "num_tokens": 141600305.0, + "step": 5509 + }, + { + "epoch": 0.6050955414012739, + "grad_norm": 2.07905912399292, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.6972283124923706, + "num_tokens": 141628058.0, + "step": 5510 + }, + { + "epoch": 0.6052053591038875, + "grad_norm": 1.9988876581192017, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7229679822921753, + "num_tokens": 141657976.0, + "step": 5511 + }, + { + "epoch": 0.6053151768065012, + "grad_norm": 2.3350253105163574, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7069960832595825, + "num_tokens": 141679291.0, + "step": 5512 + }, + { + "epoch": 0.6054249945091149, + "grad_norm": 2.090608596801758, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6963783502578735, + "num_tokens": 141708591.0, + "step": 5513 + }, + { + "epoch": 0.6055348122117286, + "grad_norm": 2.5771331787109375, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7140277624130249, + "num_tokens": 141728854.0, + "step": 5514 + }, + { + "epoch": 0.6056446299143422, + "grad_norm": 2.106066942214966, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6848716735839844, + "num_tokens": 141758037.0, + "step": 5515 + }, + { + "epoch": 0.6057544476169558, + "grad_norm": 2.244215965270996, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.710732638835907, + "num_tokens": 141782093.0, + "step": 5516 + }, + { + "epoch": 0.6058642653195695, + "grad_norm": 2.1138930320739746, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7038635015487671, + "num_tokens": 141809802.0, + "step": 5517 + }, + { + "epoch": 0.6059740830221831, + "grad_norm": 2.1804959774017334, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7065610885620117, + "num_tokens": 141837912.0, + "step": 5518 + }, + { + "epoch": 0.6060839007247968, + "grad_norm": 2.1150336265563965, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.691232442855835, + "num_tokens": 141866100.0, + "step": 5519 + }, + { + "epoch": 0.6061937184274105, + "grad_norm": 2.2507991790771484, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6980525255203247, + "num_tokens": 141894642.0, + "step": 5520 + }, + { + "epoch": 0.6063035361300242, + "grad_norm": 2.385542869567871, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6985225081443787, + "num_tokens": 141914980.0, + "step": 5521 + }, + { + "epoch": 0.6064133538326378, + "grad_norm": 2.215747356414795, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.693511962890625, + "num_tokens": 141939451.0, + "step": 5522 + }, + { + "epoch": 0.6065231715352515, + "grad_norm": 2.525888442993164, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7043190002441406, + "num_tokens": 141958951.0, + "step": 5523 + }, + { + "epoch": 0.6066329892378651, + "grad_norm": 1.989469289779663, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7074024677276611, + "num_tokens": 141988289.0, + "step": 5524 + }, + { + "epoch": 0.6067428069404788, + "grad_norm": 1.9717392921447754, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7059444189071655, + "num_tokens": 142019995.0, + "step": 5525 + }, + { + "epoch": 0.6068526246430924, + "grad_norm": 2.1788852214813232, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6885019540786743, + "num_tokens": 142046221.0, + "step": 5526 + }, + { + "epoch": 0.6069624423457062, + "grad_norm": 2.2132463455200195, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7063207030296326, + "num_tokens": 142071459.0, + "step": 5527 + }, + { + "epoch": 0.6070722600483198, + "grad_norm": 2.4084911346435547, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6977552771568298, + "num_tokens": 142094513.0, + "step": 5528 + }, + { + "epoch": 0.6071820777509335, + "grad_norm": 2.420555591583252, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7158325910568237, + "num_tokens": 142115584.0, + "step": 5529 + }, + { + "epoch": 0.6072918954535471, + "grad_norm": 2.0364267826080322, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6979034543037415, + "num_tokens": 142146315.0, + "step": 5530 + }, + { + "epoch": 0.6074017131561608, + "grad_norm": 2.315727710723877, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7105914950370789, + "num_tokens": 142167742.0, + "step": 5531 + }, + { + "epoch": 0.6075115308587744, + "grad_norm": 2.1963582038879395, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.7008302211761475, + "num_tokens": 142192226.0, + "step": 5532 + }, + { + "epoch": 0.607621348561388, + "grad_norm": 2.1669535636901855, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6978564262390137, + "num_tokens": 142219428.0, + "step": 5533 + }, + { + "epoch": 0.6077311662640018, + "grad_norm": 2.288445472717285, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7093378305435181, + "num_tokens": 142241689.0, + "step": 5534 + }, + { + "epoch": 0.6078409839666155, + "grad_norm": 2.1600773334503174, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7009428143501282, + "num_tokens": 142267962.0, + "step": 5535 + }, + { + "epoch": 0.6079508016692291, + "grad_norm": 2.1764566898345947, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7082526683807373, + "num_tokens": 142295889.0, + "step": 5536 + }, + { + "epoch": 0.6080606193718427, + "grad_norm": 2.0078539848327637, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6910804510116577, + "num_tokens": 142324302.0, + "step": 5537 + }, + { + "epoch": 0.6081704370744564, + "grad_norm": 2.303534984588623, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.72295743227005, + "num_tokens": 142346717.0, + "step": 5538 + }, + { + "epoch": 0.60828025477707, + "grad_norm": 2.1282308101654053, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7203537821769714, + "num_tokens": 142376428.0, + "step": 5539 + }, + { + "epoch": 0.6083900724796837, + "grad_norm": 2.4111504554748535, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6785538196563721, + "num_tokens": 142400040.0, + "step": 5540 + }, + { + "epoch": 0.6084998901822973, + "grad_norm": 1.8918935060501099, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7030664682388306, + "num_tokens": 142430956.0, + "step": 5541 + }, + { + "epoch": 0.6086097078849111, + "grad_norm": 2.0337717533111572, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6939486265182495, + "num_tokens": 142461224.0, + "step": 5542 + }, + { + "epoch": 0.6087195255875247, + "grad_norm": 2.078838348388672, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7064653635025024, + "num_tokens": 142490207.0, + "step": 5543 + }, + { + "epoch": 0.6088293432901384, + "grad_norm": 2.356671094894409, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7159046530723572, + "num_tokens": 142512947.0, + "step": 5544 + }, + { + "epoch": 0.608939160992752, + "grad_norm": 1.9189794063568115, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.703298032283783, + "num_tokens": 142543759.0, + "step": 5545 + }, + { + "epoch": 0.6090489786953657, + "grad_norm": 2.151195526123047, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.68935626745224, + "num_tokens": 142570712.0, + "step": 5546 + }, + { + "epoch": 0.6091587963979793, + "grad_norm": 2.102118968963623, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.7017092704772949, + "num_tokens": 142598381.0, + "step": 5547 + }, + { + "epoch": 0.609268614100593, + "grad_norm": 2.2849438190460205, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7165860533714294, + "num_tokens": 142620787.0, + "step": 5548 + }, + { + "epoch": 0.6093784318032067, + "grad_norm": 2.6102395057678223, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7191269397735596, + "num_tokens": 142638708.0, + "step": 5549 + }, + { + "epoch": 0.6094882495058204, + "grad_norm": 2.270310401916504, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6864283084869385, + "num_tokens": 142664735.0, + "step": 5550 + }, + { + "epoch": 0.609598067208434, + "grad_norm": 1.988572120666504, + "learning_rate": 1e-06, + "loss": 1.1004, + "mean_token_accuracy": 0.6694650650024414, + "num_tokens": 142696934.0, + "step": 5551 + }, + { + "epoch": 0.6097078849110477, + "grad_norm": 1.9430205821990967, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6964191198348999, + "num_tokens": 142730930.0, + "step": 5552 + }, + { + "epoch": 0.6098177026136613, + "grad_norm": 2.339322805404663, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7016057968139648, + "num_tokens": 142754624.0, + "step": 5553 + }, + { + "epoch": 0.609927520316275, + "grad_norm": 2.011876106262207, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6942631006240845, + "num_tokens": 142784464.0, + "step": 5554 + }, + { + "epoch": 0.6100373380188886, + "grad_norm": 2.2175309658050537, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6874231100082397, + "num_tokens": 142810816.0, + "step": 5555 + }, + { + "epoch": 0.6101471557215024, + "grad_norm": 2.364959478378296, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7084174752235413, + "num_tokens": 142833585.0, + "step": 5556 + }, + { + "epoch": 0.610256973424116, + "grad_norm": 2.5252685546875, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7376714944839478, + "num_tokens": 142851859.0, + "step": 5557 + }, + { + "epoch": 0.6103667911267296, + "grad_norm": 2.2678380012512207, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6947087049484253, + "num_tokens": 142878669.0, + "step": 5558 + }, + { + "epoch": 0.6104766088293433, + "grad_norm": 2.074455738067627, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6917933225631714, + "num_tokens": 142908294.0, + "step": 5559 + }, + { + "epoch": 0.6105864265319569, + "grad_norm": 2.479816198348999, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.7103151082992554, + "num_tokens": 142931197.0, + "step": 5560 + }, + { + "epoch": 0.6106962442345706, + "grad_norm": 2.0689775943756104, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7100576162338257, + "num_tokens": 142958792.0, + "step": 5561 + }, + { + "epoch": 0.6108060619371842, + "grad_norm": 2.248241901397705, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6966869831085205, + "num_tokens": 142984482.0, + "step": 5562 + }, + { + "epoch": 0.610915879639798, + "grad_norm": 2.206197738647461, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6816559433937073, + "num_tokens": 143010772.0, + "step": 5563 + }, + { + "epoch": 0.6110256973424116, + "grad_norm": 2.382401466369629, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.693473219871521, + "num_tokens": 143034095.0, + "step": 5564 + }, + { + "epoch": 0.6111355150450253, + "grad_norm": 2.201317548751831, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6910931468009949, + "num_tokens": 143059594.0, + "step": 5565 + }, + { + "epoch": 0.6112453327476389, + "grad_norm": 2.197235345840454, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6814371347427368, + "num_tokens": 143086130.0, + "step": 5566 + }, + { + "epoch": 0.6113551504502526, + "grad_norm": 2.0550005435943604, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7217439413070679, + "num_tokens": 143113087.0, + "step": 5567 + }, + { + "epoch": 0.6114649681528662, + "grad_norm": 2.2848994731903076, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7157966494560242, + "num_tokens": 143136533.0, + "step": 5568 + }, + { + "epoch": 0.6115747858554799, + "grad_norm": 2.3298470973968506, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7133327126502991, + "num_tokens": 143158293.0, + "step": 5569 + }, + { + "epoch": 0.6116846035580935, + "grad_norm": 1.9236201047897339, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.705226719379425, + "num_tokens": 143190905.0, + "step": 5570 + }, + { + "epoch": 0.6117944212607073, + "grad_norm": 1.8277255296707153, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6980229616165161, + "num_tokens": 143225440.0, + "step": 5571 + }, + { + "epoch": 0.6119042389633209, + "grad_norm": 2.0282773971557617, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7300979495048523, + "num_tokens": 143253595.0, + "step": 5572 + }, + { + "epoch": 0.6120140566659346, + "grad_norm": 2.3158652782440186, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.714317262172699, + "num_tokens": 143278122.0, + "step": 5573 + }, + { + "epoch": 0.6121238743685482, + "grad_norm": 2.1492810249328613, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.704023003578186, + "num_tokens": 143304423.0, + "step": 5574 + }, + { + "epoch": 0.6122336920711619, + "grad_norm": 2.357625722885132, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6914324760437012, + "num_tokens": 143328374.0, + "step": 5575 + }, + { + "epoch": 0.6123435097737755, + "grad_norm": 2.5320558547973633, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.716352105140686, + "num_tokens": 143348710.0, + "step": 5576 + }, + { + "epoch": 0.6124533274763891, + "grad_norm": 2.504425525665283, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.700002908706665, + "num_tokens": 143369190.0, + "step": 5577 + }, + { + "epoch": 0.6125631451790029, + "grad_norm": 2.0658555030822754, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7252317667007446, + "num_tokens": 143395515.0, + "step": 5578 + }, + { + "epoch": 0.6126729628816165, + "grad_norm": 2.042612314224243, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7161375284194946, + "num_tokens": 143424918.0, + "step": 5579 + }, + { + "epoch": 0.6127827805842302, + "grad_norm": 2.1585869789123535, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6983210444450378, + "num_tokens": 143452336.0, + "step": 5580 + }, + { + "epoch": 0.6128925982868438, + "grad_norm": 1.9029229879379272, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.686493456363678, + "num_tokens": 143484841.0, + "step": 5581 + }, + { + "epoch": 0.6130024159894575, + "grad_norm": 2.1370396614074707, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.698913037776947, + "num_tokens": 143510589.0, + "step": 5582 + }, + { + "epoch": 0.6131122336920711, + "grad_norm": 2.2177700996398926, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.721598207950592, + "num_tokens": 143534966.0, + "step": 5583 + }, + { + "epoch": 0.6132220513946848, + "grad_norm": 2.3232522010803223, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6908243894577026, + "num_tokens": 143557797.0, + "step": 5584 + }, + { + "epoch": 0.6133318690972985, + "grad_norm": 2.155988931655884, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6918462514877319, + "num_tokens": 143585341.0, + "step": 5585 + }, + { + "epoch": 0.6134416867999122, + "grad_norm": 2.1498818397521973, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6860865354537964, + "num_tokens": 143611602.0, + "step": 5586 + }, + { + "epoch": 0.6135515045025258, + "grad_norm": 2.0348992347717285, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7126830220222473, + "num_tokens": 143640690.0, + "step": 5587 + }, + { + "epoch": 0.6136613222051395, + "grad_norm": 2.0141828060150146, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7050455212593079, + "num_tokens": 143670002.0, + "step": 5588 + }, + { + "epoch": 0.6137711399077531, + "grad_norm": 2.1492931842803955, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7076322436332703, + "num_tokens": 143695289.0, + "step": 5589 + }, + { + "epoch": 0.6138809576103668, + "grad_norm": 2.246851921081543, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7303246259689331, + "num_tokens": 143719413.0, + "step": 5590 + }, + { + "epoch": 0.6139907753129804, + "grad_norm": 2.335153579711914, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7131185531616211, + "num_tokens": 143740881.0, + "step": 5591 + }, + { + "epoch": 0.6141005930155942, + "grad_norm": 2.193805694580078, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.710151731967926, + "num_tokens": 143765531.0, + "step": 5592 + }, + { + "epoch": 0.6142104107182078, + "grad_norm": 2.198633909225464, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6949359774589539, + "num_tokens": 143793216.0, + "step": 5593 + }, + { + "epoch": 0.6143202284208215, + "grad_norm": 2.3991518020629883, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7198201417922974, + "num_tokens": 143814378.0, + "step": 5594 + }, + { + "epoch": 0.6144300461234351, + "grad_norm": 2.3982913494110107, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6964768171310425, + "num_tokens": 143838970.0, + "step": 5595 + }, + { + "epoch": 0.6145398638260487, + "grad_norm": 2.18733549118042, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6888879537582397, + "num_tokens": 143865355.0, + "step": 5596 + }, + { + "epoch": 0.6146496815286624, + "grad_norm": 2.2639431953430176, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7191928029060364, + "num_tokens": 143887626.0, + "step": 5597 + }, + { + "epoch": 0.614759499231276, + "grad_norm": 2.250319719314575, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7329941987991333, + "num_tokens": 143912742.0, + "step": 5598 + }, + { + "epoch": 0.6148693169338898, + "grad_norm": 2.5995237827301025, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7112253308296204, + "num_tokens": 143930785.0, + "step": 5599 + }, + { + "epoch": 0.6149791346365034, + "grad_norm": 2.025263786315918, + "learning_rate": 1e-06, + "loss": 1.0626, + "mean_token_accuracy": 0.6898744106292725, + "num_tokens": 143960338.0, + "step": 5600 + }, + { + "epoch": 0.6150889523391171, + "grad_norm": 2.025681495666504, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7211838960647583, + "num_tokens": 143987718.0, + "step": 5601 + }, + { + "epoch": 0.6151987700417307, + "grad_norm": 2.331202268600464, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7153104543685913, + "num_tokens": 144010937.0, + "step": 5602 + }, + { + "epoch": 0.6153085877443444, + "grad_norm": 2.6182868480682373, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7178654670715332, + "num_tokens": 144029340.0, + "step": 5603 + }, + { + "epoch": 0.615418405446958, + "grad_norm": 2.4984748363494873, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6914699673652649, + "num_tokens": 144051492.0, + "step": 5604 + }, + { + "epoch": 0.6155282231495717, + "grad_norm": 2.0606653690338135, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6830877661705017, + "num_tokens": 144079375.0, + "step": 5605 + }, + { + "epoch": 0.6156380408521853, + "grad_norm": 2.003840684890747, + "learning_rate": 1e-06, + "loss": 1.057, + "mean_token_accuracy": 0.6869077682495117, + "num_tokens": 144109221.0, + "step": 5606 + }, + { + "epoch": 0.6157478585547991, + "grad_norm": 2.305018663406372, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6904174089431763, + "num_tokens": 144132818.0, + "step": 5607 + }, + { + "epoch": 0.6158576762574127, + "grad_norm": 2.3791048526763916, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7115793228149414, + "num_tokens": 144154602.0, + "step": 5608 + }, + { + "epoch": 0.6159674939600264, + "grad_norm": 2.4612743854522705, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7064677476882935, + "num_tokens": 144173932.0, + "step": 5609 + }, + { + "epoch": 0.61607731166264, + "grad_norm": 2.1560099124908447, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7194075584411621, + "num_tokens": 144199731.0, + "step": 5610 + }, + { + "epoch": 0.6161871293652537, + "grad_norm": 2.3106656074523926, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7049546241760254, + "num_tokens": 144224898.0, + "step": 5611 + }, + { + "epoch": 0.6162969470678673, + "grad_norm": 2.2948479652404785, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6955323219299316, + "num_tokens": 144248141.0, + "step": 5612 + }, + { + "epoch": 0.616406764770481, + "grad_norm": 2.3374531269073486, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.729290246963501, + "num_tokens": 144270480.0, + "step": 5613 + }, + { + "epoch": 0.6165165824730947, + "grad_norm": 2.136857032775879, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7232890725135803, + "num_tokens": 144294973.0, + "step": 5614 + }, + { + "epoch": 0.6166264001757084, + "grad_norm": 2.2972989082336426, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7036730647087097, + "num_tokens": 144320203.0, + "step": 5615 + }, + { + "epoch": 0.616736217878322, + "grad_norm": 2.1218149662017822, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6912345886230469, + "num_tokens": 144346714.0, + "step": 5616 + }, + { + "epoch": 0.6168460355809356, + "grad_norm": 2.053748369216919, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6773409843444824, + "num_tokens": 144376145.0, + "step": 5617 + }, + { + "epoch": 0.6169558532835493, + "grad_norm": 2.1823644638061523, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7116848230361938, + "num_tokens": 144400175.0, + "step": 5618 + }, + { + "epoch": 0.6170656709861629, + "grad_norm": 1.951572060585022, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6946600079536438, + "num_tokens": 144433246.0, + "step": 5619 + }, + { + "epoch": 0.6171754886887766, + "grad_norm": 2.394596815109253, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6850875020027161, + "num_tokens": 144458652.0, + "step": 5620 + }, + { + "epoch": 0.6172853063913903, + "grad_norm": 1.9809118509292603, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7082707285881042, + "num_tokens": 144488285.0, + "step": 5621 + }, + { + "epoch": 0.617395124094004, + "grad_norm": 2.0647189617156982, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7023527026176453, + "num_tokens": 144516678.0, + "step": 5622 + }, + { + "epoch": 0.6175049417966176, + "grad_norm": 2.069464683532715, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7085564732551575, + "num_tokens": 144543827.0, + "step": 5623 + }, + { + "epoch": 0.6176147594992313, + "grad_norm": 2.1879942417144775, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7120183706283569, + "num_tokens": 144568841.0, + "step": 5624 + }, + { + "epoch": 0.6177245772018449, + "grad_norm": 2.039926052093506, + "learning_rate": 1e-06, + "loss": 1.0728, + "mean_token_accuracy": 0.6777477860450745, + "num_tokens": 144598643.0, + "step": 5625 + }, + { + "epoch": 0.6178343949044586, + "grad_norm": 2.3890132904052734, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7054637670516968, + "num_tokens": 144618543.0, + "step": 5626 + }, + { + "epoch": 0.6179442126070722, + "grad_norm": 2.2256627082824707, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6976339817047119, + "num_tokens": 144645156.0, + "step": 5627 + }, + { + "epoch": 0.618054030309686, + "grad_norm": 2.3094265460968018, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7260326147079468, + "num_tokens": 144669595.0, + "step": 5628 + }, + { + "epoch": 0.6181638480122996, + "grad_norm": 2.3351829051971436, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7166324853897095, + "num_tokens": 144694713.0, + "step": 5629 + }, + { + "epoch": 0.6182736657149133, + "grad_norm": 2.1344850063323975, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7048444151878357, + "num_tokens": 144721390.0, + "step": 5630 + }, + { + "epoch": 0.6183834834175269, + "grad_norm": 2.2506415843963623, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7075572609901428, + "num_tokens": 144745508.0, + "step": 5631 + }, + { + "epoch": 0.6184933011201406, + "grad_norm": 1.987412452697754, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6904861927032471, + "num_tokens": 144776635.0, + "step": 5632 + }, + { + "epoch": 0.6186031188227542, + "grad_norm": 1.8883779048919678, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7090938687324524, + "num_tokens": 144810488.0, + "step": 5633 + }, + { + "epoch": 0.6187129365253679, + "grad_norm": 2.2694623470306396, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.7022660970687866, + "num_tokens": 144834171.0, + "step": 5634 + }, + { + "epoch": 0.6188227542279815, + "grad_norm": 2.1760425567626953, + "learning_rate": 1e-06, + "loss": 1.0664, + "mean_token_accuracy": 0.6869384050369263, + "num_tokens": 144860274.0, + "step": 5635 + }, + { + "epoch": 0.6189325719305953, + "grad_norm": 2.356451988220215, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7120275497436523, + "num_tokens": 144882601.0, + "step": 5636 + }, + { + "epoch": 0.6190423896332089, + "grad_norm": 2.0925745964050293, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7223025560379028, + "num_tokens": 144909282.0, + "step": 5637 + }, + { + "epoch": 0.6191522073358225, + "grad_norm": 2.1805953979492188, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.6998299360275269, + "num_tokens": 144935039.0, + "step": 5638 + }, + { + "epoch": 0.6192620250384362, + "grad_norm": 2.041198253631592, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6839191913604736, + "num_tokens": 144964392.0, + "step": 5639 + }, + { + "epoch": 0.6193718427410498, + "grad_norm": 2.2558724880218506, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7060739398002625, + "num_tokens": 144989724.0, + "step": 5640 + }, + { + "epoch": 0.6194816604436635, + "grad_norm": 2.096328020095825, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7063120603561401, + "num_tokens": 145016108.0, + "step": 5641 + }, + { + "epoch": 0.6195914781462771, + "grad_norm": 2.5012781620025635, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6971216201782227, + "num_tokens": 145039728.0, + "step": 5642 + }, + { + "epoch": 0.6197012958488909, + "grad_norm": 2.6981089115142822, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7120670080184937, + "num_tokens": 145057985.0, + "step": 5643 + }, + { + "epoch": 0.6198111135515045, + "grad_norm": 2.1783010959625244, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6975017786026001, + "num_tokens": 145085133.0, + "step": 5644 + }, + { + "epoch": 0.6199209312541182, + "grad_norm": 2.284479856491089, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6867412328720093, + "num_tokens": 145109451.0, + "step": 5645 + }, + { + "epoch": 0.6200307489567318, + "grad_norm": 2.358227491378784, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7026889324188232, + "num_tokens": 145133263.0, + "step": 5646 + }, + { + "epoch": 0.6201405666593455, + "grad_norm": 2.6161084175109863, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7312533855438232, + "num_tokens": 145151651.0, + "step": 5647 + }, + { + "epoch": 0.6202503843619591, + "grad_norm": 2.128488302230835, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6997323632240295, + "num_tokens": 145178284.0, + "step": 5648 + }, + { + "epoch": 0.6203602020645728, + "grad_norm": 2.323965549468994, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7091972827911377, + "num_tokens": 145202075.0, + "step": 5649 + }, + { + "epoch": 0.6204700197671865, + "grad_norm": 2.445512056350708, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7367831468582153, + "num_tokens": 145223103.0, + "step": 5650 + }, + { + "epoch": 0.6205798374698002, + "grad_norm": 2.2370448112487793, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7228473424911499, + "num_tokens": 145246610.0, + "step": 5651 + }, + { + "epoch": 0.6206896551724138, + "grad_norm": 2.16479229927063, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7119157910346985, + "num_tokens": 145271659.0, + "step": 5652 + }, + { + "epoch": 0.6207994728750275, + "grad_norm": 2.068925142288208, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7002344131469727, + "num_tokens": 145300712.0, + "step": 5653 + }, + { + "epoch": 0.6209092905776411, + "grad_norm": 2.1926305294036865, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6869386434555054, + "num_tokens": 145326839.0, + "step": 5654 + }, + { + "epoch": 0.6210191082802548, + "grad_norm": 2.037553548812866, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7060335874557495, + "num_tokens": 145354853.0, + "step": 5655 + }, + { + "epoch": 0.6211289259828684, + "grad_norm": 2.182349443435669, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7060838341712952, + "num_tokens": 145379000.0, + "step": 5656 + }, + { + "epoch": 0.6212387436854822, + "grad_norm": 2.399033546447754, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6948950290679932, + "num_tokens": 145401969.0, + "step": 5657 + }, + { + "epoch": 0.6213485613880958, + "grad_norm": 2.4747977256774902, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7120791673660278, + "num_tokens": 145422595.0, + "step": 5658 + }, + { + "epoch": 0.6214583790907094, + "grad_norm": 2.0630548000335693, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7143250703811646, + "num_tokens": 145450092.0, + "step": 5659 + }, + { + "epoch": 0.6215681967933231, + "grad_norm": 2.465338706970215, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7456713318824768, + "num_tokens": 145469178.0, + "step": 5660 + }, + { + "epoch": 0.6216780144959367, + "grad_norm": 2.2944834232330322, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7282655835151672, + "num_tokens": 145491841.0, + "step": 5661 + }, + { + "epoch": 0.6217878321985504, + "grad_norm": 2.623624563217163, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.722498893737793, + "num_tokens": 145509887.0, + "step": 5662 + }, + { + "epoch": 0.621897649901164, + "grad_norm": 2.247236967086792, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7140089273452759, + "num_tokens": 145533299.0, + "step": 5663 + }, + { + "epoch": 0.6220074676037777, + "grad_norm": 1.889062762260437, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6897414922714233, + "num_tokens": 145566638.0, + "step": 5664 + }, + { + "epoch": 0.6221172853063914, + "grad_norm": 2.4478650093078613, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.696253776550293, + "num_tokens": 145589588.0, + "step": 5665 + }, + { + "epoch": 0.6222271030090051, + "grad_norm": 2.2723348140716553, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.7072021961212158, + "num_tokens": 145613824.0, + "step": 5666 + }, + { + "epoch": 0.6223369207116187, + "grad_norm": 2.014565944671631, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7227510809898376, + "num_tokens": 145641572.0, + "step": 5667 + }, + { + "epoch": 0.6224467384142324, + "grad_norm": 2.380608081817627, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7110713720321655, + "num_tokens": 145662861.0, + "step": 5668 + }, + { + "epoch": 0.622556556116846, + "grad_norm": 2.058483839035034, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6876398921012878, + "num_tokens": 145692286.0, + "step": 5669 + }, + { + "epoch": 0.6226663738194597, + "grad_norm": 2.1575872898101807, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7233005166053772, + "num_tokens": 145717704.0, + "step": 5670 + }, + { + "epoch": 0.6227761915220733, + "grad_norm": 2.187971591949463, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7267460227012634, + "num_tokens": 145741408.0, + "step": 5671 + }, + { + "epoch": 0.6228860092246871, + "grad_norm": 2.224863290786743, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7087181806564331, + "num_tokens": 145764227.0, + "step": 5672 + }, + { + "epoch": 0.6229958269273007, + "grad_norm": 2.406524181365967, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6907780766487122, + "num_tokens": 145787923.0, + "step": 5673 + }, + { + "epoch": 0.6231056446299144, + "grad_norm": 2.6116106510162354, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7299304604530334, + "num_tokens": 145806617.0, + "step": 5674 + }, + { + "epoch": 0.623215462332528, + "grad_norm": 2.1726067066192627, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7009509801864624, + "num_tokens": 145831281.0, + "step": 5675 + }, + { + "epoch": 0.6233252800351416, + "grad_norm": 2.315202474594116, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6815099716186523, + "num_tokens": 145856872.0, + "step": 5676 + }, + { + "epoch": 0.6234350977377553, + "grad_norm": 1.9834883213043213, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7079031467437744, + "num_tokens": 145888180.0, + "step": 5677 + }, + { + "epoch": 0.6235449154403689, + "grad_norm": 2.067440986633301, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7130376696586609, + "num_tokens": 145915243.0, + "step": 5678 + }, + { + "epoch": 0.6236547331429827, + "grad_norm": 2.3331730365753174, + "learning_rate": 1e-06, + "loss": 1.0491, + "mean_token_accuracy": 0.6990278959274292, + "num_tokens": 145940058.0, + "step": 5679 + }, + { + "epoch": 0.6237645508455963, + "grad_norm": 2.2144625186920166, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6998485326766968, + "num_tokens": 145965409.0, + "step": 5680 + }, + { + "epoch": 0.62387436854821, + "grad_norm": 2.0885183811187744, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7071717381477356, + "num_tokens": 145994622.0, + "step": 5681 + }, + { + "epoch": 0.6239841862508236, + "grad_norm": 2.5985910892486572, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7208929061889648, + "num_tokens": 146015332.0, + "step": 5682 + }, + { + "epoch": 0.6240940039534373, + "grad_norm": 2.0222134590148926, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7082746624946594, + "num_tokens": 146044189.0, + "step": 5683 + }, + { + "epoch": 0.6242038216560509, + "grad_norm": 1.935770869255066, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7049421072006226, + "num_tokens": 146075698.0, + "step": 5684 + }, + { + "epoch": 0.6243136393586646, + "grad_norm": 1.9857174158096313, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.713896632194519, + "num_tokens": 146103641.0, + "step": 5685 + }, + { + "epoch": 0.6244234570612783, + "grad_norm": 2.2797791957855225, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7192000150680542, + "num_tokens": 146126188.0, + "step": 5686 + }, + { + "epoch": 0.624533274763892, + "grad_norm": 2.330118179321289, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7078746557235718, + "num_tokens": 146149679.0, + "step": 5687 + }, + { + "epoch": 0.6246430924665056, + "grad_norm": 2.3878164291381836, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6835956573486328, + "num_tokens": 146172712.0, + "step": 5688 + }, + { + "epoch": 0.6247529101691193, + "grad_norm": 2.353670835494995, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6792514324188232, + "num_tokens": 146198261.0, + "step": 5689 + }, + { + "epoch": 0.6248627278717329, + "grad_norm": 2.1471798419952393, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7028444409370422, + "num_tokens": 146222804.0, + "step": 5690 + }, + { + "epoch": 0.6249725455743466, + "grad_norm": 2.174884557723999, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6827501058578491, + "num_tokens": 146250101.0, + "step": 5691 + }, + { + "epoch": 0.6250823632769602, + "grad_norm": 2.1941890716552734, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.708228588104248, + "num_tokens": 146275620.0, + "step": 5692 + }, + { + "epoch": 0.6251921809795739, + "grad_norm": 2.064408779144287, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6975805759429932, + "num_tokens": 146306121.0, + "step": 5693 + }, + { + "epoch": 0.6253019986821876, + "grad_norm": 2.002589464187622, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6938952207565308, + "num_tokens": 146336313.0, + "step": 5694 + }, + { + "epoch": 0.6254118163848013, + "grad_norm": 2.2722034454345703, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7286516427993774, + "num_tokens": 146360908.0, + "step": 5695 + }, + { + "epoch": 0.6255216340874149, + "grad_norm": 2.2665891647338867, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.706451416015625, + "num_tokens": 146383370.0, + "step": 5696 + }, + { + "epoch": 0.6256314517900285, + "grad_norm": 2.1859796047210693, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7065975666046143, + "num_tokens": 146409097.0, + "step": 5697 + }, + { + "epoch": 0.6257412694926422, + "grad_norm": 2.242103099822998, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.6988362669944763, + "num_tokens": 146433269.0, + "step": 5698 + }, + { + "epoch": 0.6258510871952558, + "grad_norm": 2.5521416664123535, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7140401005744934, + "num_tokens": 146453501.0, + "step": 5699 + }, + { + "epoch": 0.6259609048978695, + "grad_norm": 2.0924479961395264, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7039608955383301, + "num_tokens": 146481084.0, + "step": 5700 + }, + { + "epoch": 0.6260707226004832, + "grad_norm": 2.1663548946380615, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6952864527702332, + "num_tokens": 146508016.0, + "step": 5701 + }, + { + "epoch": 0.6261805403030969, + "grad_norm": 2.2207043170928955, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6876202821731567, + "num_tokens": 146535502.0, + "step": 5702 + }, + { + "epoch": 0.6262903580057105, + "grad_norm": 1.8136138916015625, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.684739351272583, + "num_tokens": 146570828.0, + "step": 5703 + }, + { + "epoch": 0.6264001757083242, + "grad_norm": 2.157228708267212, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.691582977771759, + "num_tokens": 146602059.0, + "step": 5704 + }, + { + "epoch": 0.6265099934109378, + "grad_norm": 2.2636406421661377, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7041879892349243, + "num_tokens": 146628526.0, + "step": 5705 + }, + { + "epoch": 0.6266198111135515, + "grad_norm": 2.314126968383789, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7136574983596802, + "num_tokens": 146651274.0, + "step": 5706 + }, + { + "epoch": 0.6267296288161651, + "grad_norm": 2.184943199157715, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7357232570648193, + "num_tokens": 146676052.0, + "step": 5707 + }, + { + "epoch": 0.6268394465187789, + "grad_norm": 2.2495856285095215, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6963282823562622, + "num_tokens": 146701018.0, + "step": 5708 + }, + { + "epoch": 0.6269492642213925, + "grad_norm": 2.1317098140716553, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6930973529815674, + "num_tokens": 146726857.0, + "step": 5709 + }, + { + "epoch": 0.6270590819240062, + "grad_norm": 2.2088334560394287, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7083246111869812, + "num_tokens": 146752285.0, + "step": 5710 + }, + { + "epoch": 0.6271688996266198, + "grad_norm": 2.1998326778411865, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.697964072227478, + "num_tokens": 146775340.0, + "step": 5711 + }, + { + "epoch": 0.6272787173292335, + "grad_norm": 2.112067222595215, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.7001748085021973, + "num_tokens": 146801662.0, + "step": 5712 + }, + { + "epoch": 0.6273885350318471, + "grad_norm": 2.268110752105713, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.708560585975647, + "num_tokens": 146824835.0, + "step": 5713 + }, + { + "epoch": 0.6274983527344608, + "grad_norm": 2.1336047649383545, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7294344305992126, + "num_tokens": 146852144.0, + "step": 5714 + }, + { + "epoch": 0.6276081704370745, + "grad_norm": 2.388751983642578, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6982549428939819, + "num_tokens": 146874042.0, + "step": 5715 + }, + { + "epoch": 0.6277179881396882, + "grad_norm": 2.120213031768799, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7231176495552063, + "num_tokens": 146898719.0, + "step": 5716 + }, + { + "epoch": 0.6278278058423018, + "grad_norm": 1.9992700815200806, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6944386959075928, + "num_tokens": 146928945.0, + "step": 5717 + }, + { + "epoch": 0.6279376235449154, + "grad_norm": 2.186549663543701, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7089760303497314, + "num_tokens": 146955016.0, + "step": 5718 + }, + { + "epoch": 0.6280474412475291, + "grad_norm": 2.498859405517578, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6987934112548828, + "num_tokens": 146976167.0, + "step": 5719 + }, + { + "epoch": 0.6281572589501427, + "grad_norm": 2.183366298675537, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6937148571014404, + "num_tokens": 147002602.0, + "step": 5720 + }, + { + "epoch": 0.6282670766527564, + "grad_norm": 2.2962586879730225, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6928087472915649, + "num_tokens": 147025576.0, + "step": 5721 + }, + { + "epoch": 0.62837689435537, + "grad_norm": 2.0783779621124268, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7304399609565735, + "num_tokens": 147048828.0, + "step": 5722 + }, + { + "epoch": 0.6284867120579838, + "grad_norm": 2.3421378135681152, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6983727216720581, + "num_tokens": 147072150.0, + "step": 5723 + }, + { + "epoch": 0.6285965297605974, + "grad_norm": 2.0014491081237793, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6912046670913696, + "num_tokens": 147102673.0, + "step": 5724 + }, + { + "epoch": 0.6287063474632111, + "grad_norm": 2.482969045639038, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7143335938453674, + "num_tokens": 147125698.0, + "step": 5725 + }, + { + "epoch": 0.6288161651658247, + "grad_norm": 2.3829939365386963, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6957265138626099, + "num_tokens": 147150668.0, + "step": 5726 + }, + { + "epoch": 0.6289259828684384, + "grad_norm": 2.32081937789917, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7128876447677612, + "num_tokens": 147172704.0, + "step": 5727 + }, + { + "epoch": 0.629035800571052, + "grad_norm": 1.9793510437011719, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6736851930618286, + "num_tokens": 147204515.0, + "step": 5728 + }, + { + "epoch": 0.6291456182736657, + "grad_norm": 2.1272401809692383, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6903034448623657, + "num_tokens": 147232435.0, + "step": 5729 + }, + { + "epoch": 0.6292554359762794, + "grad_norm": 2.3983874320983887, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7179885506629944, + "num_tokens": 147253241.0, + "step": 5730 + }, + { + "epoch": 0.6293652536788931, + "grad_norm": 2.2416889667510986, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6825365424156189, + "num_tokens": 147280366.0, + "step": 5731 + }, + { + "epoch": 0.6294750713815067, + "grad_norm": 2.2826619148254395, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7177855372428894, + "num_tokens": 147304080.0, + "step": 5732 + }, + { + "epoch": 0.6295848890841204, + "grad_norm": 2.533203601837158, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7116029262542725, + "num_tokens": 147325340.0, + "step": 5733 + }, + { + "epoch": 0.629694706786734, + "grad_norm": 2.171988010406494, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7064088582992554, + "num_tokens": 147352153.0, + "step": 5734 + }, + { + "epoch": 0.6298045244893477, + "grad_norm": 2.2228755950927734, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7058424949645996, + "num_tokens": 147377332.0, + "step": 5735 + }, + { + "epoch": 0.6299143421919613, + "grad_norm": 1.9472228288650513, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6941999793052673, + "num_tokens": 147408824.0, + "step": 5736 + }, + { + "epoch": 0.630024159894575, + "grad_norm": 2.355726957321167, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7062649726867676, + "num_tokens": 147430156.0, + "step": 5737 + }, + { + "epoch": 0.6301339775971887, + "grad_norm": 2.3613719940185547, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7160227298736572, + "num_tokens": 147451752.0, + "step": 5738 + }, + { + "epoch": 0.6302437952998023, + "grad_norm": 2.0406360626220703, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7141021490097046, + "num_tokens": 147478699.0, + "step": 5739 + }, + { + "epoch": 0.630353613002416, + "grad_norm": 2.401991367340088, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7355118989944458, + "num_tokens": 147499034.0, + "step": 5740 + }, + { + "epoch": 0.6304634307050296, + "grad_norm": 2.165870428085327, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7133406400680542, + "num_tokens": 147524918.0, + "step": 5741 + }, + { + "epoch": 0.6305732484076433, + "grad_norm": 2.2386717796325684, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6967405676841736, + "num_tokens": 147548324.0, + "step": 5742 + }, + { + "epoch": 0.6306830661102569, + "grad_norm": 2.725492000579834, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7121769189834595, + "num_tokens": 147565146.0, + "step": 5743 + }, + { + "epoch": 0.6307928838128707, + "grad_norm": 1.869093656539917, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6854715347290039, + "num_tokens": 147597045.0, + "step": 5744 + }, + { + "epoch": 0.6309027015154843, + "grad_norm": 2.1966285705566406, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7020476460456848, + "num_tokens": 147623389.0, + "step": 5745 + }, + { + "epoch": 0.631012519218098, + "grad_norm": 2.1196422576904297, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.7031305432319641, + "num_tokens": 147648678.0, + "step": 5746 + }, + { + "epoch": 0.6311223369207116, + "grad_norm": 1.967573881149292, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7130860090255737, + "num_tokens": 147677347.0, + "step": 5747 + }, + { + "epoch": 0.6312321546233253, + "grad_norm": 2.390322685241699, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6979942321777344, + "num_tokens": 147699717.0, + "step": 5748 + }, + { + "epoch": 0.6313419723259389, + "grad_norm": 2.109788179397583, + "learning_rate": 1e-06, + "loss": 1.063, + "mean_token_accuracy": 0.6822061538696289, + "num_tokens": 147728778.0, + "step": 5749 + }, + { + "epoch": 0.6314517900285526, + "grad_norm": 2.093275308609009, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7148153781890869, + "num_tokens": 147755106.0, + "step": 5750 + }, + { + "epoch": 0.6315616077311663, + "grad_norm": 2.012129306793213, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7090661525726318, + "num_tokens": 147783132.0, + "step": 5751 + }, + { + "epoch": 0.63167142543378, + "grad_norm": 2.007737398147583, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7164456844329834, + "num_tokens": 147813091.0, + "step": 5752 + }, + { + "epoch": 0.6317812431363936, + "grad_norm": 2.575368642807007, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7660828232765198, + "num_tokens": 147830192.0, + "step": 5753 + }, + { + "epoch": 0.6318910608390073, + "grad_norm": 2.4148175716400146, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7118779420852661, + "num_tokens": 147853039.0, + "step": 5754 + }, + { + "epoch": 0.6320008785416209, + "grad_norm": 2.3172450065612793, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6895159482955933, + "num_tokens": 147876344.0, + "step": 5755 + }, + { + "epoch": 0.6321106962442345, + "grad_norm": 2.5045933723449707, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6964089870452881, + "num_tokens": 147896792.0, + "step": 5756 + }, + { + "epoch": 0.6322205139468482, + "grad_norm": 2.180209159851074, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7058086395263672, + "num_tokens": 147921974.0, + "step": 5757 + }, + { + "epoch": 0.6323303316494618, + "grad_norm": 2.2970187664031982, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7060503959655762, + "num_tokens": 147946976.0, + "step": 5758 + }, + { + "epoch": 0.6324401493520756, + "grad_norm": 2.1737723350524902, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7126328945159912, + "num_tokens": 147969832.0, + "step": 5759 + }, + { + "epoch": 0.6325499670546892, + "grad_norm": 2.1266679763793945, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.7017150521278381, + "num_tokens": 147996079.0, + "step": 5760 + }, + { + "epoch": 0.6326597847573029, + "grad_norm": 2.021409749984741, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7071976661682129, + "num_tokens": 148023526.0, + "step": 5761 + }, + { + "epoch": 0.6327696024599165, + "grad_norm": 2.2842838764190674, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7303251028060913, + "num_tokens": 148045088.0, + "step": 5762 + }, + { + "epoch": 0.6328794201625302, + "grad_norm": 2.208265542984009, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7048976421356201, + "num_tokens": 148068905.0, + "step": 5763 + }, + { + "epoch": 0.6329892378651438, + "grad_norm": 2.434493064880371, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7333693504333496, + "num_tokens": 148089136.0, + "step": 5764 + }, + { + "epoch": 0.6330990555677575, + "grad_norm": 2.016371726989746, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.7093690633773804, + "num_tokens": 148116855.0, + "step": 5765 + }, + { + "epoch": 0.6332088732703712, + "grad_norm": 2.0387520790100098, + "learning_rate": 1e-06, + "loss": 1.0808, + "mean_token_accuracy": 0.6762064099311829, + "num_tokens": 148144761.0, + "step": 5766 + }, + { + "epoch": 0.6333186909729849, + "grad_norm": 2.1654703617095947, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.6795072555541992, + "num_tokens": 148171858.0, + "step": 5767 + }, + { + "epoch": 0.6334285086755985, + "grad_norm": 1.9557791948318481, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7091535329818726, + "num_tokens": 148200447.0, + "step": 5768 + }, + { + "epoch": 0.6335383263782122, + "grad_norm": 1.9773887395858765, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7042590975761414, + "num_tokens": 148230116.0, + "step": 5769 + }, + { + "epoch": 0.6336481440808258, + "grad_norm": 2.1041266918182373, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6911395192146301, + "num_tokens": 148256609.0, + "step": 5770 + }, + { + "epoch": 0.6337579617834395, + "grad_norm": 2.3270058631896973, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7066826820373535, + "num_tokens": 148279950.0, + "step": 5771 + }, + { + "epoch": 0.6338677794860531, + "grad_norm": 2.2214152812957764, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6846165657043457, + "num_tokens": 148304792.0, + "step": 5772 + }, + { + "epoch": 0.6339775971886669, + "grad_norm": 2.4416632652282715, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6971105337142944, + "num_tokens": 148325807.0, + "step": 5773 + }, + { + "epoch": 0.6340874148912805, + "grad_norm": 2.1989521980285645, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.717635989189148, + "num_tokens": 148349025.0, + "step": 5774 + }, + { + "epoch": 0.6341972325938942, + "grad_norm": 2.262953996658325, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7006703615188599, + "num_tokens": 148372481.0, + "step": 5775 + }, + { + "epoch": 0.6343070502965078, + "grad_norm": 2.0836985111236572, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7048270106315613, + "num_tokens": 148400269.0, + "step": 5776 + }, + { + "epoch": 0.6344168679991214, + "grad_norm": 2.2646429538726807, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7063591480255127, + "num_tokens": 148422791.0, + "step": 5777 + }, + { + "epoch": 0.6345266857017351, + "grad_norm": 2.5218119621276855, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7281978130340576, + "num_tokens": 148442270.0, + "step": 5778 + }, + { + "epoch": 0.6346365034043487, + "grad_norm": 2.36849045753479, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7119652032852173, + "num_tokens": 148464353.0, + "step": 5779 + }, + { + "epoch": 0.6347463211069625, + "grad_norm": 2.2648563385009766, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.7015315294265747, + "num_tokens": 148488744.0, + "step": 5780 + }, + { + "epoch": 0.6348561388095761, + "grad_norm": 2.455235719680786, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.716921329498291, + "num_tokens": 148509966.0, + "step": 5781 + }, + { + "epoch": 0.6349659565121898, + "grad_norm": 2.2578606605529785, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6915273666381836, + "num_tokens": 148534961.0, + "step": 5782 + }, + { + "epoch": 0.6350757742148034, + "grad_norm": 2.439065456390381, + "learning_rate": 1e-06, + "loss": 1.0707, + "mean_token_accuracy": 0.6821119785308838, + "num_tokens": 148557042.0, + "step": 5783 + }, + { + "epoch": 0.6351855919174171, + "grad_norm": 2.061458110809326, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6931711435317993, + "num_tokens": 148585315.0, + "step": 5784 + }, + { + "epoch": 0.6352954096200307, + "grad_norm": 2.233577251434326, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7196489572525024, + "num_tokens": 148607765.0, + "step": 5785 + }, + { + "epoch": 0.6354052273226444, + "grad_norm": 2.108222484588623, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6925227046012878, + "num_tokens": 148637030.0, + "step": 5786 + }, + { + "epoch": 0.635515045025258, + "grad_norm": 1.8485934734344482, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6967581510543823, + "num_tokens": 148676564.0, + "step": 5787 + }, + { + "epoch": 0.6356248627278718, + "grad_norm": 2.08321213722229, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7035617828369141, + "num_tokens": 148704169.0, + "step": 5788 + }, + { + "epoch": 0.6357346804304854, + "grad_norm": 2.271338701248169, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7069023847579956, + "num_tokens": 148728742.0, + "step": 5789 + }, + { + "epoch": 0.6358444981330991, + "grad_norm": 2.073427438735962, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7079036235809326, + "num_tokens": 148757331.0, + "step": 5790 + }, + { + "epoch": 0.6359543158357127, + "grad_norm": 2.311367988586426, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7322108149528503, + "num_tokens": 148779110.0, + "step": 5791 + }, + { + "epoch": 0.6360641335383264, + "grad_norm": 2.2523670196533203, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7207027077674866, + "num_tokens": 148802747.0, + "step": 5792 + }, + { + "epoch": 0.63617395124094, + "grad_norm": 2.2208681106567383, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7093586921691895, + "num_tokens": 148827126.0, + "step": 5793 + }, + { + "epoch": 0.6362837689435537, + "grad_norm": 2.1755831241607666, + "learning_rate": 1e-06, + "loss": 1.088, + "mean_token_accuracy": 0.6724549531936646, + "num_tokens": 148855224.0, + "step": 5794 + }, + { + "epoch": 0.6363935866461674, + "grad_norm": 2.2144856452941895, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7024656534194946, + "num_tokens": 148879914.0, + "step": 5795 + }, + { + "epoch": 0.636503404348781, + "grad_norm": 1.95274019241333, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6968704462051392, + "num_tokens": 148914568.0, + "step": 5796 + }, + { + "epoch": 0.6366132220513947, + "grad_norm": 1.976499319076538, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7140003442764282, + "num_tokens": 148946900.0, + "step": 5797 + }, + { + "epoch": 0.6367230397540083, + "grad_norm": 2.1308255195617676, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6849891543388367, + "num_tokens": 148976450.0, + "step": 5798 + }, + { + "epoch": 0.636832857456622, + "grad_norm": 2.2323074340820312, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7061997652053833, + "num_tokens": 149003895.0, + "step": 5799 + }, + { + "epoch": 0.6369426751592356, + "grad_norm": 1.9335674047470093, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6786333322525024, + "num_tokens": 149035225.0, + "step": 5800 + }, + { + "epoch": 0.6370524928618493, + "grad_norm": 2.1451570987701416, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7101387977600098, + "num_tokens": 149061115.0, + "step": 5801 + }, + { + "epoch": 0.637162310564463, + "grad_norm": 2.215052366256714, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7040379047393799, + "num_tokens": 149085671.0, + "step": 5802 + }, + { + "epoch": 0.6372721282670767, + "grad_norm": 2.1552653312683105, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7048390507698059, + "num_tokens": 149110718.0, + "step": 5803 + }, + { + "epoch": 0.6373819459696903, + "grad_norm": 2.463900566101074, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7190999984741211, + "num_tokens": 149129635.0, + "step": 5804 + }, + { + "epoch": 0.637491763672304, + "grad_norm": 2.198953151702881, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6951867938041687, + "num_tokens": 149155199.0, + "step": 5805 + }, + { + "epoch": 0.6376015813749176, + "grad_norm": 2.076911449432373, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6897826194763184, + "num_tokens": 149184738.0, + "step": 5806 + }, + { + "epoch": 0.6377113990775313, + "grad_norm": 2.4050352573394775, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7042332887649536, + "num_tokens": 149206369.0, + "step": 5807 + }, + { + "epoch": 0.6378212167801449, + "grad_norm": 2.4165401458740234, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7292842864990234, + "num_tokens": 149227021.0, + "step": 5808 + }, + { + "epoch": 0.6379310344827587, + "grad_norm": 2.451050281524658, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7205733060836792, + "num_tokens": 149248157.0, + "step": 5809 + }, + { + "epoch": 0.6380408521853723, + "grad_norm": 2.302420139312744, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6950814723968506, + "num_tokens": 149273053.0, + "step": 5810 + }, + { + "epoch": 0.638150669887986, + "grad_norm": 2.4817562103271484, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7066051959991455, + "num_tokens": 149293866.0, + "step": 5811 + }, + { + "epoch": 0.6382604875905996, + "grad_norm": 2.3455252647399902, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.703690230846405, + "num_tokens": 149321193.0, + "step": 5812 + }, + { + "epoch": 0.6383703052932133, + "grad_norm": 2.3489513397216797, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7168872356414795, + "num_tokens": 149344369.0, + "step": 5813 + }, + { + "epoch": 0.6384801229958269, + "grad_norm": 2.047287940979004, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6969255208969116, + "num_tokens": 149373997.0, + "step": 5814 + }, + { + "epoch": 0.6385899406984406, + "grad_norm": 2.1069719791412354, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7043368220329285, + "num_tokens": 149402145.0, + "step": 5815 + }, + { + "epoch": 0.6386997584010542, + "grad_norm": 2.1634340286254883, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7014013528823853, + "num_tokens": 149428495.0, + "step": 5816 + }, + { + "epoch": 0.638809576103668, + "grad_norm": 2.249422311782837, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7163525819778442, + "num_tokens": 149452235.0, + "step": 5817 + }, + { + "epoch": 0.6389193938062816, + "grad_norm": 2.325876235961914, + "learning_rate": 1e-06, + "loss": 1.0772, + "mean_token_accuracy": 0.6871780157089233, + "num_tokens": 149475745.0, + "step": 5818 + }, + { + "epoch": 0.6390292115088952, + "grad_norm": 1.9935433864593506, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6833044290542603, + "num_tokens": 149507800.0, + "step": 5819 + }, + { + "epoch": 0.6391390292115089, + "grad_norm": 2.1952035427093506, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.7048271894454956, + "num_tokens": 149534995.0, + "step": 5820 + }, + { + "epoch": 0.6392488469141225, + "grad_norm": 2.2189865112304688, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7066589593887329, + "num_tokens": 149558594.0, + "step": 5821 + }, + { + "epoch": 0.6393586646167362, + "grad_norm": 2.28054141998291, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6961429715156555, + "num_tokens": 149583729.0, + "step": 5822 + }, + { + "epoch": 0.6394684823193498, + "grad_norm": 2.429171085357666, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6881325244903564, + "num_tokens": 149605801.0, + "step": 5823 + }, + { + "epoch": 0.6395783000219636, + "grad_norm": 2.2720139026641846, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6938886642456055, + "num_tokens": 149631401.0, + "step": 5824 + }, + { + "epoch": 0.6396881177245772, + "grad_norm": 2.6949846744537354, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6924457550048828, + "num_tokens": 149658280.0, + "step": 5825 + }, + { + "epoch": 0.6397979354271909, + "grad_norm": 2.2307608127593994, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7097232341766357, + "num_tokens": 149683017.0, + "step": 5826 + }, + { + "epoch": 0.6399077531298045, + "grad_norm": 2.1821579933166504, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.685714840888977, + "num_tokens": 149710671.0, + "step": 5827 + }, + { + "epoch": 0.6400175708324182, + "grad_norm": 2.32772159576416, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7195990085601807, + "num_tokens": 149735123.0, + "step": 5828 + }, + { + "epoch": 0.6401273885350318, + "grad_norm": 2.3698813915252686, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7160214185714722, + "num_tokens": 149756259.0, + "step": 5829 + }, + { + "epoch": 0.6402372062376455, + "grad_norm": 2.1990904808044434, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7178117036819458, + "num_tokens": 149780891.0, + "step": 5830 + }, + { + "epoch": 0.6403470239402592, + "grad_norm": 2.197007179260254, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6991827487945557, + "num_tokens": 149807124.0, + "step": 5831 + }, + { + "epoch": 0.6404568416428729, + "grad_norm": 2.048095703125, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7491531372070312, + "num_tokens": 149832254.0, + "step": 5832 + }, + { + "epoch": 0.6405666593454865, + "grad_norm": 2.214191436767578, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6841966509819031, + "num_tokens": 149860985.0, + "step": 5833 + }, + { + "epoch": 0.6406764770481002, + "grad_norm": 2.386998176574707, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6957470178604126, + "num_tokens": 149885787.0, + "step": 5834 + }, + { + "epoch": 0.6407862947507138, + "grad_norm": 2.3832149505615234, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6837607622146606, + "num_tokens": 149909542.0, + "step": 5835 + }, + { + "epoch": 0.6408961124533274, + "grad_norm": 2.3074564933776855, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7133523225784302, + "num_tokens": 149932408.0, + "step": 5836 + }, + { + "epoch": 0.6410059301559411, + "grad_norm": 2.0610363483428955, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7201972007751465, + "num_tokens": 149959555.0, + "step": 5837 + }, + { + "epoch": 0.6411157478585549, + "grad_norm": 1.886359691619873, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6876460313796997, + "num_tokens": 149992996.0, + "step": 5838 + }, + { + "epoch": 0.6412255655611685, + "grad_norm": 2.365797996520996, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7041876316070557, + "num_tokens": 150014937.0, + "step": 5839 + }, + { + "epoch": 0.6413353832637821, + "grad_norm": 2.1682915687561035, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7083967328071594, + "num_tokens": 150040068.0, + "step": 5840 + }, + { + "epoch": 0.6414452009663958, + "grad_norm": 2.014575719833374, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7252762317657471, + "num_tokens": 150068294.0, + "step": 5841 + }, + { + "epoch": 0.6415550186690094, + "grad_norm": 2.0032846927642822, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6952004432678223, + "num_tokens": 150097530.0, + "step": 5842 + }, + { + "epoch": 0.6416648363716231, + "grad_norm": 2.4503333568573, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6806050539016724, + "num_tokens": 150121230.0, + "step": 5843 + }, + { + "epoch": 0.6417746540742367, + "grad_norm": 2.1960744857788086, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6970241069793701, + "num_tokens": 150147599.0, + "step": 5844 + }, + { + "epoch": 0.6418844717768504, + "grad_norm": 2.1013426780700684, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7003599405288696, + "num_tokens": 150175034.0, + "step": 5845 + }, + { + "epoch": 0.6419942894794641, + "grad_norm": 2.482799768447876, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6996283531188965, + "num_tokens": 150198355.0, + "step": 5846 + }, + { + "epoch": 0.6421041071820778, + "grad_norm": 2.025928258895874, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7315084934234619, + "num_tokens": 150224607.0, + "step": 5847 + }, + { + "epoch": 0.6422139248846914, + "grad_norm": 2.368576765060425, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7051085233688354, + "num_tokens": 150248278.0, + "step": 5848 + }, + { + "epoch": 0.6423237425873051, + "grad_norm": 2.0862936973571777, + "learning_rate": 1e-06, + "loss": 1.0843, + "mean_token_accuracy": 0.6737866401672363, + "num_tokens": 150282863.0, + "step": 5849 + }, + { + "epoch": 0.6424335602899187, + "grad_norm": 2.0477778911590576, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6920442581176758, + "num_tokens": 150311028.0, + "step": 5850 + }, + { + "epoch": 0.6425433779925324, + "grad_norm": 2.413498640060425, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7080071568489075, + "num_tokens": 150332202.0, + "step": 5851 + }, + { + "epoch": 0.642653195695146, + "grad_norm": 1.9366072416305542, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7183114290237427, + "num_tokens": 150360245.0, + "step": 5852 + }, + { + "epoch": 0.6427630133977598, + "grad_norm": 2.078423500061035, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7062695026397705, + "num_tokens": 150387399.0, + "step": 5853 + }, + { + "epoch": 0.6428728311003734, + "grad_norm": 2.351780891418457, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7135666608810425, + "num_tokens": 150409532.0, + "step": 5854 + }, + { + "epoch": 0.6429826488029871, + "grad_norm": 2.1072134971618652, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.713610827922821, + "num_tokens": 150436095.0, + "step": 5855 + }, + { + "epoch": 0.6430924665056007, + "grad_norm": 2.0875892639160156, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.7006538510322571, + "num_tokens": 150464590.0, + "step": 5856 + }, + { + "epoch": 0.6432022842082143, + "grad_norm": 1.9724719524383545, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7121924757957458, + "num_tokens": 150491612.0, + "step": 5857 + }, + { + "epoch": 0.643312101910828, + "grad_norm": 2.0169317722320557, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6892747282981873, + "num_tokens": 150520484.0, + "step": 5858 + }, + { + "epoch": 0.6434219196134416, + "grad_norm": 2.114814043045044, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6929692625999451, + "num_tokens": 150548623.0, + "step": 5859 + }, + { + "epoch": 0.6435317373160554, + "grad_norm": 2.528484344482422, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6891258955001831, + "num_tokens": 150569838.0, + "step": 5860 + }, + { + "epoch": 0.643641555018669, + "grad_norm": 2.267639398574829, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6907125115394592, + "num_tokens": 150594956.0, + "step": 5861 + }, + { + "epoch": 0.6437513727212827, + "grad_norm": 2.047719955444336, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6923242807388306, + "num_tokens": 150625464.0, + "step": 5862 + }, + { + "epoch": 0.6438611904238963, + "grad_norm": 2.273306369781494, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7096087336540222, + "num_tokens": 150650664.0, + "step": 5863 + }, + { + "epoch": 0.64397100812651, + "grad_norm": 2.0921242237091064, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7085602283477783, + "num_tokens": 150678819.0, + "step": 5864 + }, + { + "epoch": 0.6440808258291236, + "grad_norm": 1.9462999105453491, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.713702380657196, + "num_tokens": 150709137.0, + "step": 5865 + }, + { + "epoch": 0.6441906435317373, + "grad_norm": 2.229616641998291, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6941206455230713, + "num_tokens": 150735253.0, + "step": 5866 + }, + { + "epoch": 0.644300461234351, + "grad_norm": 1.9965351819992065, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7044064402580261, + "num_tokens": 150765668.0, + "step": 5867 + }, + { + "epoch": 0.6444102789369647, + "grad_norm": 2.367917776107788, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7233996987342834, + "num_tokens": 150788231.0, + "step": 5868 + }, + { + "epoch": 0.6445200966395783, + "grad_norm": 2.268282651901245, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7059661746025085, + "num_tokens": 150811122.0, + "step": 5869 + }, + { + "epoch": 0.644629914342192, + "grad_norm": 2.241387128829956, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6928941011428833, + "num_tokens": 150836444.0, + "step": 5870 + }, + { + "epoch": 0.6447397320448056, + "grad_norm": 2.2086870670318604, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.709065318107605, + "num_tokens": 150861753.0, + "step": 5871 + }, + { + "epoch": 0.6448495497474193, + "grad_norm": 2.2940313816070557, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7150343656539917, + "num_tokens": 150883888.0, + "step": 5872 + }, + { + "epoch": 0.6449593674500329, + "grad_norm": 2.372744083404541, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7205572128295898, + "num_tokens": 150906544.0, + "step": 5873 + }, + { + "epoch": 0.6450691851526466, + "grad_norm": 2.253770112991333, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7111650109291077, + "num_tokens": 150932235.0, + "step": 5874 + }, + { + "epoch": 0.6451790028552603, + "grad_norm": 2.145496368408203, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7196575403213501, + "num_tokens": 150959637.0, + "step": 5875 + }, + { + "epoch": 0.645288820557874, + "grad_norm": 2.204542875289917, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7287476062774658, + "num_tokens": 150982998.0, + "step": 5876 + }, + { + "epoch": 0.6453986382604876, + "grad_norm": 2.1012351512908936, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7030397653579712, + "num_tokens": 151009825.0, + "step": 5877 + }, + { + "epoch": 0.6455084559631012, + "grad_norm": 2.520318031311035, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7398896217346191, + "num_tokens": 151029111.0, + "step": 5878 + }, + { + "epoch": 0.6456182736657149, + "grad_norm": 1.9070736169815063, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6911921501159668, + "num_tokens": 151060360.0, + "step": 5879 + }, + { + "epoch": 0.6457280913683285, + "grad_norm": 2.4430785179138184, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7413078546524048, + "num_tokens": 151080216.0, + "step": 5880 + }, + { + "epoch": 0.6458379090709422, + "grad_norm": 2.061654806137085, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6840935945510864, + "num_tokens": 151108306.0, + "step": 5881 + }, + { + "epoch": 0.6459477267735559, + "grad_norm": 2.1273791790008545, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.686510443687439, + "num_tokens": 151136179.0, + "step": 5882 + }, + { + "epoch": 0.6460575444761696, + "grad_norm": 2.1617891788482666, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6960756778717041, + "num_tokens": 151161566.0, + "step": 5883 + }, + { + "epoch": 0.6461673621787832, + "grad_norm": 2.0817127227783203, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7041729688644409, + "num_tokens": 151190397.0, + "step": 5884 + }, + { + "epoch": 0.6462771798813969, + "grad_norm": 2.209425449371338, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7348161339759827, + "num_tokens": 151214725.0, + "step": 5885 + }, + { + "epoch": 0.6463869975840105, + "grad_norm": 2.1123220920562744, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7245970368385315, + "num_tokens": 151240319.0, + "step": 5886 + }, + { + "epoch": 0.6464968152866242, + "grad_norm": 2.1485817432403564, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6982823014259338, + "num_tokens": 151268659.0, + "step": 5887 + }, + { + "epoch": 0.6466066329892378, + "grad_norm": 2.212005138397217, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6997644901275635, + "num_tokens": 151292894.0, + "step": 5888 + }, + { + "epoch": 0.6467164506918516, + "grad_norm": 2.2230560779571533, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.705191969871521, + "num_tokens": 151317891.0, + "step": 5889 + }, + { + "epoch": 0.6468262683944652, + "grad_norm": 2.436652898788452, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6905652284622192, + "num_tokens": 151340785.0, + "step": 5890 + }, + { + "epoch": 0.6469360860970789, + "grad_norm": 2.252699375152588, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7086054086685181, + "num_tokens": 151363733.0, + "step": 5891 + }, + { + "epoch": 0.6470459037996925, + "grad_norm": 1.9467381238937378, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7031310796737671, + "num_tokens": 151393823.0, + "step": 5892 + }, + { + "epoch": 0.6471557215023062, + "grad_norm": 2.0419962406158447, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6945226192474365, + "num_tokens": 151422692.0, + "step": 5893 + }, + { + "epoch": 0.6472655392049198, + "grad_norm": 2.466552257537842, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7170877456665039, + "num_tokens": 151442217.0, + "step": 5894 + }, + { + "epoch": 0.6473753569075335, + "grad_norm": 2.0092623233795166, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.71763014793396, + "num_tokens": 151470770.0, + "step": 5895 + }, + { + "epoch": 0.6474851746101472, + "grad_norm": 2.0762393474578857, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6814419031143188, + "num_tokens": 151498618.0, + "step": 5896 + }, + { + "epoch": 0.6475949923127609, + "grad_norm": 2.4122495651245117, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7278110980987549, + "num_tokens": 151520767.0, + "step": 5897 + }, + { + "epoch": 0.6477048100153745, + "grad_norm": 2.1538658142089844, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6977006196975708, + "num_tokens": 151546627.0, + "step": 5898 + }, + { + "epoch": 0.6478146277179881, + "grad_norm": 2.0402791500091553, + "learning_rate": 1e-06, + "loss": 1.0734, + "mean_token_accuracy": 0.6787468791007996, + "num_tokens": 151576948.0, + "step": 5899 + }, + { + "epoch": 0.6479244454206018, + "grad_norm": 1.9566437005996704, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7090119123458862, + "num_tokens": 151607159.0, + "step": 5900 + }, + { + "epoch": 0.6480342631232154, + "grad_norm": 2.2946813106536865, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7102186679840088, + "num_tokens": 151630424.0, + "step": 5901 + }, + { + "epoch": 0.6481440808258291, + "grad_norm": 2.1243577003479004, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6977800130844116, + "num_tokens": 151657047.0, + "step": 5902 + }, + { + "epoch": 0.6482538985284428, + "grad_norm": 2.176225185394287, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7100498676300049, + "num_tokens": 151683480.0, + "step": 5903 + }, + { + "epoch": 0.6483637162310565, + "grad_norm": 1.825679063796997, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.6788054704666138, + "num_tokens": 151720614.0, + "step": 5904 + }, + { + "epoch": 0.6484735339336701, + "grad_norm": 2.0106606483459473, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6945480108261108, + "num_tokens": 151749062.0, + "step": 5905 + }, + { + "epoch": 0.6485833516362838, + "grad_norm": 2.4771130084991455, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7184019684791565, + "num_tokens": 151768912.0, + "step": 5906 + }, + { + "epoch": 0.6486931693388974, + "grad_norm": 2.2913079261779785, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6925132870674133, + "num_tokens": 151795168.0, + "step": 5907 + }, + { + "epoch": 0.6488029870415111, + "grad_norm": 2.2046775817871094, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7152724266052246, + "num_tokens": 151818926.0, + "step": 5908 + }, + { + "epoch": 0.6489128047441247, + "grad_norm": 2.153669834136963, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7095562219619751, + "num_tokens": 151845614.0, + "step": 5909 + }, + { + "epoch": 0.6490226224467384, + "grad_norm": 2.1739721298217773, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.725129246711731, + "num_tokens": 151870842.0, + "step": 5910 + }, + { + "epoch": 0.6491324401493521, + "grad_norm": 2.1750612258911133, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6824822425842285, + "num_tokens": 151897666.0, + "step": 5911 + }, + { + "epoch": 0.6492422578519658, + "grad_norm": 2.207111358642578, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6922537684440613, + "num_tokens": 151922986.0, + "step": 5912 + }, + { + "epoch": 0.6493520755545794, + "grad_norm": 1.7879348993301392, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.705987811088562, + "num_tokens": 151956766.0, + "step": 5913 + }, + { + "epoch": 0.6494618932571931, + "grad_norm": 1.9073431491851807, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7157642245292664, + "num_tokens": 151986781.0, + "step": 5914 + }, + { + "epoch": 0.6495717109598067, + "grad_norm": 2.2377495765686035, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7270017862319946, + "num_tokens": 152012892.0, + "step": 5915 + }, + { + "epoch": 0.6496815286624203, + "grad_norm": 2.1925203800201416, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7036628723144531, + "num_tokens": 152037212.0, + "step": 5916 + }, + { + "epoch": 0.649791346365034, + "grad_norm": 1.99183189868927, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6983870267868042, + "num_tokens": 152066608.0, + "step": 5917 + }, + { + "epoch": 0.6499011640676478, + "grad_norm": 2.1593234539031982, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6980389952659607, + "num_tokens": 152094643.0, + "step": 5918 + }, + { + "epoch": 0.6500109817702614, + "grad_norm": 2.4202446937561035, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7145711183547974, + "num_tokens": 152117278.0, + "step": 5919 + }, + { + "epoch": 0.650120799472875, + "grad_norm": 2.156336784362793, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7047451734542847, + "num_tokens": 152143032.0, + "step": 5920 + }, + { + "epoch": 0.6502306171754887, + "grad_norm": 2.2907462120056152, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7115967869758606, + "num_tokens": 152167762.0, + "step": 5921 + }, + { + "epoch": 0.6503404348781023, + "grad_norm": 2.140551805496216, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6928180456161499, + "num_tokens": 152194132.0, + "step": 5922 + }, + { + "epoch": 0.650450252580716, + "grad_norm": 2.0643768310546875, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7015436291694641, + "num_tokens": 152222974.0, + "step": 5923 + }, + { + "epoch": 0.6505600702833296, + "grad_norm": 2.296945333480835, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7329128384590149, + "num_tokens": 152244506.0, + "step": 5924 + }, + { + "epoch": 0.6506698879859434, + "grad_norm": 2.147855043411255, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6983199119567871, + "num_tokens": 152270026.0, + "step": 5925 + }, + { + "epoch": 0.650779705688557, + "grad_norm": 2.2806344032287598, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6962913274765015, + "num_tokens": 152295366.0, + "step": 5926 + }, + { + "epoch": 0.6508895233911707, + "grad_norm": 1.9341576099395752, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7070640921592712, + "num_tokens": 152325415.0, + "step": 5927 + }, + { + "epoch": 0.6509993410937843, + "grad_norm": 2.0528621673583984, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6875102519989014, + "num_tokens": 152353369.0, + "step": 5928 + }, + { + "epoch": 0.651109158796398, + "grad_norm": 1.9959412813186646, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6833574175834656, + "num_tokens": 152383008.0, + "step": 5929 + }, + { + "epoch": 0.6512189764990116, + "grad_norm": 2.0852928161621094, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6921789646148682, + "num_tokens": 152409907.0, + "step": 5930 + }, + { + "epoch": 0.6513287942016253, + "grad_norm": 2.181097984313965, + "learning_rate": 1e-06, + "loss": 1.0973, + "mean_token_accuracy": 0.6707013249397278, + "num_tokens": 152437546.0, + "step": 5931 + }, + { + "epoch": 0.651438611904239, + "grad_norm": 1.9074430465698242, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6974281072616577, + "num_tokens": 152468962.0, + "step": 5932 + }, + { + "epoch": 0.6515484296068527, + "grad_norm": 2.3469934463500977, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7014591693878174, + "num_tokens": 152490196.0, + "step": 5933 + }, + { + "epoch": 0.6516582473094663, + "grad_norm": 1.9870191812515259, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7060136795043945, + "num_tokens": 152519005.0, + "step": 5934 + }, + { + "epoch": 0.65176806501208, + "grad_norm": 2.156648874282837, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.7039724588394165, + "num_tokens": 152544875.0, + "step": 5935 + }, + { + "epoch": 0.6518778827146936, + "grad_norm": 2.513366937637329, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7321217060089111, + "num_tokens": 152563392.0, + "step": 5936 + }, + { + "epoch": 0.6519877004173072, + "grad_norm": 2.1998772621154785, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7087390422821045, + "num_tokens": 152586844.0, + "step": 5937 + }, + { + "epoch": 0.6520975181199209, + "grad_norm": 2.1356201171875, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6847825646400452, + "num_tokens": 152614442.0, + "step": 5938 + }, + { + "epoch": 0.6522073358225345, + "grad_norm": 2.6610114574432373, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7230558395385742, + "num_tokens": 152632184.0, + "step": 5939 + }, + { + "epoch": 0.6523171535251483, + "grad_norm": 2.2223119735717773, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6981542110443115, + "num_tokens": 152658346.0, + "step": 5940 + }, + { + "epoch": 0.6524269712277619, + "grad_norm": 2.263139247894287, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7077375650405884, + "num_tokens": 152684690.0, + "step": 5941 + }, + { + "epoch": 0.6525367889303756, + "grad_norm": 2.4658994674682617, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6830819249153137, + "num_tokens": 152706600.0, + "step": 5942 + }, + { + "epoch": 0.6526466066329892, + "grad_norm": 2.2014989852905273, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6893495321273804, + "num_tokens": 152734191.0, + "step": 5943 + }, + { + "epoch": 0.6527564243356029, + "grad_norm": 2.0835800170898438, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7063998579978943, + "num_tokens": 152762199.0, + "step": 5944 + }, + { + "epoch": 0.6528662420382165, + "grad_norm": 2.0208637714385986, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7078814506530762, + "num_tokens": 152789010.0, + "step": 5945 + }, + { + "epoch": 0.6529760597408302, + "grad_norm": 2.1254048347473145, + "learning_rate": 1e-06, + "loss": 1.0633, + "mean_token_accuracy": 0.677081823348999, + "num_tokens": 152816063.0, + "step": 5946 + }, + { + "epoch": 0.6530858774434439, + "grad_norm": 2.191408395767212, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7077658176422119, + "num_tokens": 152840972.0, + "step": 5947 + }, + { + "epoch": 0.6531956951460576, + "grad_norm": 2.3538658618927, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7148087024688721, + "num_tokens": 152863342.0, + "step": 5948 + }, + { + "epoch": 0.6533055128486712, + "grad_norm": 2.2283036708831787, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.702381432056427, + "num_tokens": 152888883.0, + "step": 5949 + }, + { + "epoch": 0.6534153305512849, + "grad_norm": 2.4122862815856934, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7052549123764038, + "num_tokens": 152913395.0, + "step": 5950 + }, + { + "epoch": 0.6535251482538985, + "grad_norm": 2.507575750350952, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7172402143478394, + "num_tokens": 152934026.0, + "step": 5951 + }, + { + "epoch": 0.6536349659565122, + "grad_norm": 2.074312210083008, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7179926633834839, + "num_tokens": 152961165.0, + "step": 5952 + }, + { + "epoch": 0.6537447836591258, + "grad_norm": 2.171311855316162, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7214081287384033, + "num_tokens": 152986690.0, + "step": 5953 + }, + { + "epoch": 0.6538546013617396, + "grad_norm": 2.2379374504089355, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7039153575897217, + "num_tokens": 153012392.0, + "step": 5954 + }, + { + "epoch": 0.6539644190643532, + "grad_norm": 2.1760833263397217, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7094410061836243, + "num_tokens": 153038669.0, + "step": 5955 + }, + { + "epoch": 0.6540742367669669, + "grad_norm": 2.5068349838256836, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7142679691314697, + "num_tokens": 153058923.0, + "step": 5956 + }, + { + "epoch": 0.6541840544695805, + "grad_norm": 2.1510813236236572, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7303298711776733, + "num_tokens": 153082831.0, + "step": 5957 + }, + { + "epoch": 0.6542938721721941, + "grad_norm": 2.0555896759033203, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6828585863113403, + "num_tokens": 153110032.0, + "step": 5958 + }, + { + "epoch": 0.6544036898748078, + "grad_norm": 2.2040507793426514, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7031534314155579, + "num_tokens": 153135339.0, + "step": 5959 + }, + { + "epoch": 0.6545135075774214, + "grad_norm": 2.5247902870178223, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7352709770202637, + "num_tokens": 153153923.0, + "step": 5960 + }, + { + "epoch": 0.6546233252800352, + "grad_norm": 2.0359795093536377, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7057033777236938, + "num_tokens": 153182093.0, + "step": 5961 + }, + { + "epoch": 0.6547331429826488, + "grad_norm": 2.0145931243896484, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6934101581573486, + "num_tokens": 153211665.0, + "step": 5962 + }, + { + "epoch": 0.6548429606852625, + "grad_norm": 2.1648902893066406, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7166591882705688, + "num_tokens": 153238146.0, + "step": 5963 + }, + { + "epoch": 0.6549527783878761, + "grad_norm": 2.191546678543091, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.689018726348877, + "num_tokens": 153262470.0, + "step": 5964 + }, + { + "epoch": 0.6550625960904898, + "grad_norm": 1.901597499847412, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7033898830413818, + "num_tokens": 153295269.0, + "step": 5965 + }, + { + "epoch": 0.6551724137931034, + "grad_norm": 2.1327667236328125, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7159956693649292, + "num_tokens": 153321349.0, + "step": 5966 + }, + { + "epoch": 0.6552822314957171, + "grad_norm": 2.019529104232788, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7068793773651123, + "num_tokens": 153348111.0, + "step": 5967 + }, + { + "epoch": 0.6553920491983307, + "grad_norm": 2.141570806503296, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.701189398765564, + "num_tokens": 153374694.0, + "step": 5968 + }, + { + "epoch": 0.6555018669009445, + "grad_norm": 2.3245983123779297, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7313970923423767, + "num_tokens": 153396835.0, + "step": 5969 + }, + { + "epoch": 0.6556116846035581, + "grad_norm": 2.413450241088867, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7192704081535339, + "num_tokens": 153416050.0, + "step": 5970 + }, + { + "epoch": 0.6557215023061718, + "grad_norm": 2.429974317550659, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7082066535949707, + "num_tokens": 153439029.0, + "step": 5971 + }, + { + "epoch": 0.6558313200087854, + "grad_norm": 2.0683064460754395, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7308003306388855, + "num_tokens": 153464931.0, + "step": 5972 + }, + { + "epoch": 0.6559411377113991, + "grad_norm": 2.230458974838257, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6996481418609619, + "num_tokens": 153490458.0, + "step": 5973 + }, + { + "epoch": 0.6560509554140127, + "grad_norm": 2.1812524795532227, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.6993569135665894, + "num_tokens": 153515699.0, + "step": 5974 + }, + { + "epoch": 0.6561607731166264, + "grad_norm": 2.115706443786621, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7067757844924927, + "num_tokens": 153541955.0, + "step": 5975 + }, + { + "epoch": 0.6562705908192401, + "grad_norm": 2.1775577068328857, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.6801584959030151, + "num_tokens": 153569135.0, + "step": 5976 + }, + { + "epoch": 0.6563804085218538, + "grad_norm": 2.3448362350463867, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7261605262756348, + "num_tokens": 153594303.0, + "step": 5977 + }, + { + "epoch": 0.6564902262244674, + "grad_norm": 2.032484292984009, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6855882406234741, + "num_tokens": 153622473.0, + "step": 5978 + }, + { + "epoch": 0.656600043927081, + "grad_norm": 2.075237274169922, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7052829265594482, + "num_tokens": 153649968.0, + "step": 5979 + }, + { + "epoch": 0.6567098616296947, + "grad_norm": 2.152895212173462, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7258110046386719, + "num_tokens": 153674818.0, + "step": 5980 + }, + { + "epoch": 0.6568196793323083, + "grad_norm": 2.387120246887207, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6961439251899719, + "num_tokens": 153697090.0, + "step": 5981 + }, + { + "epoch": 0.656929497034922, + "grad_norm": 2.075788974761963, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7289692759513855, + "num_tokens": 153722829.0, + "step": 5982 + }, + { + "epoch": 0.6570393147375357, + "grad_norm": 2.0610876083374023, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.708827018737793, + "num_tokens": 153750786.0, + "step": 5983 + }, + { + "epoch": 0.6571491324401494, + "grad_norm": 2.127995729446411, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6839457750320435, + "num_tokens": 153778009.0, + "step": 5984 + }, + { + "epoch": 0.657258950142763, + "grad_norm": 2.248581647872925, + "learning_rate": 1e-06, + "loss": 1.0656, + "mean_token_accuracy": 0.6767795085906982, + "num_tokens": 153803362.0, + "step": 5985 + }, + { + "epoch": 0.6573687678453767, + "grad_norm": 2.6197457313537598, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.734153687953949, + "num_tokens": 153821896.0, + "step": 5986 + }, + { + "epoch": 0.6574785855479903, + "grad_norm": 2.155226707458496, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7240647077560425, + "num_tokens": 153847442.0, + "step": 5987 + }, + { + "epoch": 0.657588403250604, + "grad_norm": 2.1869006156921387, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7258262634277344, + "num_tokens": 153872223.0, + "step": 5988 + }, + { + "epoch": 0.6576982209532176, + "grad_norm": 2.339620351791382, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7096551656723022, + "num_tokens": 153893960.0, + "step": 5989 + }, + { + "epoch": 0.6578080386558314, + "grad_norm": 2.360825777053833, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7053617238998413, + "num_tokens": 153916464.0, + "step": 5990 + }, + { + "epoch": 0.657917856358445, + "grad_norm": 2.3829503059387207, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6989688873291016, + "num_tokens": 153940118.0, + "step": 5991 + }, + { + "epoch": 0.6580276740610587, + "grad_norm": 2.201335906982422, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6968568563461304, + "num_tokens": 153965261.0, + "step": 5992 + }, + { + "epoch": 0.6581374917636723, + "grad_norm": 2.0113861560821533, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7359962463378906, + "num_tokens": 153992840.0, + "step": 5993 + }, + { + "epoch": 0.658247309466286, + "grad_norm": 2.495654344558716, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7085434794425964, + "num_tokens": 154012285.0, + "step": 5994 + }, + { + "epoch": 0.6583571271688996, + "grad_norm": 1.9904897212982178, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6991849541664124, + "num_tokens": 154041398.0, + "step": 5995 + }, + { + "epoch": 0.6584669448715132, + "grad_norm": 2.0762131214141846, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7018513083457947, + "num_tokens": 154069167.0, + "step": 5996 + }, + { + "epoch": 0.6585767625741269, + "grad_norm": 2.1620326042175293, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.727110743522644, + "num_tokens": 154094579.0, + "step": 5997 + }, + { + "epoch": 0.6586865802767407, + "grad_norm": 2.2831127643585205, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7022080421447754, + "num_tokens": 154118135.0, + "step": 5998 + }, + { + "epoch": 0.6587963979793543, + "grad_norm": 1.8900346755981445, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7120071649551392, + "num_tokens": 154150169.0, + "step": 5999 + }, + { + "epoch": 0.6589062156819679, + "grad_norm": 1.9986802339553833, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7210903167724609, + "num_tokens": 154176808.0, + "step": 6000 + }, + { + "epoch": 0.6590160333845816, + "grad_norm": 2.1363465785980225, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.6835413575172424, + "num_tokens": 154202918.0, + "step": 6001 + }, + { + "epoch": 0.6591258510871952, + "grad_norm": 2.2957558631896973, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.707668662071228, + "num_tokens": 154225611.0, + "step": 6002 + }, + { + "epoch": 0.6592356687898089, + "grad_norm": 2.4128386974334717, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7125461101531982, + "num_tokens": 154246769.0, + "step": 6003 + }, + { + "epoch": 0.6593454864924225, + "grad_norm": 2.4085896015167236, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7054516077041626, + "num_tokens": 154267534.0, + "step": 6004 + }, + { + "epoch": 0.6594553041950363, + "grad_norm": 2.198176383972168, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7176587581634521, + "num_tokens": 154290898.0, + "step": 6005 + }, + { + "epoch": 0.6595651218976499, + "grad_norm": 2.3255105018615723, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7269463539123535, + "num_tokens": 154312523.0, + "step": 6006 + }, + { + "epoch": 0.6596749396002636, + "grad_norm": 2.18544602394104, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7050352096557617, + "num_tokens": 154337745.0, + "step": 6007 + }, + { + "epoch": 0.6597847573028772, + "grad_norm": 2.1140780448913574, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.69573575258255, + "num_tokens": 154365574.0, + "step": 6008 + }, + { + "epoch": 0.6598945750054909, + "grad_norm": 2.4680094718933105, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7091742157936096, + "num_tokens": 154386834.0, + "step": 6009 + }, + { + "epoch": 0.6600043927081045, + "grad_norm": 2.347648859024048, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6925445795059204, + "num_tokens": 154409888.0, + "step": 6010 + }, + { + "epoch": 0.6601142104107182, + "grad_norm": 2.0668082237243652, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7229588031768799, + "num_tokens": 154435501.0, + "step": 6011 + }, + { + "epoch": 0.6602240281133319, + "grad_norm": 2.2765800952911377, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6901095509529114, + "num_tokens": 154459741.0, + "step": 6012 + }, + { + "epoch": 0.6603338458159456, + "grad_norm": 2.4130709171295166, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7092888355255127, + "num_tokens": 154483178.0, + "step": 6013 + }, + { + "epoch": 0.6604436635185592, + "grad_norm": 2.02252459526062, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6902542114257812, + "num_tokens": 154515279.0, + "step": 6014 + }, + { + "epoch": 0.6605534812211729, + "grad_norm": 2.2486371994018555, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6992488503456116, + "num_tokens": 154540890.0, + "step": 6015 + }, + { + "epoch": 0.6606632989237865, + "grad_norm": 2.270127534866333, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6944193840026855, + "num_tokens": 154566397.0, + "step": 6016 + }, + { + "epoch": 0.6607731166264001, + "grad_norm": 2.3374884128570557, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7039521932601929, + "num_tokens": 154588391.0, + "step": 6017 + }, + { + "epoch": 0.6608829343290138, + "grad_norm": 2.14064359664917, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.709585964679718, + "num_tokens": 154612664.0, + "step": 6018 + }, + { + "epoch": 0.6609927520316276, + "grad_norm": 2.414604663848877, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7185778617858887, + "num_tokens": 154634522.0, + "step": 6019 + }, + { + "epoch": 0.6611025697342412, + "grad_norm": 1.947690486907959, + "learning_rate": 1e-06, + "loss": 1.1057, + "mean_token_accuracy": 0.6722012162208557, + "num_tokens": 154669152.0, + "step": 6020 + }, + { + "epoch": 0.6612123874368548, + "grad_norm": 2.0581560134887695, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7056354284286499, + "num_tokens": 154697314.0, + "step": 6021 + }, + { + "epoch": 0.6613222051394685, + "grad_norm": 2.2239742279052734, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7136164307594299, + "num_tokens": 154722531.0, + "step": 6022 + }, + { + "epoch": 0.6614320228420821, + "grad_norm": 2.2732126712799072, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7099066376686096, + "num_tokens": 154746487.0, + "step": 6023 + }, + { + "epoch": 0.6615418405446958, + "grad_norm": 2.160033702850342, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7017362713813782, + "num_tokens": 154773362.0, + "step": 6024 + }, + { + "epoch": 0.6616516582473094, + "grad_norm": 2.2412638664245605, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7265880703926086, + "num_tokens": 154796347.0, + "step": 6025 + }, + { + "epoch": 0.6617614759499231, + "grad_norm": 2.146815299987793, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7186711430549622, + "num_tokens": 154821851.0, + "step": 6026 + }, + { + "epoch": 0.6618712936525368, + "grad_norm": 2.208956718444824, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.7057943344116211, + "num_tokens": 154847088.0, + "step": 6027 + }, + { + "epoch": 0.6619811113551505, + "grad_norm": 2.408992052078247, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6953722238540649, + "num_tokens": 154869709.0, + "step": 6028 + }, + { + "epoch": 0.6620909290577641, + "grad_norm": 2.2095935344696045, + "learning_rate": 1e-06, + "loss": 1.0913, + "mean_token_accuracy": 0.6747864484786987, + "num_tokens": 154897124.0, + "step": 6029 + }, + { + "epoch": 0.6622007467603778, + "grad_norm": 2.138091564178467, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.69911789894104, + "num_tokens": 154923625.0, + "step": 6030 + }, + { + "epoch": 0.6623105644629914, + "grad_norm": 2.420762777328491, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7225649356842041, + "num_tokens": 154943797.0, + "step": 6031 + }, + { + "epoch": 0.6624203821656051, + "grad_norm": 2.212695598602295, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.714956521987915, + "num_tokens": 154967631.0, + "step": 6032 + }, + { + "epoch": 0.6625301998682187, + "grad_norm": 2.2444040775299072, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6908365488052368, + "num_tokens": 154991092.0, + "step": 6033 + }, + { + "epoch": 0.6626400175708325, + "grad_norm": 1.8632949590682983, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7114742398262024, + "num_tokens": 155024903.0, + "step": 6034 + }, + { + "epoch": 0.6627498352734461, + "grad_norm": 2.267599582672119, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6983137130737305, + "num_tokens": 155049471.0, + "step": 6035 + }, + { + "epoch": 0.6628596529760598, + "grad_norm": 2.0934746265411377, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7276287078857422, + "num_tokens": 155075555.0, + "step": 6036 + }, + { + "epoch": 0.6629694706786734, + "grad_norm": 2.233353853225708, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6967719197273254, + "num_tokens": 155101349.0, + "step": 6037 + }, + { + "epoch": 0.663079288381287, + "grad_norm": 1.9704303741455078, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7266393303871155, + "num_tokens": 155129504.0, + "step": 6038 + }, + { + "epoch": 0.6631891060839007, + "grad_norm": 2.3450801372528076, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7362807989120483, + "num_tokens": 155151042.0, + "step": 6039 + }, + { + "epoch": 0.6632989237865143, + "grad_norm": 2.126981258392334, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.6847205758094788, + "num_tokens": 155179946.0, + "step": 6040 + }, + { + "epoch": 0.6634087414891281, + "grad_norm": 2.464320659637451, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7204959392547607, + "num_tokens": 155198795.0, + "step": 6041 + }, + { + "epoch": 0.6635185591917417, + "grad_norm": 1.9668971300125122, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.704639196395874, + "num_tokens": 155231707.0, + "step": 6042 + }, + { + "epoch": 0.6636283768943554, + "grad_norm": 2.593470335006714, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.725696325302124, + "num_tokens": 155250122.0, + "step": 6043 + }, + { + "epoch": 0.663738194596969, + "grad_norm": 2.4177489280700684, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7075943946838379, + "num_tokens": 155271493.0, + "step": 6044 + }, + { + "epoch": 0.6638480122995827, + "grad_norm": 2.464688777923584, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7259320616722107, + "num_tokens": 155292370.0, + "step": 6045 + }, + { + "epoch": 0.6639578300021963, + "grad_norm": 2.1637532711029053, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.7100870609283447, + "num_tokens": 155319585.0, + "step": 6046 + }, + { + "epoch": 0.66406764770481, + "grad_norm": 2.2677438259124756, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7181429266929626, + "num_tokens": 155342963.0, + "step": 6047 + }, + { + "epoch": 0.6641774654074237, + "grad_norm": 2.090512990951538, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7032124996185303, + "num_tokens": 155370530.0, + "step": 6048 + }, + { + "epoch": 0.6642872831100374, + "grad_norm": 2.876159906387329, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7236295938491821, + "num_tokens": 155387793.0, + "step": 6049 + }, + { + "epoch": 0.664397100812651, + "grad_norm": 2.141113519668579, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6937271356582642, + "num_tokens": 155413657.0, + "step": 6050 + }, + { + "epoch": 0.6645069185152647, + "grad_norm": 2.176335573196411, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7148252129554749, + "num_tokens": 155439202.0, + "step": 6051 + }, + { + "epoch": 0.6646167362178783, + "grad_norm": 2.0146901607513428, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.701850175857544, + "num_tokens": 155468262.0, + "step": 6052 + }, + { + "epoch": 0.664726553920492, + "grad_norm": 2.103444814682007, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.714849054813385, + "num_tokens": 155494121.0, + "step": 6053 + }, + { + "epoch": 0.6648363716231056, + "grad_norm": 2.1524240970611572, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7217886447906494, + "num_tokens": 155517156.0, + "step": 6054 + }, + { + "epoch": 0.6649461893257194, + "grad_norm": 2.153139591217041, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6817508339881897, + "num_tokens": 155546872.0, + "step": 6055 + }, + { + "epoch": 0.665056007028333, + "grad_norm": 1.8588894605636597, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6997827291488647, + "num_tokens": 155582161.0, + "step": 6056 + }, + { + "epoch": 0.6651658247309467, + "grad_norm": 2.4984607696533203, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7255814075469971, + "num_tokens": 155600731.0, + "step": 6057 + }, + { + "epoch": 0.6652756424335603, + "grad_norm": 2.2866721153259277, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7245043516159058, + "num_tokens": 155621767.0, + "step": 6058 + }, + { + "epoch": 0.665385460136174, + "grad_norm": 2.235170364379883, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7087489366531372, + "num_tokens": 155645143.0, + "step": 6059 + }, + { + "epoch": 0.6654952778387876, + "grad_norm": 2.2488975524902344, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7278182506561279, + "num_tokens": 155667349.0, + "step": 6060 + }, + { + "epoch": 0.6656050955414012, + "grad_norm": 2.0510032176971436, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7242535352706909, + "num_tokens": 155693747.0, + "step": 6061 + }, + { + "epoch": 0.6657149132440149, + "grad_norm": 1.9618635177612305, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6789790391921997, + "num_tokens": 155724227.0, + "step": 6062 + }, + { + "epoch": 0.6658247309466286, + "grad_norm": 2.141413688659668, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6909907460212708, + "num_tokens": 155751449.0, + "step": 6063 + }, + { + "epoch": 0.6659345486492423, + "grad_norm": 2.3306190967559814, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7127695083618164, + "num_tokens": 155774268.0, + "step": 6064 + }, + { + "epoch": 0.6660443663518559, + "grad_norm": 2.4334821701049805, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7310097217559814, + "num_tokens": 155796293.0, + "step": 6065 + }, + { + "epoch": 0.6661541840544696, + "grad_norm": 2.189222812652588, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7275886535644531, + "num_tokens": 155820421.0, + "step": 6066 + }, + { + "epoch": 0.6662640017570832, + "grad_norm": 2.297102212905884, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7095769643783569, + "num_tokens": 155843741.0, + "step": 6067 + }, + { + "epoch": 0.6663738194596969, + "grad_norm": 2.471231698989868, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.738489031791687, + "num_tokens": 155864184.0, + "step": 6068 + }, + { + "epoch": 0.6664836371623105, + "grad_norm": 2.343447208404541, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6852750182151794, + "num_tokens": 155887678.0, + "step": 6069 + }, + { + "epoch": 0.6665934548649243, + "grad_norm": 1.9294238090515137, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.697590708732605, + "num_tokens": 155916920.0, + "step": 6070 + }, + { + "epoch": 0.6667032725675379, + "grad_norm": 2.18865704536438, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.726447582244873, + "num_tokens": 155941062.0, + "step": 6071 + }, + { + "epoch": 0.6668130902701516, + "grad_norm": 1.9013950824737549, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6963120698928833, + "num_tokens": 155973604.0, + "step": 6072 + }, + { + "epoch": 0.6669229079727652, + "grad_norm": 1.9108201265335083, + "learning_rate": 1e-06, + "loss": 1.1042, + "mean_token_accuracy": 0.6675754189491272, + "num_tokens": 156007947.0, + "step": 6073 + }, + { + "epoch": 0.6670327256753789, + "grad_norm": 2.1210286617279053, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7072262763977051, + "num_tokens": 156036040.0, + "step": 6074 + }, + { + "epoch": 0.6671425433779925, + "grad_norm": 2.1357340812683105, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6903481483459473, + "num_tokens": 156063596.0, + "step": 6075 + }, + { + "epoch": 0.6672523610806061, + "grad_norm": 2.2244300842285156, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.712088942527771, + "num_tokens": 156088377.0, + "step": 6076 + }, + { + "epoch": 0.6673621787832199, + "grad_norm": 2.0650980472564697, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7139722108840942, + "num_tokens": 156116570.0, + "step": 6077 + }, + { + "epoch": 0.6674719964858336, + "grad_norm": 2.3988046646118164, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6955084800720215, + "num_tokens": 156139372.0, + "step": 6078 + }, + { + "epoch": 0.6675818141884472, + "grad_norm": 2.235915184020996, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7064470052719116, + "num_tokens": 156163156.0, + "step": 6079 + }, + { + "epoch": 0.6676916318910608, + "grad_norm": 2.1336357593536377, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6904046535491943, + "num_tokens": 156189639.0, + "step": 6080 + }, + { + "epoch": 0.6678014495936745, + "grad_norm": 2.2238543033599854, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7057117223739624, + "num_tokens": 156213919.0, + "step": 6081 + }, + { + "epoch": 0.6679112672962881, + "grad_norm": 2.3055601119995117, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6973234415054321, + "num_tokens": 156238890.0, + "step": 6082 + }, + { + "epoch": 0.6680210849989018, + "grad_norm": 2.3293955326080322, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7014713287353516, + "num_tokens": 156263209.0, + "step": 6083 + }, + { + "epoch": 0.6681309027015155, + "grad_norm": 2.21983003616333, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7142947316169739, + "num_tokens": 156288743.0, + "step": 6084 + }, + { + "epoch": 0.6682407204041292, + "grad_norm": 2.0549044609069824, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6924505233764648, + "num_tokens": 156318537.0, + "step": 6085 + }, + { + "epoch": 0.6683505381067428, + "grad_norm": 2.280548095703125, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7152082324028015, + "num_tokens": 156343227.0, + "step": 6086 + }, + { + "epoch": 0.6684603558093565, + "grad_norm": 1.9013981819152832, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.702707827091217, + "num_tokens": 156375057.0, + "step": 6087 + }, + { + "epoch": 0.6685701735119701, + "grad_norm": 2.0540826320648193, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7142788171768188, + "num_tokens": 156402713.0, + "step": 6088 + }, + { + "epoch": 0.6686799912145838, + "grad_norm": 2.289944648742676, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7174997329711914, + "num_tokens": 156425465.0, + "step": 6089 + }, + { + "epoch": 0.6687898089171974, + "grad_norm": 2.000087261199951, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7003167867660522, + "num_tokens": 156456526.0, + "step": 6090 + }, + { + "epoch": 0.6688996266198111, + "grad_norm": 2.005937337875366, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6927121877670288, + "num_tokens": 156487388.0, + "step": 6091 + }, + { + "epoch": 0.6690094443224248, + "grad_norm": 2.185244560241699, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7124610543251038, + "num_tokens": 156512688.0, + "step": 6092 + }, + { + "epoch": 0.6691192620250385, + "grad_norm": 2.2533154487609863, + "learning_rate": 1e-06, + "loss": 1.0495, + "mean_token_accuracy": 0.6860960721969604, + "num_tokens": 156537824.0, + "step": 6093 + }, + { + "epoch": 0.6692290797276521, + "grad_norm": 2.0281994342803955, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7223078608512878, + "num_tokens": 156565352.0, + "step": 6094 + }, + { + "epoch": 0.6693388974302658, + "grad_norm": 2.2003722190856934, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7364141345024109, + "num_tokens": 156588637.0, + "step": 6095 + }, + { + "epoch": 0.6694487151328794, + "grad_norm": 2.11706280708313, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6899212598800659, + "num_tokens": 156618376.0, + "step": 6096 + }, + { + "epoch": 0.669558532835493, + "grad_norm": 2.0601871013641357, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7109769582748413, + "num_tokens": 156645973.0, + "step": 6097 + }, + { + "epoch": 0.6696683505381067, + "grad_norm": 1.9826905727386475, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6976467370986938, + "num_tokens": 156675812.0, + "step": 6098 + }, + { + "epoch": 0.6697781682407205, + "grad_norm": 2.0485265254974365, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6886858940124512, + "num_tokens": 156706105.0, + "step": 6099 + }, + { + "epoch": 0.6698879859433341, + "grad_norm": 2.232905864715576, + "learning_rate": 1e-06, + "loss": 1.0798, + "mean_token_accuracy": 0.6683961153030396, + "num_tokens": 156733097.0, + "step": 6100 + }, + { + "epoch": 0.6699978036459477, + "grad_norm": 2.0436203479766846, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6969264149665833, + "num_tokens": 156760309.0, + "step": 6101 + }, + { + "epoch": 0.6701076213485614, + "grad_norm": 2.331087827682495, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7206284403800964, + "num_tokens": 156782700.0, + "step": 6102 + }, + { + "epoch": 0.670217439051175, + "grad_norm": 2.3470282554626465, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.717915952205658, + "num_tokens": 156804634.0, + "step": 6103 + }, + { + "epoch": 0.6703272567537887, + "grad_norm": 2.172288417816162, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.706039547920227, + "num_tokens": 156829762.0, + "step": 6104 + }, + { + "epoch": 0.6704370744564023, + "grad_norm": 2.544213056564331, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7391417026519775, + "num_tokens": 156849669.0, + "step": 6105 + }, + { + "epoch": 0.6705468921590161, + "grad_norm": 2.433025360107422, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7132745981216431, + "num_tokens": 156869829.0, + "step": 6106 + }, + { + "epoch": 0.6706567098616297, + "grad_norm": 2.4205639362335205, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7210277318954468, + "num_tokens": 156891519.0, + "step": 6107 + }, + { + "epoch": 0.6707665275642434, + "grad_norm": 2.333096742630005, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7034340500831604, + "num_tokens": 156914250.0, + "step": 6108 + }, + { + "epoch": 0.670876345266857, + "grad_norm": 2.242419719696045, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6935027837753296, + "num_tokens": 156938592.0, + "step": 6109 + }, + { + "epoch": 0.6709861629694707, + "grad_norm": 2.419241189956665, + "learning_rate": 1e-06, + "loss": 1.07, + "mean_token_accuracy": 0.684424102306366, + "num_tokens": 156960777.0, + "step": 6110 + }, + { + "epoch": 0.6710959806720843, + "grad_norm": 2.056340456008911, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6968389749526978, + "num_tokens": 156989229.0, + "step": 6111 + }, + { + "epoch": 0.671205798374698, + "grad_norm": 2.1764323711395264, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7007207274436951, + "num_tokens": 157017772.0, + "step": 6112 + }, + { + "epoch": 0.6713156160773117, + "grad_norm": 2.1655941009521484, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6933202743530273, + "num_tokens": 157044109.0, + "step": 6113 + }, + { + "epoch": 0.6714254337799254, + "grad_norm": 2.3543896675109863, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7177225351333618, + "num_tokens": 157065667.0, + "step": 6114 + }, + { + "epoch": 0.671535251482539, + "grad_norm": 2.4529592990875244, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7189284563064575, + "num_tokens": 157086280.0, + "step": 6115 + }, + { + "epoch": 0.6716450691851527, + "grad_norm": 2.1488282680511475, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7212597131729126, + "num_tokens": 157111236.0, + "step": 6116 + }, + { + "epoch": 0.6717548868877663, + "grad_norm": 2.1488280296325684, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7074733972549438, + "num_tokens": 157136525.0, + "step": 6117 + }, + { + "epoch": 0.67186470459038, + "grad_norm": 2.0837600231170654, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6985068917274475, + "num_tokens": 157165648.0, + "step": 6118 + }, + { + "epoch": 0.6719745222929936, + "grad_norm": 2.1444315910339355, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7059142589569092, + "num_tokens": 157190453.0, + "step": 6119 + }, + { + "epoch": 0.6720843399956072, + "grad_norm": 1.8984752893447876, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7116544842720032, + "num_tokens": 157222268.0, + "step": 6120 + }, + { + "epoch": 0.672194157698221, + "grad_norm": 2.117455244064331, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7104478478431702, + "num_tokens": 157248312.0, + "step": 6121 + }, + { + "epoch": 0.6723039754008346, + "grad_norm": 2.18054461479187, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7171893119812012, + "num_tokens": 157271885.0, + "step": 6122 + }, + { + "epoch": 0.6724137931034483, + "grad_norm": 2.2229418754577637, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7007701396942139, + "num_tokens": 157298518.0, + "step": 6123 + }, + { + "epoch": 0.6725236108060619, + "grad_norm": 2.206252098083496, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7261710166931152, + "num_tokens": 157320942.0, + "step": 6124 + }, + { + "epoch": 0.6726334285086756, + "grad_norm": 2.0899744033813477, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7036139965057373, + "num_tokens": 157348327.0, + "step": 6125 + }, + { + "epoch": 0.6727432462112892, + "grad_norm": 2.545544147491455, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7277398705482483, + "num_tokens": 157366933.0, + "step": 6126 + }, + { + "epoch": 0.6728530639139029, + "grad_norm": 2.1595895290374756, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7275938987731934, + "num_tokens": 157391184.0, + "step": 6127 + }, + { + "epoch": 0.6729628816165166, + "grad_norm": 2.346733808517456, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7265152335166931, + "num_tokens": 157413822.0, + "step": 6128 + }, + { + "epoch": 0.6730726993191303, + "grad_norm": 2.0625953674316406, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7067139148712158, + "num_tokens": 157441681.0, + "step": 6129 + }, + { + "epoch": 0.6731825170217439, + "grad_norm": 2.2483506202697754, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6959778070449829, + "num_tokens": 157466652.0, + "step": 6130 + }, + { + "epoch": 0.6732923347243576, + "grad_norm": 2.2102975845336914, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7441152334213257, + "num_tokens": 157489369.0, + "step": 6131 + }, + { + "epoch": 0.6734021524269712, + "grad_norm": 2.3223235607147217, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6974287629127502, + "num_tokens": 157512744.0, + "step": 6132 + }, + { + "epoch": 0.6735119701295849, + "grad_norm": 2.6164913177490234, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.719326913356781, + "num_tokens": 157531441.0, + "step": 6133 + }, + { + "epoch": 0.6736217878321985, + "grad_norm": 2.128187417984009, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6927839517593384, + "num_tokens": 157560328.0, + "step": 6134 + }, + { + "epoch": 0.6737316055348123, + "grad_norm": 2.0672218799591064, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7192472219467163, + "num_tokens": 157587743.0, + "step": 6135 + }, + { + "epoch": 0.6738414232374259, + "grad_norm": 2.3800241947174072, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7077750563621521, + "num_tokens": 157611262.0, + "step": 6136 + }, + { + "epoch": 0.6739512409400396, + "grad_norm": 2.3977694511413574, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.681553840637207, + "num_tokens": 157634005.0, + "step": 6137 + }, + { + "epoch": 0.6740610586426532, + "grad_norm": 2.4369781017303467, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7222431302070618, + "num_tokens": 157655076.0, + "step": 6138 + }, + { + "epoch": 0.6741708763452668, + "grad_norm": 2.2472102642059326, + "learning_rate": 1e-06, + "loss": 1.0808, + "mean_token_accuracy": 0.6776971817016602, + "num_tokens": 157679814.0, + "step": 6139 + }, + { + "epoch": 0.6742806940478805, + "grad_norm": 2.2422072887420654, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6930567622184753, + "num_tokens": 157705063.0, + "step": 6140 + }, + { + "epoch": 0.6743905117504941, + "grad_norm": 2.1610686779022217, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7309561371803284, + "num_tokens": 157730941.0, + "step": 6141 + }, + { + "epoch": 0.6745003294531079, + "grad_norm": 2.1038835048675537, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6852196455001831, + "num_tokens": 157762172.0, + "step": 6142 + }, + { + "epoch": 0.6746101471557215, + "grad_norm": 2.1823348999023438, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.685234785079956, + "num_tokens": 157787658.0, + "step": 6143 + }, + { + "epoch": 0.6747199648583352, + "grad_norm": 2.4174184799194336, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7040741443634033, + "num_tokens": 157810140.0, + "step": 6144 + }, + { + "epoch": 0.6748297825609488, + "grad_norm": 2.461949110031128, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7206955552101135, + "num_tokens": 157829680.0, + "step": 6145 + }, + { + "epoch": 0.6749396002635625, + "grad_norm": 2.377122640609741, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7101430892944336, + "num_tokens": 157852194.0, + "step": 6146 + }, + { + "epoch": 0.6750494179661761, + "grad_norm": 2.015062093734741, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.699893593788147, + "num_tokens": 157882951.0, + "step": 6147 + }, + { + "epoch": 0.6751592356687898, + "grad_norm": 2.211930274963379, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7075036764144897, + "num_tokens": 157906649.0, + "step": 6148 + }, + { + "epoch": 0.6752690533714034, + "grad_norm": 1.9921073913574219, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7141567468643188, + "num_tokens": 157936642.0, + "step": 6149 + }, + { + "epoch": 0.6753788710740172, + "grad_norm": 2.3637588024139404, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7292081117630005, + "num_tokens": 157958981.0, + "step": 6150 + }, + { + "epoch": 0.6754886887766308, + "grad_norm": 2.308281183242798, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6891244649887085, + "num_tokens": 157983435.0, + "step": 6151 + }, + { + "epoch": 0.6755985064792445, + "grad_norm": 1.891961932182312, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7155336141586304, + "num_tokens": 158014959.0, + "step": 6152 + }, + { + "epoch": 0.6757083241818581, + "grad_norm": 2.274498701095581, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7056654095649719, + "num_tokens": 158038481.0, + "step": 6153 + }, + { + "epoch": 0.6758181418844718, + "grad_norm": 1.9233747720718384, + "learning_rate": 1e-06, + "loss": 1.063, + "mean_token_accuracy": 0.682704508304596, + "num_tokens": 158071260.0, + "step": 6154 + }, + { + "epoch": 0.6759279595870854, + "grad_norm": 2.53359055519104, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7155864238739014, + "num_tokens": 158090566.0, + "step": 6155 + }, + { + "epoch": 0.676037777289699, + "grad_norm": 2.0838375091552734, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7057311534881592, + "num_tokens": 158117285.0, + "step": 6156 + }, + { + "epoch": 0.6761475949923128, + "grad_norm": 2.2313482761383057, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7034488320350647, + "num_tokens": 158141614.0, + "step": 6157 + }, + { + "epoch": 0.6762574126949265, + "grad_norm": 1.964897632598877, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6878546476364136, + "num_tokens": 158174388.0, + "step": 6158 + }, + { + "epoch": 0.6763672303975401, + "grad_norm": 2.420851230621338, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7192220687866211, + "num_tokens": 158194388.0, + "step": 6159 + }, + { + "epoch": 0.6764770481001537, + "grad_norm": 2.453550100326538, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.716559648513794, + "num_tokens": 158215607.0, + "step": 6160 + }, + { + "epoch": 0.6765868658027674, + "grad_norm": 2.1750996112823486, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7021385431289673, + "num_tokens": 158240472.0, + "step": 6161 + }, + { + "epoch": 0.676696683505381, + "grad_norm": 1.9603204727172852, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7112042307853699, + "num_tokens": 158270560.0, + "step": 6162 + }, + { + "epoch": 0.6768065012079947, + "grad_norm": 2.3201751708984375, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7069019079208374, + "num_tokens": 158293161.0, + "step": 6163 + }, + { + "epoch": 0.6769163189106084, + "grad_norm": 2.3597545623779297, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.717518150806427, + "num_tokens": 158316384.0, + "step": 6164 + }, + { + "epoch": 0.6770261366132221, + "grad_norm": 2.225816011428833, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6984140872955322, + "num_tokens": 158341031.0, + "step": 6165 + }, + { + "epoch": 0.6771359543158357, + "grad_norm": 2.234895706176758, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7044469714164734, + "num_tokens": 158365978.0, + "step": 6166 + }, + { + "epoch": 0.6772457720184494, + "grad_norm": 1.8866368532180786, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.717695951461792, + "num_tokens": 158398883.0, + "step": 6167 + }, + { + "epoch": 0.677355589721063, + "grad_norm": 2.0613296031951904, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.680875301361084, + "num_tokens": 158429200.0, + "step": 6168 + }, + { + "epoch": 0.6774654074236767, + "grad_norm": 1.9334471225738525, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6907797455787659, + "num_tokens": 158461558.0, + "step": 6169 + }, + { + "epoch": 0.6775752251262903, + "grad_norm": 2.2876341342926025, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7261612415313721, + "num_tokens": 158485060.0, + "step": 6170 + }, + { + "epoch": 0.6776850428289041, + "grad_norm": 2.1446874141693115, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7174164652824402, + "num_tokens": 158509643.0, + "step": 6171 + }, + { + "epoch": 0.6777948605315177, + "grad_norm": 1.8756818771362305, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7025682926177979, + "num_tokens": 158542792.0, + "step": 6172 + }, + { + "epoch": 0.6779046782341314, + "grad_norm": 2.269787073135376, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6856651306152344, + "num_tokens": 158567631.0, + "step": 6173 + }, + { + "epoch": 0.678014495936745, + "grad_norm": 2.0300538539886475, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.702013373374939, + "num_tokens": 158597522.0, + "step": 6174 + }, + { + "epoch": 0.6781243136393587, + "grad_norm": 1.994384765625, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.711164116859436, + "num_tokens": 158628619.0, + "step": 6175 + }, + { + "epoch": 0.6782341313419723, + "grad_norm": 2.2508630752563477, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7242304086685181, + "num_tokens": 158652306.0, + "step": 6176 + }, + { + "epoch": 0.678343949044586, + "grad_norm": 1.8980827331542969, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.705109715461731, + "num_tokens": 158683632.0, + "step": 6177 + }, + { + "epoch": 0.6784537667471996, + "grad_norm": 2.084073066711426, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7101059556007385, + "num_tokens": 158711002.0, + "step": 6178 + }, + { + "epoch": 0.6785635844498134, + "grad_norm": 2.2014169692993164, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6839167475700378, + "num_tokens": 158738430.0, + "step": 6179 + }, + { + "epoch": 0.678673402152427, + "grad_norm": 2.4704973697662354, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7303640842437744, + "num_tokens": 158758714.0, + "step": 6180 + }, + { + "epoch": 0.6787832198550406, + "grad_norm": 1.9136086702346802, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7115024328231812, + "num_tokens": 158789446.0, + "step": 6181 + }, + { + "epoch": 0.6788930375576543, + "grad_norm": 2.2526919841766357, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7058049440383911, + "num_tokens": 158812762.0, + "step": 6182 + }, + { + "epoch": 0.6790028552602679, + "grad_norm": 2.165893316268921, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7075259685516357, + "num_tokens": 158838585.0, + "step": 6183 + }, + { + "epoch": 0.6791126729628816, + "grad_norm": 2.356940984725952, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7272191047668457, + "num_tokens": 158860152.0, + "step": 6184 + }, + { + "epoch": 0.6792224906654952, + "grad_norm": 2.023751735687256, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6932222843170166, + "num_tokens": 158890703.0, + "step": 6185 + }, + { + "epoch": 0.679332308368109, + "grad_norm": 2.132888078689575, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6851812601089478, + "num_tokens": 158919494.0, + "step": 6186 + }, + { + "epoch": 0.6794421260707226, + "grad_norm": 2.1873714923858643, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7161789536476135, + "num_tokens": 158942953.0, + "step": 6187 + }, + { + "epoch": 0.6795519437733363, + "grad_norm": 2.2051684856414795, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.75356125831604, + "num_tokens": 158964819.0, + "step": 6188 + }, + { + "epoch": 0.6796617614759499, + "grad_norm": 2.1177618503570557, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.716825008392334, + "num_tokens": 158991571.0, + "step": 6189 + }, + { + "epoch": 0.6797715791785636, + "grad_norm": 2.3267548084259033, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7209490537643433, + "num_tokens": 159014583.0, + "step": 6190 + }, + { + "epoch": 0.6798813968811772, + "grad_norm": 2.4280731678009033, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7226024270057678, + "num_tokens": 159035965.0, + "step": 6191 + }, + { + "epoch": 0.6799912145837909, + "grad_norm": 2.2390096187591553, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7129378318786621, + "num_tokens": 159059575.0, + "step": 6192 + }, + { + "epoch": 0.6801010322864046, + "grad_norm": 2.2756481170654297, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.717242419719696, + "num_tokens": 159082702.0, + "step": 6193 + }, + { + "epoch": 0.6802108499890183, + "grad_norm": 2.2378084659576416, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.695767879486084, + "num_tokens": 159108698.0, + "step": 6194 + }, + { + "epoch": 0.6803206676916319, + "grad_norm": 2.100186824798584, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6979609131813049, + "num_tokens": 159136776.0, + "step": 6195 + }, + { + "epoch": 0.6804304853942456, + "grad_norm": 2.103116989135742, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6989322900772095, + "num_tokens": 159165841.0, + "step": 6196 + }, + { + "epoch": 0.6805403030968592, + "grad_norm": 2.149329423904419, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.695716381072998, + "num_tokens": 159190891.0, + "step": 6197 + }, + { + "epoch": 0.6806501207994728, + "grad_norm": 2.137573719024658, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7100659012794495, + "num_tokens": 159217103.0, + "step": 6198 + }, + { + "epoch": 0.6807599385020865, + "grad_norm": 2.647512435913086, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6919803619384766, + "num_tokens": 159238150.0, + "step": 6199 + }, + { + "epoch": 0.6808697562047002, + "grad_norm": 2.2558376789093018, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6969180107116699, + "num_tokens": 159262900.0, + "step": 6200 + }, + { + "epoch": 0.6809795739073139, + "grad_norm": 1.9384955167770386, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6890579462051392, + "num_tokens": 159293860.0, + "step": 6201 + }, + { + "epoch": 0.6810893916099275, + "grad_norm": 2.3460023403167725, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7293940186500549, + "num_tokens": 159315967.0, + "step": 6202 + }, + { + "epoch": 0.6811992093125412, + "grad_norm": 2.1066248416900635, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.691074013710022, + "num_tokens": 159344553.0, + "step": 6203 + }, + { + "epoch": 0.6813090270151548, + "grad_norm": 2.189656972885132, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7101185917854309, + "num_tokens": 159369028.0, + "step": 6204 + }, + { + "epoch": 0.6814188447177685, + "grad_norm": 2.1587469577789307, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7028317451477051, + "num_tokens": 159393454.0, + "step": 6205 + }, + { + "epoch": 0.6815286624203821, + "grad_norm": 1.9547348022460938, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6797451972961426, + "num_tokens": 159423334.0, + "step": 6206 + }, + { + "epoch": 0.6816384801229959, + "grad_norm": 2.1901044845581055, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6857343912124634, + "num_tokens": 159449184.0, + "step": 6207 + }, + { + "epoch": 0.6817482978256095, + "grad_norm": 2.140669345855713, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7119021415710449, + "num_tokens": 159476297.0, + "step": 6208 + }, + { + "epoch": 0.6818581155282232, + "grad_norm": 2.2168772220611572, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7371678948402405, + "num_tokens": 159500647.0, + "step": 6209 + }, + { + "epoch": 0.6819679332308368, + "grad_norm": 2.103351593017578, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6940608620643616, + "num_tokens": 159527200.0, + "step": 6210 + }, + { + "epoch": 0.6820777509334505, + "grad_norm": 2.1575286388397217, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6841055154800415, + "num_tokens": 159553476.0, + "step": 6211 + }, + { + "epoch": 0.6821875686360641, + "grad_norm": 2.278353691101074, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6894144415855408, + "num_tokens": 159579004.0, + "step": 6212 + }, + { + "epoch": 0.6822973863386778, + "grad_norm": 2.1570048332214355, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.690123975276947, + "num_tokens": 159606376.0, + "step": 6213 + }, + { + "epoch": 0.6824072040412914, + "grad_norm": 2.3380889892578125, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.70644611120224, + "num_tokens": 159629747.0, + "step": 6214 + }, + { + "epoch": 0.6825170217439052, + "grad_norm": 2.4051196575164795, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7316058874130249, + "num_tokens": 159650252.0, + "step": 6215 + }, + { + "epoch": 0.6826268394465188, + "grad_norm": 2.4576528072357178, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7271366715431213, + "num_tokens": 159670138.0, + "step": 6216 + }, + { + "epoch": 0.6827366571491325, + "grad_norm": 2.427412271499634, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7084811925888062, + "num_tokens": 159693050.0, + "step": 6217 + }, + { + "epoch": 0.6828464748517461, + "grad_norm": 2.0125350952148438, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.680761456489563, + "num_tokens": 159724388.0, + "step": 6218 + }, + { + "epoch": 0.6829562925543597, + "grad_norm": 2.2863168716430664, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7111961841583252, + "num_tokens": 159747733.0, + "step": 6219 + }, + { + "epoch": 0.6830661102569734, + "grad_norm": 2.355964422225952, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7034872770309448, + "num_tokens": 159770289.0, + "step": 6220 + }, + { + "epoch": 0.683175927959587, + "grad_norm": 1.9093979597091675, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6958822011947632, + "num_tokens": 159801828.0, + "step": 6221 + }, + { + "epoch": 0.6832857456622008, + "grad_norm": 2.1465117931365967, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7183764576911926, + "num_tokens": 159829527.0, + "step": 6222 + }, + { + "epoch": 0.6833955633648144, + "grad_norm": 2.35611891746521, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7102323770523071, + "num_tokens": 159851605.0, + "step": 6223 + }, + { + "epoch": 0.6835053810674281, + "grad_norm": 2.2784264087677, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7072893977165222, + "num_tokens": 159877367.0, + "step": 6224 + }, + { + "epoch": 0.6836151987700417, + "grad_norm": 2.3688676357269287, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7119855284690857, + "num_tokens": 159898991.0, + "step": 6225 + }, + { + "epoch": 0.6837250164726554, + "grad_norm": 2.188995838165283, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7026933431625366, + "num_tokens": 159924358.0, + "step": 6226 + }, + { + "epoch": 0.683834834175269, + "grad_norm": 2.155424118041992, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.685022234916687, + "num_tokens": 159952958.0, + "step": 6227 + }, + { + "epoch": 0.6839446518778827, + "grad_norm": 2.2735860347747803, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7199703454971313, + "num_tokens": 159976075.0, + "step": 6228 + }, + { + "epoch": 0.6840544695804964, + "grad_norm": 2.1597278118133545, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7108120918273926, + "num_tokens": 160002910.0, + "step": 6229 + }, + { + "epoch": 0.6841642872831101, + "grad_norm": 2.0096077919006348, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7191162109375, + "num_tokens": 160031411.0, + "step": 6230 + }, + { + "epoch": 0.6842741049857237, + "grad_norm": 2.178555727005005, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7211243510246277, + "num_tokens": 160058169.0, + "step": 6231 + }, + { + "epoch": 0.6843839226883374, + "grad_norm": 1.962859034538269, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6909503936767578, + "num_tokens": 160090726.0, + "step": 6232 + }, + { + "epoch": 0.684493740390951, + "grad_norm": 2.5388193130493164, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7294930815696716, + "num_tokens": 160110055.0, + "step": 6233 + }, + { + "epoch": 0.6846035580935647, + "grad_norm": 2.261550188064575, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7175226211547852, + "num_tokens": 160133311.0, + "step": 6234 + }, + { + "epoch": 0.6847133757961783, + "grad_norm": 2.4294207096099854, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7021328210830688, + "num_tokens": 160155786.0, + "step": 6235 + }, + { + "epoch": 0.6848231934987921, + "grad_norm": 2.078679084777832, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7156630754470825, + "num_tokens": 160181874.0, + "step": 6236 + }, + { + "epoch": 0.6849330112014057, + "grad_norm": 2.326719284057617, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7247296571731567, + "num_tokens": 160203918.0, + "step": 6237 + }, + { + "epoch": 0.6850428289040194, + "grad_norm": 2.204585552215576, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7264248132705688, + "num_tokens": 160228156.0, + "step": 6238 + }, + { + "epoch": 0.685152646606633, + "grad_norm": 2.442296028137207, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7091579437255859, + "num_tokens": 160250208.0, + "step": 6239 + }, + { + "epoch": 0.6852624643092466, + "grad_norm": 2.2455601692199707, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7103155851364136, + "num_tokens": 160274913.0, + "step": 6240 + }, + { + "epoch": 0.6853722820118603, + "grad_norm": 2.0573506355285645, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7125948071479797, + "num_tokens": 160302513.0, + "step": 6241 + }, + { + "epoch": 0.6854820997144739, + "grad_norm": 2.1233115196228027, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7076767683029175, + "num_tokens": 160328257.0, + "step": 6242 + }, + { + "epoch": 0.6855919174170876, + "grad_norm": 1.9440358877182007, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6942334771156311, + "num_tokens": 160359654.0, + "step": 6243 + }, + { + "epoch": 0.6857017351197013, + "grad_norm": 2.304410457611084, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7281396985054016, + "num_tokens": 160383659.0, + "step": 6244 + }, + { + "epoch": 0.685811552822315, + "grad_norm": 2.1423444747924805, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6784635782241821, + "num_tokens": 160410421.0, + "step": 6245 + }, + { + "epoch": 0.6859213705249286, + "grad_norm": 2.059776544570923, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7081403732299805, + "num_tokens": 160440460.0, + "step": 6246 + }, + { + "epoch": 0.6860311882275423, + "grad_norm": 2.3526675701141357, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7180866003036499, + "num_tokens": 160462991.0, + "step": 6247 + }, + { + "epoch": 0.6861410059301559, + "grad_norm": 2.113185167312622, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7332332134246826, + "num_tokens": 160487366.0, + "step": 6248 + }, + { + "epoch": 0.6862508236327696, + "grad_norm": 2.0251386165618896, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7042384147644043, + "num_tokens": 160514688.0, + "step": 6249 + }, + { + "epoch": 0.6863606413353832, + "grad_norm": 2.1809144020080566, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7129295468330383, + "num_tokens": 160539068.0, + "step": 6250 + }, + { + "epoch": 0.686470459037997, + "grad_norm": 1.9520859718322754, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6797064542770386, + "num_tokens": 160573143.0, + "step": 6251 + }, + { + "epoch": 0.6865802767406106, + "grad_norm": 2.044630289077759, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6963400840759277, + "num_tokens": 160602448.0, + "step": 6252 + }, + { + "epoch": 0.6866900944432243, + "grad_norm": 2.2375805377960205, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7028205394744873, + "num_tokens": 160626816.0, + "step": 6253 + }, + { + "epoch": 0.6867999121458379, + "grad_norm": 2.2076802253723145, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6793379187583923, + "num_tokens": 160653775.0, + "step": 6254 + }, + { + "epoch": 0.6869097298484516, + "grad_norm": 2.208597183227539, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6971260905265808, + "num_tokens": 160680328.0, + "step": 6255 + }, + { + "epoch": 0.6870195475510652, + "grad_norm": 2.2992818355560303, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7221260070800781, + "num_tokens": 160702786.0, + "step": 6256 + }, + { + "epoch": 0.6871293652536788, + "grad_norm": 2.0703930854797363, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6926441192626953, + "num_tokens": 160731134.0, + "step": 6257 + }, + { + "epoch": 0.6872391829562926, + "grad_norm": 1.9983536005020142, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.683056116104126, + "num_tokens": 160760917.0, + "step": 6258 + }, + { + "epoch": 0.6873490006589063, + "grad_norm": 1.9626502990722656, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6962721943855286, + "num_tokens": 160791642.0, + "step": 6259 + }, + { + "epoch": 0.6874588183615199, + "grad_norm": 2.098297595977783, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7307560443878174, + "num_tokens": 160817609.0, + "step": 6260 + }, + { + "epoch": 0.6875686360641335, + "grad_norm": 2.0833799839019775, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.692894458770752, + "num_tokens": 160846094.0, + "step": 6261 + }, + { + "epoch": 0.6876784537667472, + "grad_norm": 2.364168167114258, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7130953073501587, + "num_tokens": 160868791.0, + "step": 6262 + }, + { + "epoch": 0.6877882714693608, + "grad_norm": 2.1196069717407227, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7153174877166748, + "num_tokens": 160894917.0, + "step": 6263 + }, + { + "epoch": 0.6878980891719745, + "grad_norm": 2.8594818115234375, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7373206615447998, + "num_tokens": 160911839.0, + "step": 6264 + }, + { + "epoch": 0.6880079068745882, + "grad_norm": 2.438671112060547, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7096959352493286, + "num_tokens": 160932649.0, + "step": 6265 + }, + { + "epoch": 0.6881177245772019, + "grad_norm": 1.949427843093872, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6921982765197754, + "num_tokens": 160964425.0, + "step": 6266 + }, + { + "epoch": 0.6882275422798155, + "grad_norm": 2.166928291320801, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7246031165122986, + "num_tokens": 160989160.0, + "step": 6267 + }, + { + "epoch": 0.6883373599824292, + "grad_norm": 2.2308881282806396, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7088176608085632, + "num_tokens": 161014883.0, + "step": 6268 + }, + { + "epoch": 0.6884471776850428, + "grad_norm": 2.0835061073303223, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7055542469024658, + "num_tokens": 161042441.0, + "step": 6269 + }, + { + "epoch": 0.6885569953876565, + "grad_norm": 2.197355031967163, + "learning_rate": 1e-06, + "loss": 1.057, + "mean_token_accuracy": 0.679590106010437, + "num_tokens": 161068578.0, + "step": 6270 + }, + { + "epoch": 0.6886668130902701, + "grad_norm": 2.3481714725494385, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7172608375549316, + "num_tokens": 161090898.0, + "step": 6271 + }, + { + "epoch": 0.6887766307928838, + "grad_norm": 2.2864367961883545, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7230360507965088, + "num_tokens": 161113637.0, + "step": 6272 + }, + { + "epoch": 0.6888864484954975, + "grad_norm": 1.8507729768753052, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6961787939071655, + "num_tokens": 161146741.0, + "step": 6273 + }, + { + "epoch": 0.6889962661981112, + "grad_norm": 2.299703598022461, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7011205554008484, + "num_tokens": 161171227.0, + "step": 6274 + }, + { + "epoch": 0.6891060839007248, + "grad_norm": 2.1792287826538086, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7225943207740784, + "num_tokens": 161196677.0, + "step": 6275 + }, + { + "epoch": 0.6892159016033385, + "grad_norm": 1.9647313356399536, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7196018695831299, + "num_tokens": 161226008.0, + "step": 6276 + }, + { + "epoch": 0.6893257193059521, + "grad_norm": 2.089384078979492, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6941981911659241, + "num_tokens": 161254412.0, + "step": 6277 + }, + { + "epoch": 0.6894355370085657, + "grad_norm": 1.998559594154358, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7056195139884949, + "num_tokens": 161283762.0, + "step": 6278 + }, + { + "epoch": 0.6895453547111794, + "grad_norm": 2.033315420150757, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6997727155685425, + "num_tokens": 161314491.0, + "step": 6279 + }, + { + "epoch": 0.6896551724137931, + "grad_norm": 2.305907726287842, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7168283462524414, + "num_tokens": 161336543.0, + "step": 6280 + }, + { + "epoch": 0.6897649901164068, + "grad_norm": 2.4577958583831787, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7024213671684265, + "num_tokens": 161358760.0, + "step": 6281 + }, + { + "epoch": 0.6898748078190204, + "grad_norm": 2.1133432388305664, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7046890258789062, + "num_tokens": 161383933.0, + "step": 6282 + }, + { + "epoch": 0.6899846255216341, + "grad_norm": 1.8038203716278076, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6830081343650818, + "num_tokens": 161423173.0, + "step": 6283 + }, + { + "epoch": 0.6900944432242477, + "grad_norm": 2.4261343479156494, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7303029298782349, + "num_tokens": 161444118.0, + "step": 6284 + }, + { + "epoch": 0.6902042609268614, + "grad_norm": 2.475351095199585, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7256661057472229, + "num_tokens": 161464406.0, + "step": 6285 + }, + { + "epoch": 0.690314078629475, + "grad_norm": 2.178187131881714, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.70789635181427, + "num_tokens": 161488189.0, + "step": 6286 + }, + { + "epoch": 0.6904238963320888, + "grad_norm": 2.0794620513916016, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7243047952651978, + "num_tokens": 161515872.0, + "step": 6287 + }, + { + "epoch": 0.6905337140347024, + "grad_norm": 2.257023572921753, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7099905014038086, + "num_tokens": 161537748.0, + "step": 6288 + }, + { + "epoch": 0.6906435317373161, + "grad_norm": 2.3063948154449463, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7131111025810242, + "num_tokens": 161562348.0, + "step": 6289 + }, + { + "epoch": 0.6907533494399297, + "grad_norm": 1.8372701406478882, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.713857889175415, + "num_tokens": 161594674.0, + "step": 6290 + }, + { + "epoch": 0.6908631671425434, + "grad_norm": 2.1094117164611816, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7055776119232178, + "num_tokens": 161623275.0, + "step": 6291 + }, + { + "epoch": 0.690972984845157, + "grad_norm": 1.9920858144760132, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6863762140274048, + "num_tokens": 161654767.0, + "step": 6292 + }, + { + "epoch": 0.6910828025477707, + "grad_norm": 2.2579996585845947, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7292671203613281, + "num_tokens": 161678037.0, + "step": 6293 + }, + { + "epoch": 0.6911926202503844, + "grad_norm": 2.200134038925171, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7087337970733643, + "num_tokens": 161703946.0, + "step": 6294 + }, + { + "epoch": 0.6913024379529981, + "grad_norm": 2.3766634464263916, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7192480564117432, + "num_tokens": 161724845.0, + "step": 6295 + }, + { + "epoch": 0.6914122556556117, + "grad_norm": 2.2419090270996094, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7102140784263611, + "num_tokens": 161750700.0, + "step": 6296 + }, + { + "epoch": 0.6915220733582254, + "grad_norm": 2.2374770641326904, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6996399164199829, + "num_tokens": 161775442.0, + "step": 6297 + }, + { + "epoch": 0.691631891060839, + "grad_norm": 2.235291004180908, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7076172828674316, + "num_tokens": 161799937.0, + "step": 6298 + }, + { + "epoch": 0.6917417087634526, + "grad_norm": 2.016368865966797, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7359957098960876, + "num_tokens": 161827897.0, + "step": 6299 + }, + { + "epoch": 0.6918515264660663, + "grad_norm": 2.2153193950653076, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.698462963104248, + "num_tokens": 161853727.0, + "step": 6300 + }, + { + "epoch": 0.6919613441686799, + "grad_norm": 1.7256455421447754, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7147272825241089, + "num_tokens": 161892503.0, + "step": 6301 + }, + { + "epoch": 0.6920711618712937, + "grad_norm": 2.5610463619232178, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7315781116485596, + "num_tokens": 161912713.0, + "step": 6302 + }, + { + "epoch": 0.6921809795739073, + "grad_norm": 2.2139132022857666, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7181061506271362, + "num_tokens": 161937324.0, + "step": 6303 + }, + { + "epoch": 0.692290797276521, + "grad_norm": 1.8146644830703735, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.692579984664917, + "num_tokens": 161971369.0, + "step": 6304 + }, + { + "epoch": 0.6924006149791346, + "grad_norm": 1.9730411767959595, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7224975824356079, + "num_tokens": 161999857.0, + "step": 6305 + }, + { + "epoch": 0.6925104326817483, + "grad_norm": 2.1832616329193115, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7217661142349243, + "num_tokens": 162025643.0, + "step": 6306 + }, + { + "epoch": 0.6926202503843619, + "grad_norm": 2.1251626014709473, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7135993242263794, + "num_tokens": 162055058.0, + "step": 6307 + }, + { + "epoch": 0.6927300680869756, + "grad_norm": 2.240518808364868, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.699372410774231, + "num_tokens": 162080153.0, + "step": 6308 + }, + { + "epoch": 0.6928398857895893, + "grad_norm": 2.272423267364502, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7012399435043335, + "num_tokens": 162108669.0, + "step": 6309 + }, + { + "epoch": 0.692949703492203, + "grad_norm": 2.453573703765869, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6836119294166565, + "num_tokens": 162132538.0, + "step": 6310 + }, + { + "epoch": 0.6930595211948166, + "grad_norm": 2.542332172393799, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7244076132774353, + "num_tokens": 162151396.0, + "step": 6311 + }, + { + "epoch": 0.6931693388974303, + "grad_norm": 2.193462371826172, + "learning_rate": 1e-06, + "loss": 1.1016, + "mean_token_accuracy": 0.6801773309707642, + "num_tokens": 162178786.0, + "step": 6312 + }, + { + "epoch": 0.6932791566000439, + "grad_norm": 2.2720296382904053, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.706507682800293, + "num_tokens": 162202496.0, + "step": 6313 + }, + { + "epoch": 0.6933889743026576, + "grad_norm": 2.24543833732605, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7050827741622925, + "num_tokens": 162227476.0, + "step": 6314 + }, + { + "epoch": 0.6934987920052712, + "grad_norm": 2.1813578605651855, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6974430084228516, + "num_tokens": 162252571.0, + "step": 6315 + }, + { + "epoch": 0.693608609707885, + "grad_norm": 2.1931071281433105, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.699530303478241, + "num_tokens": 162278924.0, + "step": 6316 + }, + { + "epoch": 0.6937184274104986, + "grad_norm": 2.2226059436798096, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6877589821815491, + "num_tokens": 162303228.0, + "step": 6317 + }, + { + "epoch": 0.6938282451131123, + "grad_norm": 2.1711063385009766, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7001473307609558, + "num_tokens": 162329627.0, + "step": 6318 + }, + { + "epoch": 0.6939380628157259, + "grad_norm": 2.1039958000183105, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.689703643321991, + "num_tokens": 162360411.0, + "step": 6319 + }, + { + "epoch": 0.6940478805183395, + "grad_norm": 2.3531692028045654, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6983745098114014, + "num_tokens": 162383532.0, + "step": 6320 + }, + { + "epoch": 0.6941576982209532, + "grad_norm": 2.1120173931121826, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7177578210830688, + "num_tokens": 162411159.0, + "step": 6321 + }, + { + "epoch": 0.6942675159235668, + "grad_norm": 2.211751937866211, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7066845297813416, + "num_tokens": 162436756.0, + "step": 6322 + }, + { + "epoch": 0.6943773336261806, + "grad_norm": 2.1491949558258057, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7323977947235107, + "num_tokens": 162459946.0, + "step": 6323 + }, + { + "epoch": 0.6944871513287942, + "grad_norm": 2.435253858566284, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7146438360214233, + "num_tokens": 162482123.0, + "step": 6324 + }, + { + "epoch": 0.6945969690314079, + "grad_norm": 2.100665807723999, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7418639659881592, + "num_tokens": 162508317.0, + "step": 6325 + }, + { + "epoch": 0.6947067867340215, + "grad_norm": 2.341299533843994, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7107622027397156, + "num_tokens": 162532121.0, + "step": 6326 + }, + { + "epoch": 0.6948166044366352, + "grad_norm": 2.209333658218384, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6981008052825928, + "num_tokens": 162557621.0, + "step": 6327 + }, + { + "epoch": 0.6949264221392488, + "grad_norm": 2.282405138015747, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7117613554000854, + "num_tokens": 162581933.0, + "step": 6328 + }, + { + "epoch": 0.6950362398418625, + "grad_norm": 2.1138312816619873, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6960517764091492, + "num_tokens": 162608956.0, + "step": 6329 + }, + { + "epoch": 0.6951460575444761, + "grad_norm": 2.0680649280548096, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6958585977554321, + "num_tokens": 162637804.0, + "step": 6330 + }, + { + "epoch": 0.6952558752470899, + "grad_norm": 2.059638738632202, + "learning_rate": 1e-06, + "loss": 1.1314, + "mean_token_accuracy": 0.6653361320495605, + "num_tokens": 162669174.0, + "step": 6331 + }, + { + "epoch": 0.6953656929497035, + "grad_norm": 2.2153139114379883, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7058703899383545, + "num_tokens": 162694414.0, + "step": 6332 + }, + { + "epoch": 0.6954755106523172, + "grad_norm": 2.0797219276428223, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.68781578540802, + "num_tokens": 162723249.0, + "step": 6333 + }, + { + "epoch": 0.6955853283549308, + "grad_norm": 2.141791820526123, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.713497519493103, + "num_tokens": 162750874.0, + "step": 6334 + }, + { + "epoch": 0.6956951460575445, + "grad_norm": 2.2173898220062256, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6906980276107788, + "num_tokens": 162777820.0, + "step": 6335 + }, + { + "epoch": 0.6958049637601581, + "grad_norm": 2.1661133766174316, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7171028852462769, + "num_tokens": 162801722.0, + "step": 6336 + }, + { + "epoch": 0.6959147814627717, + "grad_norm": 2.1485977172851562, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6843178272247314, + "num_tokens": 162829830.0, + "step": 6337 + }, + { + "epoch": 0.6960245991653855, + "grad_norm": 2.3620119094848633, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7238448262214661, + "num_tokens": 162851724.0, + "step": 6338 + }, + { + "epoch": 0.6961344168679992, + "grad_norm": 2.256479263305664, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7297225594520569, + "num_tokens": 162872971.0, + "step": 6339 + }, + { + "epoch": 0.6962442345706128, + "grad_norm": 2.117495059967041, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6764901876449585, + "num_tokens": 162902155.0, + "step": 6340 + }, + { + "epoch": 0.6963540522732264, + "grad_norm": 1.9800169467926025, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7195419073104858, + "num_tokens": 162931450.0, + "step": 6341 + }, + { + "epoch": 0.6964638699758401, + "grad_norm": 2.1550121307373047, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6911083459854126, + "num_tokens": 162956893.0, + "step": 6342 + }, + { + "epoch": 0.6965736876784537, + "grad_norm": 2.1628408432006836, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7082393169403076, + "num_tokens": 162982639.0, + "step": 6343 + }, + { + "epoch": 0.6966835053810674, + "grad_norm": 2.505242347717285, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7149136066436768, + "num_tokens": 163002636.0, + "step": 6344 + }, + { + "epoch": 0.6967933230836811, + "grad_norm": 2.1264407634735107, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.697166383266449, + "num_tokens": 163029359.0, + "step": 6345 + }, + { + "epoch": 0.6969031407862948, + "grad_norm": 2.531677484512329, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7120102047920227, + "num_tokens": 163049735.0, + "step": 6346 + }, + { + "epoch": 0.6970129584889084, + "grad_norm": 2.234144926071167, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6870243549346924, + "num_tokens": 163075443.0, + "step": 6347 + }, + { + "epoch": 0.6971227761915221, + "grad_norm": 2.246431827545166, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7060233354568481, + "num_tokens": 163101118.0, + "step": 6348 + }, + { + "epoch": 0.6972325938941357, + "grad_norm": 2.19643235206604, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7279677391052246, + "num_tokens": 163126408.0, + "step": 6349 + }, + { + "epoch": 0.6973424115967494, + "grad_norm": 2.4101128578186035, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.6978634595870972, + "num_tokens": 163148250.0, + "step": 6350 + }, + { + "epoch": 0.697452229299363, + "grad_norm": 2.2287518978118896, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7018163204193115, + "num_tokens": 163172558.0, + "step": 6351 + }, + { + "epoch": 0.6975620470019768, + "grad_norm": 2.113162040710449, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6921806335449219, + "num_tokens": 163200813.0, + "step": 6352 + }, + { + "epoch": 0.6976718647045904, + "grad_norm": 2.1610183715820312, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7293033003807068, + "num_tokens": 163226235.0, + "step": 6353 + }, + { + "epoch": 0.6977816824072041, + "grad_norm": 2.4706413745880127, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7081714868545532, + "num_tokens": 163246122.0, + "step": 6354 + }, + { + "epoch": 0.6978915001098177, + "grad_norm": 2.23057222366333, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7016571760177612, + "num_tokens": 163273200.0, + "step": 6355 + }, + { + "epoch": 0.6980013178124314, + "grad_norm": 2.3070952892303467, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7186819314956665, + "num_tokens": 163295975.0, + "step": 6356 + }, + { + "epoch": 0.698111135515045, + "grad_norm": 2.2260000705718994, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.718234658241272, + "num_tokens": 163320925.0, + "step": 6357 + }, + { + "epoch": 0.6982209532176586, + "grad_norm": 2.154175281524658, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7029296159744263, + "num_tokens": 163348926.0, + "step": 6358 + }, + { + "epoch": 0.6983307709202724, + "grad_norm": 2.256391763687134, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6967685222625732, + "num_tokens": 163374676.0, + "step": 6359 + }, + { + "epoch": 0.698440588622886, + "grad_norm": 2.2379066944122314, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7039893865585327, + "num_tokens": 163399567.0, + "step": 6360 + }, + { + "epoch": 0.6985504063254997, + "grad_norm": 2.4005284309387207, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7163660526275635, + "num_tokens": 163422376.0, + "step": 6361 + }, + { + "epoch": 0.6986602240281133, + "grad_norm": 2.103837490081787, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7080591320991516, + "num_tokens": 163451525.0, + "step": 6362 + }, + { + "epoch": 0.698770041730727, + "grad_norm": 2.2236404418945312, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7194196581840515, + "num_tokens": 163476697.0, + "step": 6363 + }, + { + "epoch": 0.6988798594333406, + "grad_norm": 2.133342742919922, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6947474479675293, + "num_tokens": 163504088.0, + "step": 6364 + }, + { + "epoch": 0.6989896771359543, + "grad_norm": 2.1393611431121826, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.70912104845047, + "num_tokens": 163528576.0, + "step": 6365 + }, + { + "epoch": 0.6990994948385679, + "grad_norm": 2.4756863117218018, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7173798680305481, + "num_tokens": 163549775.0, + "step": 6366 + }, + { + "epoch": 0.6992093125411817, + "grad_norm": 2.3453214168548584, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.697268009185791, + "num_tokens": 163573241.0, + "step": 6367 + }, + { + "epoch": 0.6993191302437953, + "grad_norm": 2.093531370162964, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7163058519363403, + "num_tokens": 163599622.0, + "step": 6368 + }, + { + "epoch": 0.699428947946409, + "grad_norm": 2.229965925216675, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7146995663642883, + "num_tokens": 163626724.0, + "step": 6369 + }, + { + "epoch": 0.6995387656490226, + "grad_norm": 2.139260768890381, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7104266881942749, + "num_tokens": 163654695.0, + "step": 6370 + }, + { + "epoch": 0.6996485833516363, + "grad_norm": 1.9708476066589355, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6812073588371277, + "num_tokens": 163686712.0, + "step": 6371 + }, + { + "epoch": 0.6997584010542499, + "grad_norm": 2.2650434970855713, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.721254825592041, + "num_tokens": 163710550.0, + "step": 6372 + }, + { + "epoch": 0.6998682187568636, + "grad_norm": 2.669764280319214, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.725717306137085, + "num_tokens": 163728710.0, + "step": 6373 + }, + { + "epoch": 0.6999780364594773, + "grad_norm": 2.160552501678467, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.712597131729126, + "num_tokens": 163753398.0, + "step": 6374 + }, + { + "epoch": 0.700087854162091, + "grad_norm": 2.051814556121826, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.700393795967102, + "num_tokens": 163780122.0, + "step": 6375 + }, + { + "epoch": 0.7001976718647046, + "grad_norm": 2.34826397895813, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7068220376968384, + "num_tokens": 163801745.0, + "step": 6376 + }, + { + "epoch": 0.7003074895673183, + "grad_norm": 2.720414400100708, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7114408016204834, + "num_tokens": 163819745.0, + "step": 6377 + }, + { + "epoch": 0.7004173072699319, + "grad_norm": 2.301861047744751, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7133829593658447, + "num_tokens": 163843180.0, + "step": 6378 + }, + { + "epoch": 0.7005271249725455, + "grad_norm": 1.9936676025390625, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.706960916519165, + "num_tokens": 163876143.0, + "step": 6379 + }, + { + "epoch": 0.7006369426751592, + "grad_norm": 2.174208879470825, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6894959211349487, + "num_tokens": 163903100.0, + "step": 6380 + }, + { + "epoch": 0.700746760377773, + "grad_norm": 1.9983069896697998, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7401920557022095, + "num_tokens": 163934110.0, + "step": 6381 + }, + { + "epoch": 0.7008565780803866, + "grad_norm": 2.2304883003234863, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.712577223777771, + "num_tokens": 163959280.0, + "step": 6382 + }, + { + "epoch": 0.7009663957830002, + "grad_norm": 2.0076494216918945, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.693120002746582, + "num_tokens": 163988855.0, + "step": 6383 + }, + { + "epoch": 0.7010762134856139, + "grad_norm": 2.1182432174682617, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.7007978558540344, + "num_tokens": 164016674.0, + "step": 6384 + }, + { + "epoch": 0.7011860311882275, + "grad_norm": 2.249622106552124, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6929528117179871, + "num_tokens": 164044576.0, + "step": 6385 + }, + { + "epoch": 0.7012958488908412, + "grad_norm": 2.0064210891723633, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7007923722267151, + "num_tokens": 164073095.0, + "step": 6386 + }, + { + "epoch": 0.7014056665934548, + "grad_norm": 2.11930513381958, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7119659185409546, + "num_tokens": 164098359.0, + "step": 6387 + }, + { + "epoch": 0.7015154842960686, + "grad_norm": 2.127349615097046, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7044897675514221, + "num_tokens": 164123921.0, + "step": 6388 + }, + { + "epoch": 0.7016253019986822, + "grad_norm": 2.443722724914551, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7163705825805664, + "num_tokens": 164145528.0, + "step": 6389 + }, + { + "epoch": 0.7017351197012959, + "grad_norm": 2.333327054977417, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7379189729690552, + "num_tokens": 164168183.0, + "step": 6390 + }, + { + "epoch": 0.7018449374039095, + "grad_norm": 1.9997069835662842, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7240992784500122, + "num_tokens": 164197129.0, + "step": 6391 + }, + { + "epoch": 0.7019547551065232, + "grad_norm": 1.9914640188217163, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6946961879730225, + "num_tokens": 164228060.0, + "step": 6392 + }, + { + "epoch": 0.7020645728091368, + "grad_norm": 2.1536219120025635, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7093769907951355, + "num_tokens": 164254390.0, + "step": 6393 + }, + { + "epoch": 0.7021743905117505, + "grad_norm": 2.1241674423217773, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6986984610557556, + "num_tokens": 164282356.0, + "step": 6394 + }, + { + "epoch": 0.7022842082143641, + "grad_norm": 2.0634231567382812, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7027988433837891, + "num_tokens": 164309642.0, + "step": 6395 + }, + { + "epoch": 0.7023940259169779, + "grad_norm": 2.164276361465454, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.693736732006073, + "num_tokens": 164336540.0, + "step": 6396 + }, + { + "epoch": 0.7025038436195915, + "grad_norm": 2.071712017059326, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7068469524383545, + "num_tokens": 164363113.0, + "step": 6397 + }, + { + "epoch": 0.7026136613222052, + "grad_norm": 1.9066681861877441, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.7002902030944824, + "num_tokens": 164397129.0, + "step": 6398 + }, + { + "epoch": 0.7027234790248188, + "grad_norm": 2.2908504009246826, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7078073024749756, + "num_tokens": 164420658.0, + "step": 6399 + }, + { + "epoch": 0.7028332967274324, + "grad_norm": 1.985180139541626, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7078270316123962, + "num_tokens": 164448980.0, + "step": 6400 + }, + { + "epoch": 0.7029431144300461, + "grad_norm": 2.350740432739258, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6901669502258301, + "num_tokens": 164472052.0, + "step": 6401 + }, + { + "epoch": 0.7030529321326597, + "grad_norm": 2.0961759090423584, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7081127166748047, + "num_tokens": 164499552.0, + "step": 6402 + }, + { + "epoch": 0.7031627498352735, + "grad_norm": 2.7028958797454834, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7254544496536255, + "num_tokens": 164517097.0, + "step": 6403 + }, + { + "epoch": 0.7032725675378871, + "grad_norm": 2.2649457454681396, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.696099042892456, + "num_tokens": 164542814.0, + "step": 6404 + }, + { + "epoch": 0.7033823852405008, + "grad_norm": 2.157257556915283, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7150473594665527, + "num_tokens": 164568628.0, + "step": 6405 + }, + { + "epoch": 0.7034922029431144, + "grad_norm": 2.5080251693725586, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7087422609329224, + "num_tokens": 164587422.0, + "step": 6406 + }, + { + "epoch": 0.7036020206457281, + "grad_norm": 2.2400705814361572, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7050641775131226, + "num_tokens": 164611234.0, + "step": 6407 + }, + { + "epoch": 0.7037118383483417, + "grad_norm": 2.35854434967041, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7108844518661499, + "num_tokens": 164632645.0, + "step": 6408 + }, + { + "epoch": 0.7038216560509554, + "grad_norm": 2.159846782684326, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6908475756645203, + "num_tokens": 164659491.0, + "step": 6409 + }, + { + "epoch": 0.7039314737535691, + "grad_norm": 2.12506103515625, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7044491767883301, + "num_tokens": 164685803.0, + "step": 6410 + }, + { + "epoch": 0.7040412914561828, + "grad_norm": 2.391716718673706, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7121464014053345, + "num_tokens": 164708498.0, + "step": 6411 + }, + { + "epoch": 0.7041511091587964, + "grad_norm": 2.1085894107818604, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.721908688545227, + "num_tokens": 164735533.0, + "step": 6412 + }, + { + "epoch": 0.7042609268614101, + "grad_norm": 2.2629592418670654, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.736571192741394, + "num_tokens": 164757995.0, + "step": 6413 + }, + { + "epoch": 0.7043707445640237, + "grad_norm": 2.119086742401123, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7183876037597656, + "num_tokens": 164783224.0, + "step": 6414 + }, + { + "epoch": 0.7044805622666374, + "grad_norm": 2.115809202194214, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7202500104904175, + "num_tokens": 164809598.0, + "step": 6415 + }, + { + "epoch": 0.704590379969251, + "grad_norm": 2.161301374435425, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6972047090530396, + "num_tokens": 164835910.0, + "step": 6416 + }, + { + "epoch": 0.7047001976718648, + "grad_norm": 1.8117971420288086, + "learning_rate": 1e-06, + "loss": 1.1122, + "mean_token_accuracy": 0.6719090938568115, + "num_tokens": 164873491.0, + "step": 6417 + }, + { + "epoch": 0.7048100153744784, + "grad_norm": 2.2941465377807617, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6996648907661438, + "num_tokens": 164896540.0, + "step": 6418 + }, + { + "epoch": 0.704919833077092, + "grad_norm": 2.3375864028930664, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7153331637382507, + "num_tokens": 164917528.0, + "step": 6419 + }, + { + "epoch": 0.7050296507797057, + "grad_norm": 2.440985918045044, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6868098378181458, + "num_tokens": 164940819.0, + "step": 6420 + }, + { + "epoch": 0.7051394684823193, + "grad_norm": 2.243835687637329, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7003135681152344, + "num_tokens": 164966214.0, + "step": 6421 + }, + { + "epoch": 0.705249286184933, + "grad_norm": 2.0059945583343506, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.718697190284729, + "num_tokens": 164995949.0, + "step": 6422 + }, + { + "epoch": 0.7053591038875466, + "grad_norm": 2.1743016242980957, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7133765816688538, + "num_tokens": 165022996.0, + "step": 6423 + }, + { + "epoch": 0.7054689215901603, + "grad_norm": 2.4492101669311523, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7108986377716064, + "num_tokens": 165044411.0, + "step": 6424 + }, + { + "epoch": 0.705578739292774, + "grad_norm": 2.477168321609497, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6970330476760864, + "num_tokens": 165065152.0, + "step": 6425 + }, + { + "epoch": 0.7056885569953877, + "grad_norm": 2.344438076019287, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7210425138473511, + "num_tokens": 165086321.0, + "step": 6426 + }, + { + "epoch": 0.7057983746980013, + "grad_norm": 2.0668697357177734, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6803666353225708, + "num_tokens": 165116354.0, + "step": 6427 + }, + { + "epoch": 0.705908192400615, + "grad_norm": 2.0695974826812744, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.7048291563987732, + "num_tokens": 165145690.0, + "step": 6428 + }, + { + "epoch": 0.7060180101032286, + "grad_norm": 2.204285144805908, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7036266326904297, + "num_tokens": 165169346.0, + "step": 6429 + }, + { + "epoch": 0.7061278278058423, + "grad_norm": 2.125480890274048, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7328020334243774, + "num_tokens": 165195027.0, + "step": 6430 + }, + { + "epoch": 0.7062376455084559, + "grad_norm": 2.016826868057251, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7269752025604248, + "num_tokens": 165221731.0, + "step": 6431 + }, + { + "epoch": 0.7063474632110697, + "grad_norm": 2.121516704559326, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7065003514289856, + "num_tokens": 165250295.0, + "step": 6432 + }, + { + "epoch": 0.7064572809136833, + "grad_norm": 2.216769218444824, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6989525556564331, + "num_tokens": 165274817.0, + "step": 6433 + }, + { + "epoch": 0.706567098616297, + "grad_norm": 2.06278920173645, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.7121293544769287, + "num_tokens": 165302575.0, + "step": 6434 + }, + { + "epoch": 0.7066769163189106, + "grad_norm": 2.2846832275390625, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7043828964233398, + "num_tokens": 165325152.0, + "step": 6435 + }, + { + "epoch": 0.7067867340215243, + "grad_norm": 2.0124237537384033, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6851256489753723, + "num_tokens": 165357868.0, + "step": 6436 + }, + { + "epoch": 0.7068965517241379, + "grad_norm": 2.044027805328369, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.697272002696991, + "num_tokens": 165388384.0, + "step": 6437 + }, + { + "epoch": 0.7070063694267515, + "grad_norm": 2.377720832824707, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6799320578575134, + "num_tokens": 165413458.0, + "step": 6438 + }, + { + "epoch": 0.7071161871293653, + "grad_norm": 2.2161030769348145, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6904100179672241, + "num_tokens": 165439512.0, + "step": 6439 + }, + { + "epoch": 0.707226004831979, + "grad_norm": 1.9281857013702393, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7073133587837219, + "num_tokens": 165470221.0, + "step": 6440 + }, + { + "epoch": 0.7073358225345926, + "grad_norm": 2.211491823196411, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7358235716819763, + "num_tokens": 165491855.0, + "step": 6441 + }, + { + "epoch": 0.7074456402372062, + "grad_norm": 2.088550567626953, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7123574018478394, + "num_tokens": 165517862.0, + "step": 6442 + }, + { + "epoch": 0.7075554579398199, + "grad_norm": 2.1893293857574463, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.7024092674255371, + "num_tokens": 165544295.0, + "step": 6443 + }, + { + "epoch": 0.7076652756424335, + "grad_norm": 2.1390304565429688, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6969496011734009, + "num_tokens": 165570722.0, + "step": 6444 + }, + { + "epoch": 0.7077750933450472, + "grad_norm": 2.373141050338745, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6902856826782227, + "num_tokens": 165592773.0, + "step": 6445 + }, + { + "epoch": 0.7078849110476609, + "grad_norm": 2.255479574203491, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6979330778121948, + "num_tokens": 165616848.0, + "step": 6446 + }, + { + "epoch": 0.7079947287502746, + "grad_norm": 1.9308295249938965, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.6817434430122375, + "num_tokens": 165650686.0, + "step": 6447 + }, + { + "epoch": 0.7081045464528882, + "grad_norm": 1.9666125774383545, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6841530203819275, + "num_tokens": 165683790.0, + "step": 6448 + }, + { + "epoch": 0.7082143641555019, + "grad_norm": 2.322719097137451, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7375005483627319, + "num_tokens": 165706025.0, + "step": 6449 + }, + { + "epoch": 0.7083241818581155, + "grad_norm": 2.1331326961517334, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7074906826019287, + "num_tokens": 165730347.0, + "step": 6450 + }, + { + "epoch": 0.7084339995607292, + "grad_norm": 2.4576573371887207, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7157502174377441, + "num_tokens": 165750810.0, + "step": 6451 + }, + { + "epoch": 0.7085438172633428, + "grad_norm": 2.021777391433716, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6845494508743286, + "num_tokens": 165782569.0, + "step": 6452 + }, + { + "epoch": 0.7086536349659565, + "grad_norm": 2.3607680797576904, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7042368650436401, + "num_tokens": 165806797.0, + "step": 6453 + }, + { + "epoch": 0.7087634526685702, + "grad_norm": 2.6252663135528564, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7085463404655457, + "num_tokens": 165827043.0, + "step": 6454 + }, + { + "epoch": 0.7088732703711839, + "grad_norm": 2.6697330474853516, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7231154441833496, + "num_tokens": 165844967.0, + "step": 6455 + }, + { + "epoch": 0.7089830880737975, + "grad_norm": 2.167120933532715, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.701231837272644, + "num_tokens": 165870089.0, + "step": 6456 + }, + { + "epoch": 0.7090929057764112, + "grad_norm": 2.1669933795928955, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7238174676895142, + "num_tokens": 165895069.0, + "step": 6457 + }, + { + "epoch": 0.7092027234790248, + "grad_norm": 2.1091535091400146, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7129480242729187, + "num_tokens": 165922926.0, + "step": 6458 + }, + { + "epoch": 0.7093125411816384, + "grad_norm": 2.1400365829467773, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6991580724716187, + "num_tokens": 165949661.0, + "step": 6459 + }, + { + "epoch": 0.7094223588842521, + "grad_norm": 2.1098906993865967, + "learning_rate": 1e-06, + "loss": 1.0875, + "mean_token_accuracy": 0.6682402491569519, + "num_tokens": 165978025.0, + "step": 6460 + }, + { + "epoch": 0.7095321765868658, + "grad_norm": 2.140219211578369, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.708199679851532, + "num_tokens": 166002289.0, + "step": 6461 + }, + { + "epoch": 0.7096419942894795, + "grad_norm": 2.029337167739868, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6947171688079834, + "num_tokens": 166029664.0, + "step": 6462 + }, + { + "epoch": 0.7097518119920931, + "grad_norm": 2.4337453842163086, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7011495232582092, + "num_tokens": 166052165.0, + "step": 6463 + }, + { + "epoch": 0.7098616296947068, + "grad_norm": 2.03292179107666, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6954559087753296, + "num_tokens": 166082136.0, + "step": 6464 + }, + { + "epoch": 0.7099714473973204, + "grad_norm": 2.1468122005462646, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6684055328369141, + "num_tokens": 166107360.0, + "step": 6465 + }, + { + "epoch": 0.7100812650999341, + "grad_norm": 2.075777292251587, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6900792121887207, + "num_tokens": 166135575.0, + "step": 6466 + }, + { + "epoch": 0.7101910828025477, + "grad_norm": 2.169351577758789, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6949017643928528, + "num_tokens": 166162555.0, + "step": 6467 + }, + { + "epoch": 0.7103009005051615, + "grad_norm": 2.0577738285064697, + "learning_rate": 1e-06, + "loss": 1.089, + "mean_token_accuracy": 0.6788307428359985, + "num_tokens": 166193397.0, + "step": 6468 + }, + { + "epoch": 0.7104107182077751, + "grad_norm": 2.040750503540039, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6884754300117493, + "num_tokens": 166225214.0, + "step": 6469 + }, + { + "epoch": 0.7105205359103888, + "grad_norm": 1.888776421546936, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7017906308174133, + "num_tokens": 166258953.0, + "step": 6470 + }, + { + "epoch": 0.7106303536130024, + "grad_norm": 2.203648805618286, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7032526731491089, + "num_tokens": 166284209.0, + "step": 6471 + }, + { + "epoch": 0.7107401713156161, + "grad_norm": 2.403019666671753, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7052016258239746, + "num_tokens": 166305918.0, + "step": 6472 + }, + { + "epoch": 0.7108499890182297, + "grad_norm": 1.9928334951400757, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.721472442150116, + "num_tokens": 166332933.0, + "step": 6473 + }, + { + "epoch": 0.7109598067208434, + "grad_norm": 2.0604164600372314, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6871574521064758, + "num_tokens": 166363934.0, + "step": 6474 + }, + { + "epoch": 0.7110696244234571, + "grad_norm": 2.2307095527648926, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7089745998382568, + "num_tokens": 166387169.0, + "step": 6475 + }, + { + "epoch": 0.7111794421260708, + "grad_norm": 2.022862434387207, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7129876017570496, + "num_tokens": 166414524.0, + "step": 6476 + }, + { + "epoch": 0.7112892598286844, + "grad_norm": 2.0395193099975586, + "learning_rate": 1e-06, + "loss": 1.0869, + "mean_token_accuracy": 0.6819856762886047, + "num_tokens": 166444702.0, + "step": 6477 + }, + { + "epoch": 0.711399077531298, + "grad_norm": 2.3312621116638184, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7008259296417236, + "num_tokens": 166469669.0, + "step": 6478 + }, + { + "epoch": 0.7115088952339117, + "grad_norm": 1.8635375499725342, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.68864506483078, + "num_tokens": 166503514.0, + "step": 6479 + }, + { + "epoch": 0.7116187129365253, + "grad_norm": 2.233098268508911, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7114719748497009, + "num_tokens": 166529786.0, + "step": 6480 + }, + { + "epoch": 0.711728530639139, + "grad_norm": 1.9508116245269775, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.696021318435669, + "num_tokens": 166560572.0, + "step": 6481 + }, + { + "epoch": 0.7118383483417526, + "grad_norm": 2.0923306941986084, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.699819803237915, + "num_tokens": 166587901.0, + "step": 6482 + }, + { + "epoch": 0.7119481660443664, + "grad_norm": 2.2651236057281494, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7040475606918335, + "num_tokens": 166611361.0, + "step": 6483 + }, + { + "epoch": 0.71205798374698, + "grad_norm": 2.2634801864624023, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6960439085960388, + "num_tokens": 166633950.0, + "step": 6484 + }, + { + "epoch": 0.7121678014495937, + "grad_norm": 2.119333505630493, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6992515325546265, + "num_tokens": 166662458.0, + "step": 6485 + }, + { + "epoch": 0.7122776191522073, + "grad_norm": 2.069005012512207, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7014533877372742, + "num_tokens": 166688401.0, + "step": 6486 + }, + { + "epoch": 0.712387436854821, + "grad_norm": 2.320204019546509, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7149983644485474, + "num_tokens": 166711798.0, + "step": 6487 + }, + { + "epoch": 0.7124972545574346, + "grad_norm": 2.3772201538085938, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.716363787651062, + "num_tokens": 166734126.0, + "step": 6488 + }, + { + "epoch": 0.7126070722600483, + "grad_norm": 2.584627389907837, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7267165184020996, + "num_tokens": 166753434.0, + "step": 6489 + }, + { + "epoch": 0.712716889962662, + "grad_norm": 2.210542678833008, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7079654335975647, + "num_tokens": 166777260.0, + "step": 6490 + }, + { + "epoch": 0.7128267076652757, + "grad_norm": 2.1920218467712402, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7185642123222351, + "num_tokens": 166804143.0, + "step": 6491 + }, + { + "epoch": 0.7129365253678893, + "grad_norm": 2.449198007583618, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7196977734565735, + "num_tokens": 166823846.0, + "step": 6492 + }, + { + "epoch": 0.713046343070503, + "grad_norm": 2.0397748947143555, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.7036637663841248, + "num_tokens": 166855750.0, + "step": 6493 + }, + { + "epoch": 0.7131561607731166, + "grad_norm": 2.160858631134033, + "learning_rate": 1e-06, + "loss": 1.0891, + "mean_token_accuracy": 0.6747516393661499, + "num_tokens": 166883025.0, + "step": 6494 + }, + { + "epoch": 0.7132659784757303, + "grad_norm": 2.3444108963012695, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6856775879859924, + "num_tokens": 166911916.0, + "step": 6495 + }, + { + "epoch": 0.7133757961783439, + "grad_norm": 2.032111883163452, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6917669177055359, + "num_tokens": 166942262.0, + "step": 6496 + }, + { + "epoch": 0.7134856138809577, + "grad_norm": 2.1939926147460938, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.726962149143219, + "num_tokens": 166967859.0, + "step": 6497 + }, + { + "epoch": 0.7135954315835713, + "grad_norm": 2.3295395374298096, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7047715187072754, + "num_tokens": 166992579.0, + "step": 6498 + }, + { + "epoch": 0.713705249286185, + "grad_norm": 2.319145679473877, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6994280219078064, + "num_tokens": 167017149.0, + "step": 6499 + }, + { + "epoch": 0.7138150669887986, + "grad_norm": 1.9244660139083862, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7077018022537231, + "num_tokens": 167047466.0, + "step": 6500 + }, + { + "epoch": 0.7139248846914122, + "grad_norm": 2.0320231914520264, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6984681487083435, + "num_tokens": 167076709.0, + "step": 6501 + }, + { + "epoch": 0.7140347023940259, + "grad_norm": 2.1593830585479736, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6980949640274048, + "num_tokens": 167103466.0, + "step": 6502 + }, + { + "epoch": 0.7141445200966395, + "grad_norm": 2.169260263442993, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7098393440246582, + "num_tokens": 167127817.0, + "step": 6503 + }, + { + "epoch": 0.7142543377992533, + "grad_norm": 2.309684991836548, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7108149528503418, + "num_tokens": 167151613.0, + "step": 6504 + }, + { + "epoch": 0.7143641555018669, + "grad_norm": 1.8808714151382446, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7094117999076843, + "num_tokens": 167183014.0, + "step": 6505 + }, + { + "epoch": 0.7144739732044806, + "grad_norm": 2.43385648727417, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7365236282348633, + "num_tokens": 167203310.0, + "step": 6506 + }, + { + "epoch": 0.7145837909070942, + "grad_norm": 2.13657283782959, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7046079635620117, + "num_tokens": 167229461.0, + "step": 6507 + }, + { + "epoch": 0.7146936086097079, + "grad_norm": 1.9935357570648193, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6953544020652771, + "num_tokens": 167257997.0, + "step": 6508 + }, + { + "epoch": 0.7148034263123215, + "grad_norm": 2.108914852142334, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.701021134853363, + "num_tokens": 167285340.0, + "step": 6509 + }, + { + "epoch": 0.7149132440149352, + "grad_norm": 1.877974271774292, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7058395147323608, + "num_tokens": 167318360.0, + "step": 6510 + }, + { + "epoch": 0.7150230617175488, + "grad_norm": 2.3452131748199463, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7018961310386658, + "num_tokens": 167342447.0, + "step": 6511 + }, + { + "epoch": 0.7151328794201626, + "grad_norm": 2.181274175643921, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6793350577354431, + "num_tokens": 167371726.0, + "step": 6512 + }, + { + "epoch": 0.7152426971227762, + "grad_norm": 2.2781875133514404, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6915202140808105, + "num_tokens": 167398190.0, + "step": 6513 + }, + { + "epoch": 0.7153525148253899, + "grad_norm": 2.227565288543701, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6803690195083618, + "num_tokens": 167425503.0, + "step": 6514 + }, + { + "epoch": 0.7154623325280035, + "grad_norm": 2.0965514183044434, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6890960931777954, + "num_tokens": 167453833.0, + "step": 6515 + }, + { + "epoch": 0.7155721502306172, + "grad_norm": 2.248082160949707, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7188985347747803, + "num_tokens": 167476463.0, + "step": 6516 + }, + { + "epoch": 0.7156819679332308, + "grad_norm": 2.079355478286743, + "learning_rate": 1e-06, + "loss": 1.1209, + "mean_token_accuracy": 0.6604610085487366, + "num_tokens": 167505666.0, + "step": 6517 + }, + { + "epoch": 0.7157917856358444, + "grad_norm": 2.222137212753296, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7085956335067749, + "num_tokens": 167528879.0, + "step": 6518 + }, + { + "epoch": 0.7159016033384582, + "grad_norm": 2.362616777420044, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7159035801887512, + "num_tokens": 167551784.0, + "step": 6519 + }, + { + "epoch": 0.7160114210410718, + "grad_norm": 2.3496193885803223, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7201346755027771, + "num_tokens": 167574468.0, + "step": 6520 + }, + { + "epoch": 0.7161212387436855, + "grad_norm": 1.9149926900863647, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7230685949325562, + "num_tokens": 167605485.0, + "step": 6521 + }, + { + "epoch": 0.7162310564462991, + "grad_norm": 2.2313072681427, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7015953063964844, + "num_tokens": 167630654.0, + "step": 6522 + }, + { + "epoch": 0.7163408741489128, + "grad_norm": 1.9881504774093628, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7193682193756104, + "num_tokens": 167658378.0, + "step": 6523 + }, + { + "epoch": 0.7164506918515264, + "grad_norm": 2.2231605052948, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7100964784622192, + "num_tokens": 167683816.0, + "step": 6524 + }, + { + "epoch": 0.7165605095541401, + "grad_norm": 2.4477880001068115, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7149526476860046, + "num_tokens": 167704915.0, + "step": 6525 + }, + { + "epoch": 0.7166703272567538, + "grad_norm": 2.550659418106079, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.723391592502594, + "num_tokens": 167724038.0, + "step": 6526 + }, + { + "epoch": 0.7167801449593675, + "grad_norm": 2.1677470207214355, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7006731629371643, + "num_tokens": 167748316.0, + "step": 6527 + }, + { + "epoch": 0.7168899626619811, + "grad_norm": 2.3012008666992188, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7074432969093323, + "num_tokens": 167770952.0, + "step": 6528 + }, + { + "epoch": 0.7169997803645948, + "grad_norm": 1.927014946937561, + "learning_rate": 1e-06, + "loss": 1.0495, + "mean_token_accuracy": 0.6828428506851196, + "num_tokens": 167803629.0, + "step": 6529 + }, + { + "epoch": 0.7171095980672084, + "grad_norm": 2.0856494903564453, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6919927000999451, + "num_tokens": 167831643.0, + "step": 6530 + }, + { + "epoch": 0.7172194157698221, + "grad_norm": 2.2079594135284424, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7122358083724976, + "num_tokens": 167855308.0, + "step": 6531 + }, + { + "epoch": 0.7173292334724357, + "grad_norm": 2.2514307498931885, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6915096640586853, + "num_tokens": 167880542.0, + "step": 6532 + }, + { + "epoch": 0.7174390511750495, + "grad_norm": 2.437840461730957, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7143853902816772, + "num_tokens": 167903709.0, + "step": 6533 + }, + { + "epoch": 0.7175488688776631, + "grad_norm": 2.294368267059326, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7048429846763611, + "num_tokens": 167928099.0, + "step": 6534 + }, + { + "epoch": 0.7176586865802768, + "grad_norm": 2.173943281173706, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7393336296081543, + "num_tokens": 167952518.0, + "step": 6535 + }, + { + "epoch": 0.7177685042828904, + "grad_norm": 2.2938385009765625, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7115268707275391, + "num_tokens": 167976489.0, + "step": 6536 + }, + { + "epoch": 0.717878321985504, + "grad_norm": 2.295290231704712, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7075946927070618, + "num_tokens": 168000396.0, + "step": 6537 + }, + { + "epoch": 0.7179881396881177, + "grad_norm": 2.578179121017456, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7174398303031921, + "num_tokens": 168018530.0, + "step": 6538 + }, + { + "epoch": 0.7180979573907313, + "grad_norm": 2.3240807056427, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7017380595207214, + "num_tokens": 168042012.0, + "step": 6539 + }, + { + "epoch": 0.7182077750933451, + "grad_norm": 2.4882655143737793, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7268549203872681, + "num_tokens": 168060541.0, + "step": 6540 + }, + { + "epoch": 0.7183175927959587, + "grad_norm": 2.235201120376587, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.689262866973877, + "num_tokens": 168086040.0, + "step": 6541 + }, + { + "epoch": 0.7184274104985724, + "grad_norm": 2.22524094581604, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7128624320030212, + "num_tokens": 168110210.0, + "step": 6542 + }, + { + "epoch": 0.718537228201186, + "grad_norm": 2.128404140472412, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7356522679328918, + "num_tokens": 168134310.0, + "step": 6543 + }, + { + "epoch": 0.7186470459037997, + "grad_norm": 2.5023765563964844, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7109472751617432, + "num_tokens": 168155946.0, + "step": 6544 + }, + { + "epoch": 0.7187568636064133, + "grad_norm": 2.4073867797851562, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7323669195175171, + "num_tokens": 168176494.0, + "step": 6545 + }, + { + "epoch": 0.718866681309027, + "grad_norm": 2.144498109817505, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.682274580001831, + "num_tokens": 168202446.0, + "step": 6546 + }, + { + "epoch": 0.7189764990116406, + "grad_norm": 2.588756799697876, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7240357398986816, + "num_tokens": 168223086.0, + "step": 6547 + }, + { + "epoch": 0.7190863167142544, + "grad_norm": 2.181626319885254, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6905516386032104, + "num_tokens": 168250642.0, + "step": 6548 + }, + { + "epoch": 0.719196134416868, + "grad_norm": 2.3993897438049316, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7193603515625, + "num_tokens": 168273171.0, + "step": 6549 + }, + { + "epoch": 0.7193059521194817, + "grad_norm": 2.0638837814331055, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6988215446472168, + "num_tokens": 168301672.0, + "step": 6550 + }, + { + "epoch": 0.7194157698220953, + "grad_norm": 2.1409196853637695, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7137255668640137, + "num_tokens": 168327750.0, + "step": 6551 + }, + { + "epoch": 0.719525587524709, + "grad_norm": 2.2353460788726807, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7037684917449951, + "num_tokens": 168352577.0, + "step": 6552 + }, + { + "epoch": 0.7196354052273226, + "grad_norm": 2.0278968811035156, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.6968474388122559, + "num_tokens": 168380863.0, + "step": 6553 + }, + { + "epoch": 0.7197452229299363, + "grad_norm": 2.210505962371826, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.6955839395523071, + "num_tokens": 168406508.0, + "step": 6554 + }, + { + "epoch": 0.71985504063255, + "grad_norm": 2.4244701862335205, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7011211514472961, + "num_tokens": 168429004.0, + "step": 6555 + }, + { + "epoch": 0.7199648583351637, + "grad_norm": 2.174633026123047, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.716069757938385, + "num_tokens": 168453481.0, + "step": 6556 + }, + { + "epoch": 0.7200746760377773, + "grad_norm": 2.0527493953704834, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7172960042953491, + "num_tokens": 168480549.0, + "step": 6557 + }, + { + "epoch": 0.720184493740391, + "grad_norm": 2.2963433265686035, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7042705416679382, + "num_tokens": 168505444.0, + "step": 6558 + }, + { + "epoch": 0.7202943114430046, + "grad_norm": 2.5428690910339355, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7202130556106567, + "num_tokens": 168524282.0, + "step": 6559 + }, + { + "epoch": 0.7204041291456182, + "grad_norm": 2.283068895339966, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7195926904678345, + "num_tokens": 168548761.0, + "step": 6560 + }, + { + "epoch": 0.7205139468482319, + "grad_norm": 2.1651675701141357, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7164449095726013, + "num_tokens": 168572631.0, + "step": 6561 + }, + { + "epoch": 0.7206237645508456, + "grad_norm": 2.091902256011963, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7216802835464478, + "num_tokens": 168597626.0, + "step": 6562 + }, + { + "epoch": 0.7207335822534593, + "grad_norm": 2.4038913249969482, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7034537196159363, + "num_tokens": 168618961.0, + "step": 6563 + }, + { + "epoch": 0.7208433999560729, + "grad_norm": 2.0266902446746826, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.720826268196106, + "num_tokens": 168647842.0, + "step": 6564 + }, + { + "epoch": 0.7209532176586866, + "grad_norm": 2.0992531776428223, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7269383668899536, + "num_tokens": 168675372.0, + "step": 6565 + }, + { + "epoch": 0.7210630353613002, + "grad_norm": 2.2695200443267822, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.6977104544639587, + "num_tokens": 168700305.0, + "step": 6566 + }, + { + "epoch": 0.7211728530639139, + "grad_norm": 2.032543420791626, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.708829939365387, + "num_tokens": 168729413.0, + "step": 6567 + }, + { + "epoch": 0.7212826707665275, + "grad_norm": 2.1547417640686035, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6911249160766602, + "num_tokens": 168758518.0, + "step": 6568 + }, + { + "epoch": 0.7213924884691413, + "grad_norm": 2.4009768962860107, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7215399146080017, + "num_tokens": 168779784.0, + "step": 6569 + }, + { + "epoch": 0.7215023061717549, + "grad_norm": 2.128492832183838, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6936904788017273, + "num_tokens": 168808727.0, + "step": 6570 + }, + { + "epoch": 0.7216121238743686, + "grad_norm": 2.1256473064422607, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7154794335365295, + "num_tokens": 168835778.0, + "step": 6571 + }, + { + "epoch": 0.7217219415769822, + "grad_norm": 2.1455817222595215, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7086458206176758, + "num_tokens": 168860781.0, + "step": 6572 + }, + { + "epoch": 0.7218317592795959, + "grad_norm": 2.328474283218384, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7111037373542786, + "num_tokens": 168883510.0, + "step": 6573 + }, + { + "epoch": 0.7219415769822095, + "grad_norm": 1.9571388959884644, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7095125913619995, + "num_tokens": 168914254.0, + "step": 6574 + }, + { + "epoch": 0.7220513946848232, + "grad_norm": 2.187795877456665, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.69777512550354, + "num_tokens": 168939108.0, + "step": 6575 + }, + { + "epoch": 0.7221612123874368, + "grad_norm": 2.2946341037750244, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7092688083648682, + "num_tokens": 168963799.0, + "step": 6576 + }, + { + "epoch": 0.7222710300900506, + "grad_norm": 2.0994656085968018, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6995006799697876, + "num_tokens": 168991898.0, + "step": 6577 + }, + { + "epoch": 0.7223808477926642, + "grad_norm": 2.363672971725464, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7138022184371948, + "num_tokens": 169015580.0, + "step": 6578 + }, + { + "epoch": 0.7224906654952779, + "grad_norm": 2.306593418121338, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6972157955169678, + "num_tokens": 169040920.0, + "step": 6579 + }, + { + "epoch": 0.7226004831978915, + "grad_norm": 2.1108925342559814, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7154484987258911, + "num_tokens": 169068747.0, + "step": 6580 + }, + { + "epoch": 0.7227103009005051, + "grad_norm": 2.2079873085021973, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.679138720035553, + "num_tokens": 169095709.0, + "step": 6581 + }, + { + "epoch": 0.7228201186031188, + "grad_norm": 2.3356001377105713, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7045830488204956, + "num_tokens": 169118359.0, + "step": 6582 + }, + { + "epoch": 0.7229299363057324, + "grad_norm": 2.4576172828674316, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.704781711101532, + "num_tokens": 169137958.0, + "step": 6583 + }, + { + "epoch": 0.7230397540083462, + "grad_norm": 2.1739342212677, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7024451494216919, + "num_tokens": 169164349.0, + "step": 6584 + }, + { + "epoch": 0.7231495717109598, + "grad_norm": 2.0007355213165283, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6879175305366516, + "num_tokens": 169194013.0, + "step": 6585 + }, + { + "epoch": 0.7232593894135735, + "grad_norm": 2.149984121322632, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6917397379875183, + "num_tokens": 169221370.0, + "step": 6586 + }, + { + "epoch": 0.7233692071161871, + "grad_norm": 1.9907861948013306, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6949634552001953, + "num_tokens": 169250483.0, + "step": 6587 + }, + { + "epoch": 0.7234790248188008, + "grad_norm": 2.281029224395752, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7218223810195923, + "num_tokens": 169274364.0, + "step": 6588 + }, + { + "epoch": 0.7235888425214144, + "grad_norm": 2.188272714614868, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7097389101982117, + "num_tokens": 169300145.0, + "step": 6589 + }, + { + "epoch": 0.7236986602240281, + "grad_norm": 2.246983528137207, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7116805911064148, + "num_tokens": 169322688.0, + "step": 6590 + }, + { + "epoch": 0.7238084779266418, + "grad_norm": 2.3745205402374268, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.702838659286499, + "num_tokens": 169346118.0, + "step": 6591 + }, + { + "epoch": 0.7239182956292555, + "grad_norm": 2.2360005378723145, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7016329765319824, + "num_tokens": 169370762.0, + "step": 6592 + }, + { + "epoch": 0.7240281133318691, + "grad_norm": 2.3047754764556885, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.724467396736145, + "num_tokens": 169394507.0, + "step": 6593 + }, + { + "epoch": 0.7241379310344828, + "grad_norm": 2.0382559299468994, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7061548233032227, + "num_tokens": 169423220.0, + "step": 6594 + }, + { + "epoch": 0.7242477487370964, + "grad_norm": 2.511857748031616, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7157999277114868, + "num_tokens": 169442826.0, + "step": 6595 + }, + { + "epoch": 0.72435756643971, + "grad_norm": 2.270895481109619, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7023909091949463, + "num_tokens": 169465490.0, + "step": 6596 + }, + { + "epoch": 0.7244673841423237, + "grad_norm": 2.2078185081481934, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7138057351112366, + "num_tokens": 169490876.0, + "step": 6597 + }, + { + "epoch": 0.7245772018449375, + "grad_norm": 2.045339345932007, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7087892293930054, + "num_tokens": 169518792.0, + "step": 6598 + }, + { + "epoch": 0.7246870195475511, + "grad_norm": 2.0475199222564697, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6931437253952026, + "num_tokens": 169548313.0, + "step": 6599 + }, + { + "epoch": 0.7247968372501647, + "grad_norm": 2.2595837116241455, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.721208393573761, + "num_tokens": 169572425.0, + "step": 6600 + }, + { + "epoch": 0.7249066549527784, + "grad_norm": 2.3383169174194336, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7013977766036987, + "num_tokens": 169597852.0, + "step": 6601 + }, + { + "epoch": 0.725016472655392, + "grad_norm": 2.274077892303467, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7089249491691589, + "num_tokens": 169621633.0, + "step": 6602 + }, + { + "epoch": 0.7251262903580057, + "grad_norm": 2.1410562992095947, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6868776082992554, + "num_tokens": 169648088.0, + "step": 6603 + }, + { + "epoch": 0.7252361080606193, + "grad_norm": 2.1801507472991943, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7352457046508789, + "num_tokens": 169670718.0, + "step": 6604 + }, + { + "epoch": 0.725345925763233, + "grad_norm": 2.2218427658081055, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7124884724617004, + "num_tokens": 169694344.0, + "step": 6605 + }, + { + "epoch": 0.7254557434658467, + "grad_norm": 2.0392329692840576, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7077475786209106, + "num_tokens": 169723699.0, + "step": 6606 + }, + { + "epoch": 0.7255655611684604, + "grad_norm": 2.38295316696167, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7221683263778687, + "num_tokens": 169745901.0, + "step": 6607 + }, + { + "epoch": 0.725675378871074, + "grad_norm": 2.1474947929382324, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.718900203704834, + "num_tokens": 169772053.0, + "step": 6608 + }, + { + "epoch": 0.7257851965736877, + "grad_norm": 2.0337042808532715, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6940959692001343, + "num_tokens": 169802375.0, + "step": 6609 + }, + { + "epoch": 0.7258950142763013, + "grad_norm": 2.433478355407715, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7306361198425293, + "num_tokens": 169821003.0, + "step": 6610 + }, + { + "epoch": 0.726004831978915, + "grad_norm": 2.1140646934509277, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7246905565261841, + "num_tokens": 169845623.0, + "step": 6611 + }, + { + "epoch": 0.7261146496815286, + "grad_norm": 2.302572011947632, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.703650712966919, + "num_tokens": 169869192.0, + "step": 6612 + }, + { + "epoch": 0.7262244673841424, + "grad_norm": 2.1959519386291504, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6953182220458984, + "num_tokens": 169895457.0, + "step": 6613 + }, + { + "epoch": 0.726334285086756, + "grad_norm": 2.1266467571258545, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6792318820953369, + "num_tokens": 169921527.0, + "step": 6614 + }, + { + "epoch": 0.7264441027893697, + "grad_norm": 2.6374905109405518, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7069167494773865, + "num_tokens": 169941904.0, + "step": 6615 + }, + { + "epoch": 0.7265539204919833, + "grad_norm": 1.9904496669769287, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7230097055435181, + "num_tokens": 169972238.0, + "step": 6616 + }, + { + "epoch": 0.726663738194597, + "grad_norm": 2.4299864768981934, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7206142544746399, + "num_tokens": 169993434.0, + "step": 6617 + }, + { + "epoch": 0.7267735558972106, + "grad_norm": 2.285829782485962, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6895454525947571, + "num_tokens": 170016661.0, + "step": 6618 + }, + { + "epoch": 0.7268833735998242, + "grad_norm": 2.3171143531799316, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6951857209205627, + "num_tokens": 170040476.0, + "step": 6619 + }, + { + "epoch": 0.726993191302438, + "grad_norm": 1.9874624013900757, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7200524806976318, + "num_tokens": 170068847.0, + "step": 6620 + }, + { + "epoch": 0.7271030090050516, + "grad_norm": 1.9787763357162476, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7173995971679688, + "num_tokens": 170100204.0, + "step": 6621 + }, + { + "epoch": 0.7272128267076653, + "grad_norm": 1.9449830055236816, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7002049684524536, + "num_tokens": 170130647.0, + "step": 6622 + }, + { + "epoch": 0.7273226444102789, + "grad_norm": 2.223304271697998, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6960545182228088, + "num_tokens": 170158535.0, + "step": 6623 + }, + { + "epoch": 0.7274324621128926, + "grad_norm": 2.1543612480163574, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6749194860458374, + "num_tokens": 170184337.0, + "step": 6624 + }, + { + "epoch": 0.7275422798155062, + "grad_norm": 2.311575412750244, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6779940128326416, + "num_tokens": 170207997.0, + "step": 6625 + }, + { + "epoch": 0.7276520975181199, + "grad_norm": 2.3526899814605713, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7288720607757568, + "num_tokens": 170228355.0, + "step": 6626 + }, + { + "epoch": 0.7277619152207336, + "grad_norm": 1.8820419311523438, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6959539651870728, + "num_tokens": 170261652.0, + "step": 6627 + }, + { + "epoch": 0.7278717329233473, + "grad_norm": 2.432725429534912, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7226937413215637, + "num_tokens": 170283821.0, + "step": 6628 + }, + { + "epoch": 0.7279815506259609, + "grad_norm": 2.067396402359009, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6977829933166504, + "num_tokens": 170312429.0, + "step": 6629 + }, + { + "epoch": 0.7280913683285746, + "grad_norm": 2.3701155185699463, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7087001800537109, + "num_tokens": 170333563.0, + "step": 6630 + }, + { + "epoch": 0.7282011860311882, + "grad_norm": 1.904833197593689, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7105984091758728, + "num_tokens": 170365228.0, + "step": 6631 + }, + { + "epoch": 0.7283110037338019, + "grad_norm": 2.1112022399902344, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6967681646347046, + "num_tokens": 170389409.0, + "step": 6632 + }, + { + "epoch": 0.7284208214364155, + "grad_norm": 2.356186866760254, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7442846894264221, + "num_tokens": 170410180.0, + "step": 6633 + }, + { + "epoch": 0.7285306391390292, + "grad_norm": 2.3943724632263184, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7167020440101624, + "num_tokens": 170433012.0, + "step": 6634 + }, + { + "epoch": 0.7286404568416429, + "grad_norm": 2.2351584434509277, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7160787582397461, + "num_tokens": 170457308.0, + "step": 6635 + }, + { + "epoch": 0.7287502745442566, + "grad_norm": 1.9425022602081299, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6943397521972656, + "num_tokens": 170488887.0, + "step": 6636 + }, + { + "epoch": 0.7288600922468702, + "grad_norm": 2.2786295413970947, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7074236869812012, + "num_tokens": 170512071.0, + "step": 6637 + }, + { + "epoch": 0.7289699099494839, + "grad_norm": 1.9675934314727783, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7145540714263916, + "num_tokens": 170540906.0, + "step": 6638 + }, + { + "epoch": 0.7290797276520975, + "grad_norm": 2.1632797718048096, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7220458984375, + "num_tokens": 170564602.0, + "step": 6639 + }, + { + "epoch": 0.7291895453547111, + "grad_norm": 2.1856393814086914, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.701282262802124, + "num_tokens": 170589774.0, + "step": 6640 + }, + { + "epoch": 0.7292993630573248, + "grad_norm": 2.2819175720214844, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7138707041740417, + "num_tokens": 170613211.0, + "step": 6641 + }, + { + "epoch": 0.7294091807599385, + "grad_norm": 2.1343421936035156, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6901756525039673, + "num_tokens": 170639228.0, + "step": 6642 + }, + { + "epoch": 0.7295189984625522, + "grad_norm": 2.0116097927093506, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6988546848297119, + "num_tokens": 170668844.0, + "step": 6643 + }, + { + "epoch": 0.7296288161651658, + "grad_norm": 2.069200277328491, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7025598287582397, + "num_tokens": 170695975.0, + "step": 6644 + }, + { + "epoch": 0.7297386338677795, + "grad_norm": 2.1142475605010986, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7033945918083191, + "num_tokens": 170724561.0, + "step": 6645 + }, + { + "epoch": 0.7298484515703931, + "grad_norm": 1.8930679559707642, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7060906887054443, + "num_tokens": 170756687.0, + "step": 6646 + }, + { + "epoch": 0.7299582692730068, + "grad_norm": 2.118095874786377, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.703814685344696, + "num_tokens": 170780277.0, + "step": 6647 + }, + { + "epoch": 0.7300680869756204, + "grad_norm": 2.444669485092163, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7155393362045288, + "num_tokens": 170801355.0, + "step": 6648 + }, + { + "epoch": 0.7301779046782342, + "grad_norm": 2.4707844257354736, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7414363622665405, + "num_tokens": 170819863.0, + "step": 6649 + }, + { + "epoch": 0.7302877223808478, + "grad_norm": 2.098512887954712, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7262165546417236, + "num_tokens": 170844015.0, + "step": 6650 + }, + { + "epoch": 0.7303975400834615, + "grad_norm": 2.0293281078338623, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.6788378357887268, + "num_tokens": 170876612.0, + "step": 6651 + }, + { + "epoch": 0.7305073577860751, + "grad_norm": 2.1698248386383057, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7199816107749939, + "num_tokens": 170899703.0, + "step": 6652 + }, + { + "epoch": 0.7306171754886888, + "grad_norm": 2.356874704360962, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7108023762702942, + "num_tokens": 170921123.0, + "step": 6653 + }, + { + "epoch": 0.7307269931913024, + "grad_norm": 2.211242914199829, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7270817756652832, + "num_tokens": 170945516.0, + "step": 6654 + }, + { + "epoch": 0.7308368108939161, + "grad_norm": 2.351300001144409, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.699292004108429, + "num_tokens": 170968796.0, + "step": 6655 + }, + { + "epoch": 0.7309466285965298, + "grad_norm": 2.261591911315918, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7096753120422363, + "num_tokens": 170991889.0, + "step": 6656 + }, + { + "epoch": 0.7310564462991435, + "grad_norm": 2.179093837738037, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6877633929252625, + "num_tokens": 171020329.0, + "step": 6657 + }, + { + "epoch": 0.7311662640017571, + "grad_norm": 2.2831709384918213, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6889926195144653, + "num_tokens": 171046386.0, + "step": 6658 + }, + { + "epoch": 0.7312760817043708, + "grad_norm": 2.0706946849823, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7034729719161987, + "num_tokens": 171076084.0, + "step": 6659 + }, + { + "epoch": 0.7313858994069844, + "grad_norm": 2.1377787590026855, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6972940564155579, + "num_tokens": 171103886.0, + "step": 6660 + }, + { + "epoch": 0.731495717109598, + "grad_norm": 2.0874786376953125, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6994961500167847, + "num_tokens": 171133365.0, + "step": 6661 + }, + { + "epoch": 0.7316055348122117, + "grad_norm": 2.392524003982544, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7083466053009033, + "num_tokens": 171155816.0, + "step": 6662 + }, + { + "epoch": 0.7317153525148253, + "grad_norm": 2.060210704803467, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6941389441490173, + "num_tokens": 171183696.0, + "step": 6663 + }, + { + "epoch": 0.7318251702174391, + "grad_norm": 2.2429022789001465, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7267173528671265, + "num_tokens": 171207733.0, + "step": 6664 + }, + { + "epoch": 0.7319349879200527, + "grad_norm": 2.204629421234131, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6981777548789978, + "num_tokens": 171235984.0, + "step": 6665 + }, + { + "epoch": 0.7320448056226664, + "grad_norm": 2.592649221420288, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7379822731018066, + "num_tokens": 171254434.0, + "step": 6666 + }, + { + "epoch": 0.73215462332528, + "grad_norm": 2.7172837257385254, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7293174266815186, + "num_tokens": 171270735.0, + "step": 6667 + }, + { + "epoch": 0.7322644410278937, + "grad_norm": 1.97494375705719, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6993923187255859, + "num_tokens": 171300328.0, + "step": 6668 + }, + { + "epoch": 0.7323742587305073, + "grad_norm": 2.14628267288208, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6922727823257446, + "num_tokens": 171328362.0, + "step": 6669 + }, + { + "epoch": 0.732484076433121, + "grad_norm": 2.2361834049224854, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7159200310707092, + "num_tokens": 171351539.0, + "step": 6670 + }, + { + "epoch": 0.7325938941357347, + "grad_norm": 2.1165573596954346, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7047193646430969, + "num_tokens": 171378007.0, + "step": 6671 + }, + { + "epoch": 0.7327037118383484, + "grad_norm": 2.2506346702575684, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7159832715988159, + "num_tokens": 171401649.0, + "step": 6672 + }, + { + "epoch": 0.732813529540962, + "grad_norm": 2.000408887863159, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7284618020057678, + "num_tokens": 171430465.0, + "step": 6673 + }, + { + "epoch": 0.7329233472435757, + "grad_norm": 1.8864659070968628, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.687143087387085, + "num_tokens": 171464959.0, + "step": 6674 + }, + { + "epoch": 0.7330331649461893, + "grad_norm": 2.48350191116333, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7133280634880066, + "num_tokens": 171486285.0, + "step": 6675 + }, + { + "epoch": 0.733142982648803, + "grad_norm": 2.332415819168091, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7163747549057007, + "num_tokens": 171509221.0, + "step": 6676 + }, + { + "epoch": 0.7332528003514166, + "grad_norm": 2.116281509399414, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7136020660400391, + "num_tokens": 171535473.0, + "step": 6677 + }, + { + "epoch": 0.7333626180540304, + "grad_norm": 2.1058952808380127, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7254410982131958, + "num_tokens": 171561387.0, + "step": 6678 + }, + { + "epoch": 0.733472435756644, + "grad_norm": 2.4336254596710205, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7112365365028381, + "num_tokens": 171581210.0, + "step": 6679 + }, + { + "epoch": 0.7335822534592576, + "grad_norm": 2.0984039306640625, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6820370554924011, + "num_tokens": 171608756.0, + "step": 6680 + }, + { + "epoch": 0.7336920711618713, + "grad_norm": 2.0829079151153564, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6831233501434326, + "num_tokens": 171637774.0, + "step": 6681 + }, + { + "epoch": 0.7338018888644849, + "grad_norm": 1.917814016342163, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7111111879348755, + "num_tokens": 171668958.0, + "step": 6682 + }, + { + "epoch": 0.7339117065670986, + "grad_norm": 1.8048847913742065, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6916969418525696, + "num_tokens": 171707224.0, + "step": 6683 + }, + { + "epoch": 0.7340215242697122, + "grad_norm": 2.1638195514678955, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6928995847702026, + "num_tokens": 171734016.0, + "step": 6684 + }, + { + "epoch": 0.734131341972326, + "grad_norm": 2.244403600692749, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6974524259567261, + "num_tokens": 171759369.0, + "step": 6685 + }, + { + "epoch": 0.7342411596749396, + "grad_norm": 2.002481460571289, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7243510484695435, + "num_tokens": 171788595.0, + "step": 6686 + }, + { + "epoch": 0.7343509773775533, + "grad_norm": 2.0767390727996826, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6701866984367371, + "num_tokens": 171817981.0, + "step": 6687 + }, + { + "epoch": 0.7344607950801669, + "grad_norm": 2.335564613342285, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7161319255828857, + "num_tokens": 171843904.0, + "step": 6688 + }, + { + "epoch": 0.7345706127827806, + "grad_norm": 2.0139336585998535, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6961231231689453, + "num_tokens": 171872822.0, + "step": 6689 + }, + { + "epoch": 0.7346804304853942, + "grad_norm": 2.164036989212036, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7081615924835205, + "num_tokens": 171899521.0, + "step": 6690 + }, + { + "epoch": 0.7347902481880079, + "grad_norm": 2.3109283447265625, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7205747961997986, + "num_tokens": 171921633.0, + "step": 6691 + }, + { + "epoch": 0.7349000658906216, + "grad_norm": 2.113142490386963, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6878172159194946, + "num_tokens": 171949695.0, + "step": 6692 + }, + { + "epoch": 0.7350098835932353, + "grad_norm": 2.017688035964966, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.704509437084198, + "num_tokens": 171978223.0, + "step": 6693 + }, + { + "epoch": 0.7351197012958489, + "grad_norm": 2.028869152069092, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7200735807418823, + "num_tokens": 172006839.0, + "step": 6694 + }, + { + "epoch": 0.7352295189984626, + "grad_norm": 2.4422404766082764, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7098193168640137, + "num_tokens": 172026628.0, + "step": 6695 + }, + { + "epoch": 0.7353393367010762, + "grad_norm": 2.290055513381958, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6865524053573608, + "num_tokens": 172052412.0, + "step": 6696 + }, + { + "epoch": 0.7354491544036899, + "grad_norm": 2.2616944313049316, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7015483975410461, + "num_tokens": 172076558.0, + "step": 6697 + }, + { + "epoch": 0.7355589721063035, + "grad_norm": 2.2190182209014893, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6965328454971313, + "num_tokens": 172104054.0, + "step": 6698 + }, + { + "epoch": 0.7356687898089171, + "grad_norm": 2.226425886154175, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7140588760375977, + "num_tokens": 172128933.0, + "step": 6699 + }, + { + "epoch": 0.7357786075115309, + "grad_norm": 2.364675998687744, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.723307728767395, + "num_tokens": 172149427.0, + "step": 6700 + }, + { + "epoch": 0.7358884252141445, + "grad_norm": 2.294865369796753, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7088857889175415, + "num_tokens": 172171888.0, + "step": 6701 + }, + { + "epoch": 0.7359982429167582, + "grad_norm": 1.8740004301071167, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7130714058876038, + "num_tokens": 172205931.0, + "step": 6702 + }, + { + "epoch": 0.7361080606193718, + "grad_norm": 1.8767787218093872, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7125819325447083, + "num_tokens": 172237478.0, + "step": 6703 + }, + { + "epoch": 0.7362178783219855, + "grad_norm": 2.4589810371398926, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7323777675628662, + "num_tokens": 172258036.0, + "step": 6704 + }, + { + "epoch": 0.7363276960245991, + "grad_norm": 2.277585983276367, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7099306583404541, + "num_tokens": 172281728.0, + "step": 6705 + }, + { + "epoch": 0.7364375137272128, + "grad_norm": 2.4743549823760986, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7076958417892456, + "num_tokens": 172302918.0, + "step": 6706 + }, + { + "epoch": 0.7365473314298265, + "grad_norm": 2.1963539123535156, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7270846366882324, + "num_tokens": 172326507.0, + "step": 6707 + }, + { + "epoch": 0.7366571491324402, + "grad_norm": 2.391508102416992, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7165666818618774, + "num_tokens": 172347650.0, + "step": 6708 + }, + { + "epoch": 0.7367669668350538, + "grad_norm": 2.1304545402526855, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.692110002040863, + "num_tokens": 172376829.0, + "step": 6709 + }, + { + "epoch": 0.7368767845376675, + "grad_norm": 2.3749001026153564, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6910374164581299, + "num_tokens": 172400220.0, + "step": 6710 + }, + { + "epoch": 0.7369866022402811, + "grad_norm": 2.4501407146453857, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7194392681121826, + "num_tokens": 172420359.0, + "step": 6711 + }, + { + "epoch": 0.7370964199428948, + "grad_norm": 2.0376837253570557, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6895228624343872, + "num_tokens": 172450060.0, + "step": 6712 + }, + { + "epoch": 0.7372062376455084, + "grad_norm": 2.117985963821411, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7024292945861816, + "num_tokens": 172475560.0, + "step": 6713 + }, + { + "epoch": 0.7373160553481222, + "grad_norm": 2.1110422611236572, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7181231379508972, + "num_tokens": 172501500.0, + "step": 6714 + }, + { + "epoch": 0.7374258730507358, + "grad_norm": 2.044189691543579, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7015458941459656, + "num_tokens": 172531477.0, + "step": 6715 + }, + { + "epoch": 0.7375356907533495, + "grad_norm": 2.0799200534820557, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6980795860290527, + "num_tokens": 172558596.0, + "step": 6716 + }, + { + "epoch": 0.7376455084559631, + "grad_norm": 2.4645144939422607, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6907273530960083, + "num_tokens": 172580986.0, + "step": 6717 + }, + { + "epoch": 0.7377553261585768, + "grad_norm": 2.285306215286255, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.711241602897644, + "num_tokens": 172604198.0, + "step": 6718 + }, + { + "epoch": 0.7378651438611904, + "grad_norm": 2.341909885406494, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7094298601150513, + "num_tokens": 172627452.0, + "step": 6719 + }, + { + "epoch": 0.737974961563804, + "grad_norm": 2.321138381958008, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7253393530845642, + "num_tokens": 172650090.0, + "step": 6720 + }, + { + "epoch": 0.7380847792664178, + "grad_norm": 2.2389566898345947, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.712260901927948, + "num_tokens": 172674365.0, + "step": 6721 + }, + { + "epoch": 0.7381945969690314, + "grad_norm": 2.2323224544525146, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7001148462295532, + "num_tokens": 172697687.0, + "step": 6722 + }, + { + "epoch": 0.7383044146716451, + "grad_norm": 2.2604498863220215, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7108354568481445, + "num_tokens": 172721952.0, + "step": 6723 + }, + { + "epoch": 0.7384142323742587, + "grad_norm": 2.317180871963501, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7137183547019958, + "num_tokens": 172745292.0, + "step": 6724 + }, + { + "epoch": 0.7385240500768724, + "grad_norm": 2.396775960922241, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7391995191574097, + "num_tokens": 172763370.0, + "step": 6725 + }, + { + "epoch": 0.738633867779486, + "grad_norm": 2.1988489627838135, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6983394622802734, + "num_tokens": 172789113.0, + "step": 6726 + }, + { + "epoch": 0.7387436854820997, + "grad_norm": 2.2743515968322754, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7233299016952515, + "num_tokens": 172811793.0, + "step": 6727 + }, + { + "epoch": 0.7388535031847133, + "grad_norm": 1.8851691484451294, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7080508470535278, + "num_tokens": 172843899.0, + "step": 6728 + }, + { + "epoch": 0.7389633208873271, + "grad_norm": 2.5836985111236572, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7207328081130981, + "num_tokens": 172863269.0, + "step": 6729 + }, + { + "epoch": 0.7390731385899407, + "grad_norm": 2.130152463912964, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7031583786010742, + "num_tokens": 172889762.0, + "step": 6730 + }, + { + "epoch": 0.7391829562925544, + "grad_norm": 2.122840404510498, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7123113870620728, + "num_tokens": 172915346.0, + "step": 6731 + }, + { + "epoch": 0.739292773995168, + "grad_norm": 2.481769561767578, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7022659778594971, + "num_tokens": 172935271.0, + "step": 6732 + }, + { + "epoch": 0.7394025916977817, + "grad_norm": 2.3313958644866943, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7219860553741455, + "num_tokens": 172957498.0, + "step": 6733 + }, + { + "epoch": 0.7395124094003953, + "grad_norm": 2.3538835048675537, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6923648118972778, + "num_tokens": 172981887.0, + "step": 6734 + }, + { + "epoch": 0.739622227103009, + "grad_norm": 2.119023561477661, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7269382476806641, + "num_tokens": 173009116.0, + "step": 6735 + }, + { + "epoch": 0.7397320448056227, + "grad_norm": 2.2438883781433105, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.693683385848999, + "num_tokens": 173037546.0, + "step": 6736 + }, + { + "epoch": 0.7398418625082364, + "grad_norm": 2.6219537258148193, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7000927925109863, + "num_tokens": 173056668.0, + "step": 6737 + }, + { + "epoch": 0.73995168021085, + "grad_norm": 2.2674312591552734, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7017163038253784, + "num_tokens": 173081212.0, + "step": 6738 + }, + { + "epoch": 0.7400614979134637, + "grad_norm": 2.2428677082061768, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7126284241676331, + "num_tokens": 173107348.0, + "step": 6739 + }, + { + "epoch": 0.7401713156160773, + "grad_norm": 2.196495294570923, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7204889059066772, + "num_tokens": 173133133.0, + "step": 6740 + }, + { + "epoch": 0.7402811333186909, + "grad_norm": 2.342437267303467, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7369384765625, + "num_tokens": 173154767.0, + "step": 6741 + }, + { + "epoch": 0.7403909510213046, + "grad_norm": 2.300124168395996, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7300525903701782, + "num_tokens": 173177216.0, + "step": 6742 + }, + { + "epoch": 0.7405007687239183, + "grad_norm": 1.9993352890014648, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7115724086761475, + "num_tokens": 173206818.0, + "step": 6743 + }, + { + "epoch": 0.740610586426532, + "grad_norm": 1.9981738328933716, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.698334813117981, + "num_tokens": 173237791.0, + "step": 6744 + }, + { + "epoch": 0.7407204041291456, + "grad_norm": 2.4440834522247314, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7219120264053345, + "num_tokens": 173256135.0, + "step": 6745 + }, + { + "epoch": 0.7408302218317593, + "grad_norm": 2.288465738296509, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7124712467193604, + "num_tokens": 173281168.0, + "step": 6746 + }, + { + "epoch": 0.7409400395343729, + "grad_norm": 2.257523775100708, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7078641653060913, + "num_tokens": 173306664.0, + "step": 6747 + }, + { + "epoch": 0.7410498572369866, + "grad_norm": 2.0202243328094482, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6840950846672058, + "num_tokens": 173337871.0, + "step": 6748 + }, + { + "epoch": 0.7411596749396002, + "grad_norm": 2.042494058609009, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6905858516693115, + "num_tokens": 173367184.0, + "step": 6749 + }, + { + "epoch": 0.741269492642214, + "grad_norm": 2.328998565673828, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.7066284418106079, + "num_tokens": 173389677.0, + "step": 6750 + }, + { + "epoch": 0.7413793103448276, + "grad_norm": 2.3135385513305664, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.693179726600647, + "num_tokens": 173412926.0, + "step": 6751 + }, + { + "epoch": 0.7414891280474413, + "grad_norm": 2.3580565452575684, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7365021705627441, + "num_tokens": 173433910.0, + "step": 6752 + }, + { + "epoch": 0.7415989457500549, + "grad_norm": 2.4049010276794434, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7075878381729126, + "num_tokens": 173455482.0, + "step": 6753 + }, + { + "epoch": 0.7417087634526686, + "grad_norm": 2.4971635341644287, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.6873259544372559, + "num_tokens": 173478805.0, + "step": 6754 + }, + { + "epoch": 0.7418185811552822, + "grad_norm": 2.12668514251709, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7040278911590576, + "num_tokens": 173504522.0, + "step": 6755 + }, + { + "epoch": 0.7419283988578959, + "grad_norm": 2.0947186946868896, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7142879962921143, + "num_tokens": 173530774.0, + "step": 6756 + }, + { + "epoch": 0.7420382165605095, + "grad_norm": 2.1248691082000732, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6974116563796997, + "num_tokens": 173558548.0, + "step": 6757 + }, + { + "epoch": 0.7421480342631233, + "grad_norm": 2.2082014083862305, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6883969902992249, + "num_tokens": 173584676.0, + "step": 6758 + }, + { + "epoch": 0.7422578519657369, + "grad_norm": 2.1647260189056396, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6890524625778198, + "num_tokens": 173610372.0, + "step": 6759 + }, + { + "epoch": 0.7423676696683505, + "grad_norm": 2.100667715072632, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7048014998435974, + "num_tokens": 173637519.0, + "step": 6760 + }, + { + "epoch": 0.7424774873709642, + "grad_norm": 2.50606107711792, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7018113732337952, + "num_tokens": 173658793.0, + "step": 6761 + }, + { + "epoch": 0.7425873050735778, + "grad_norm": 1.9835129976272583, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6992924213409424, + "num_tokens": 173687492.0, + "step": 6762 + }, + { + "epoch": 0.7426971227761915, + "grad_norm": 2.0718984603881836, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7034153938293457, + "num_tokens": 173714038.0, + "step": 6763 + }, + { + "epoch": 0.7428069404788051, + "grad_norm": 2.2438337802886963, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6820464134216309, + "num_tokens": 173739327.0, + "step": 6764 + }, + { + "epoch": 0.7429167581814189, + "grad_norm": 2.123169422149658, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.6783099174499512, + "num_tokens": 173767627.0, + "step": 6765 + }, + { + "epoch": 0.7430265758840325, + "grad_norm": 2.1751039028167725, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7058572173118591, + "num_tokens": 173793767.0, + "step": 6766 + }, + { + "epoch": 0.7431363935866462, + "grad_norm": 2.0754663944244385, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6924406290054321, + "num_tokens": 173821666.0, + "step": 6767 + }, + { + "epoch": 0.7432462112892598, + "grad_norm": 2.098231792449951, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6961241364479065, + "num_tokens": 173849304.0, + "step": 6768 + }, + { + "epoch": 0.7433560289918735, + "grad_norm": 2.297194719314575, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7208658456802368, + "num_tokens": 173871548.0, + "step": 6769 + }, + { + "epoch": 0.7434658466944871, + "grad_norm": 2.2714076042175293, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.694210410118103, + "num_tokens": 173896040.0, + "step": 6770 + }, + { + "epoch": 0.7435756643971008, + "grad_norm": 2.2423272132873535, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7251697182655334, + "num_tokens": 173919397.0, + "step": 6771 + }, + { + "epoch": 0.7436854820997145, + "grad_norm": 2.0961382389068604, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7285411357879639, + "num_tokens": 173945441.0, + "step": 6772 + }, + { + "epoch": 0.7437952998023282, + "grad_norm": 2.226022243499756, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7251135110855103, + "num_tokens": 173968884.0, + "step": 6773 + }, + { + "epoch": 0.7439051175049418, + "grad_norm": 2.005746841430664, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.710159420967102, + "num_tokens": 173998153.0, + "step": 6774 + }, + { + "epoch": 0.7440149352075555, + "grad_norm": 2.196343421936035, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7101512551307678, + "num_tokens": 174022883.0, + "step": 6775 + }, + { + "epoch": 0.7441247529101691, + "grad_norm": 2.1809115409851074, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.7005500793457031, + "num_tokens": 174048209.0, + "step": 6776 + }, + { + "epoch": 0.7442345706127828, + "grad_norm": 2.180788993835449, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.7039729952812195, + "num_tokens": 174074127.0, + "step": 6777 + }, + { + "epoch": 0.7443443883153964, + "grad_norm": 2.5054054260253906, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7095927000045776, + "num_tokens": 174094317.0, + "step": 6778 + }, + { + "epoch": 0.7444542060180102, + "grad_norm": 2.2525999546051025, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7086428999900818, + "num_tokens": 174119077.0, + "step": 6779 + }, + { + "epoch": 0.7445640237206238, + "grad_norm": 2.201779365539551, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7341281175613403, + "num_tokens": 174143510.0, + "step": 6780 + }, + { + "epoch": 0.7446738414232374, + "grad_norm": 2.148043632507324, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6986396312713623, + "num_tokens": 174169906.0, + "step": 6781 + }, + { + "epoch": 0.7447836591258511, + "grad_norm": 2.3319103717803955, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6883444786071777, + "num_tokens": 174192200.0, + "step": 6782 + }, + { + "epoch": 0.7448934768284647, + "grad_norm": 2.7834692001342773, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7306715250015259, + "num_tokens": 174210710.0, + "step": 6783 + }, + { + "epoch": 0.7450032945310784, + "grad_norm": 2.272488594055176, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6930310726165771, + "num_tokens": 174236079.0, + "step": 6784 + }, + { + "epoch": 0.745113112233692, + "grad_norm": 2.1688642501831055, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6967461109161377, + "num_tokens": 174263003.0, + "step": 6785 + }, + { + "epoch": 0.7452229299363057, + "grad_norm": 2.208402395248413, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6916472911834717, + "num_tokens": 174290126.0, + "step": 6786 + }, + { + "epoch": 0.7453327476389194, + "grad_norm": 2.8048839569091797, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6989994049072266, + "num_tokens": 174309530.0, + "step": 6787 + }, + { + "epoch": 0.7454425653415331, + "grad_norm": 1.9942970275878906, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6936528086662292, + "num_tokens": 174341044.0, + "step": 6788 + }, + { + "epoch": 0.7455523830441467, + "grad_norm": 2.2276721000671387, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7027108669281006, + "num_tokens": 174365124.0, + "step": 6789 + }, + { + "epoch": 0.7456622007467604, + "grad_norm": 2.529832124710083, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7207242250442505, + "num_tokens": 174384154.0, + "step": 6790 + }, + { + "epoch": 0.745772018449374, + "grad_norm": 2.625791311264038, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7272458076477051, + "num_tokens": 174403226.0, + "step": 6791 + }, + { + "epoch": 0.7458818361519877, + "grad_norm": 2.097078323364258, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7090067863464355, + "num_tokens": 174430777.0, + "step": 6792 + }, + { + "epoch": 0.7459916538546013, + "grad_norm": 1.8753511905670166, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6960574388504028, + "num_tokens": 174465355.0, + "step": 6793 + }, + { + "epoch": 0.7461014715572151, + "grad_norm": 2.2125000953674316, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6908633708953857, + "num_tokens": 174492782.0, + "step": 6794 + }, + { + "epoch": 0.7462112892598287, + "grad_norm": 2.44260573387146, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.694149374961853, + "num_tokens": 174516786.0, + "step": 6795 + }, + { + "epoch": 0.7463211069624424, + "grad_norm": 2.3740310668945312, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7152748107910156, + "num_tokens": 174540981.0, + "step": 6796 + }, + { + "epoch": 0.746430924665056, + "grad_norm": 1.9837557077407837, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.743678629398346, + "num_tokens": 174567890.0, + "step": 6797 + }, + { + "epoch": 0.7465407423676697, + "grad_norm": 2.2247369289398193, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7321337461471558, + "num_tokens": 174591235.0, + "step": 6798 + }, + { + "epoch": 0.7466505600702833, + "grad_norm": 1.9438703060150146, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7256696224212646, + "num_tokens": 174621207.0, + "step": 6799 + }, + { + "epoch": 0.7467603777728969, + "grad_norm": 2.214052438735962, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7338061928749084, + "num_tokens": 174644836.0, + "step": 6800 + }, + { + "epoch": 0.7468701954755107, + "grad_norm": 2.3336143493652344, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6985063552856445, + "num_tokens": 174667436.0, + "step": 6801 + }, + { + "epoch": 0.7469800131781243, + "grad_norm": 2.4540762901306152, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7042743563652039, + "num_tokens": 174690944.0, + "step": 6802 + }, + { + "epoch": 0.747089830880738, + "grad_norm": 2.230630397796631, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7271794080734253, + "num_tokens": 174715086.0, + "step": 6803 + }, + { + "epoch": 0.7471996485833516, + "grad_norm": 2.671769142150879, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7190929651260376, + "num_tokens": 174735571.0, + "step": 6804 + }, + { + "epoch": 0.7473094662859653, + "grad_norm": 2.2505311965942383, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6955416202545166, + "num_tokens": 174759600.0, + "step": 6805 + }, + { + "epoch": 0.7474192839885789, + "grad_norm": 2.4298741817474365, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7104741334915161, + "num_tokens": 174781969.0, + "step": 6806 + }, + { + "epoch": 0.7475291016911926, + "grad_norm": 2.085461378097534, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7002121210098267, + "num_tokens": 174810291.0, + "step": 6807 + }, + { + "epoch": 0.7476389193938063, + "grad_norm": 2.1860058307647705, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6798875331878662, + "num_tokens": 174837728.0, + "step": 6808 + }, + { + "epoch": 0.74774873709642, + "grad_norm": 2.1889185905456543, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7258930206298828, + "num_tokens": 174862523.0, + "step": 6809 + }, + { + "epoch": 0.7478585547990336, + "grad_norm": 1.9888625144958496, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7107559442520142, + "num_tokens": 174892045.0, + "step": 6810 + }, + { + "epoch": 0.7479683725016473, + "grad_norm": 2.5069611072540283, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7237343788146973, + "num_tokens": 174911792.0, + "step": 6811 + }, + { + "epoch": 0.7480781902042609, + "grad_norm": 2.2872955799102783, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7231784462928772, + "num_tokens": 174933885.0, + "step": 6812 + }, + { + "epoch": 0.7481880079068746, + "grad_norm": 2.2318737506866455, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7067722082138062, + "num_tokens": 174961056.0, + "step": 6813 + }, + { + "epoch": 0.7482978256094882, + "grad_norm": 2.101522445678711, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7191550135612488, + "num_tokens": 174987317.0, + "step": 6814 + }, + { + "epoch": 0.7484076433121019, + "grad_norm": 2.0268537998199463, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7100623846054077, + "num_tokens": 175015238.0, + "step": 6815 + }, + { + "epoch": 0.7485174610147156, + "grad_norm": 2.4329850673675537, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7019691467285156, + "num_tokens": 175037141.0, + "step": 6816 + }, + { + "epoch": 0.7486272787173293, + "grad_norm": 2.4185056686401367, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6940432786941528, + "num_tokens": 175058885.0, + "step": 6817 + }, + { + "epoch": 0.7487370964199429, + "grad_norm": 1.982279896736145, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6935874223709106, + "num_tokens": 175087207.0, + "step": 6818 + }, + { + "epoch": 0.7488469141225566, + "grad_norm": 2.2083280086517334, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7209396362304688, + "num_tokens": 175109500.0, + "step": 6819 + }, + { + "epoch": 0.7489567318251702, + "grad_norm": 2.363532781600952, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7078007459640503, + "num_tokens": 175131533.0, + "step": 6820 + }, + { + "epoch": 0.7490665495277838, + "grad_norm": 2.3700904846191406, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6919218897819519, + "num_tokens": 175155481.0, + "step": 6821 + }, + { + "epoch": 0.7491763672303975, + "grad_norm": 2.1538848876953125, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.690775990486145, + "num_tokens": 175185432.0, + "step": 6822 + }, + { + "epoch": 0.7492861849330112, + "grad_norm": 2.318368434906006, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6935266256332397, + "num_tokens": 175209847.0, + "step": 6823 + }, + { + "epoch": 0.7493960026356249, + "grad_norm": 2.074497699737549, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.717386782169342, + "num_tokens": 175237073.0, + "step": 6824 + }, + { + "epoch": 0.7495058203382385, + "grad_norm": 2.3665852546691895, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7178829908370972, + "num_tokens": 175260681.0, + "step": 6825 + }, + { + "epoch": 0.7496156380408522, + "grad_norm": 2.1910977363586426, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.704861581325531, + "num_tokens": 175287697.0, + "step": 6826 + }, + { + "epoch": 0.7497254557434658, + "grad_norm": 2.16085147857666, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7099974751472473, + "num_tokens": 175314095.0, + "step": 6827 + }, + { + "epoch": 0.7498352734460795, + "grad_norm": 2.1266403198242188, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7030069828033447, + "num_tokens": 175340534.0, + "step": 6828 + }, + { + "epoch": 0.7499450911486931, + "grad_norm": 2.03080153465271, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7182019352912903, + "num_tokens": 175368129.0, + "step": 6829 + }, + { + "epoch": 0.7500549088513069, + "grad_norm": 2.0198144912719727, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7257706522941589, + "num_tokens": 175395064.0, + "step": 6830 + }, + { + "epoch": 0.7501647265539205, + "grad_norm": 2.2177233695983887, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6870042085647583, + "num_tokens": 175421117.0, + "step": 6831 + }, + { + "epoch": 0.7502745442565342, + "grad_norm": 1.972261667251587, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7101480960845947, + "num_tokens": 175449021.0, + "step": 6832 + }, + { + "epoch": 0.7503843619591478, + "grad_norm": 2.302478313446045, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7087156772613525, + "num_tokens": 175471913.0, + "step": 6833 + }, + { + "epoch": 0.7504941796617615, + "grad_norm": 2.2536027431488037, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7351617813110352, + "num_tokens": 175495518.0, + "step": 6834 + }, + { + "epoch": 0.7506039973643751, + "grad_norm": 2.326993465423584, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7072856426239014, + "num_tokens": 175517595.0, + "step": 6835 + }, + { + "epoch": 0.7507138150669888, + "grad_norm": 2.1889162063598633, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6910908222198486, + "num_tokens": 175546096.0, + "step": 6836 + }, + { + "epoch": 0.7508236327696025, + "grad_norm": 2.320943593978882, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6852920055389404, + "num_tokens": 175570545.0, + "step": 6837 + }, + { + "epoch": 0.7509334504722162, + "grad_norm": 2.0834274291992188, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7205071449279785, + "num_tokens": 175599408.0, + "step": 6838 + }, + { + "epoch": 0.7510432681748298, + "grad_norm": 1.9671616554260254, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7104708552360535, + "num_tokens": 175631859.0, + "step": 6839 + }, + { + "epoch": 0.7511530858774434, + "grad_norm": 2.133244514465332, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7103307843208313, + "num_tokens": 175656254.0, + "step": 6840 + }, + { + "epoch": 0.7512629035800571, + "grad_norm": 2.2939274311065674, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7132480144500732, + "num_tokens": 175678555.0, + "step": 6841 + }, + { + "epoch": 0.7513727212826707, + "grad_norm": 2.2937779426574707, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.728614091873169, + "num_tokens": 175701079.0, + "step": 6842 + }, + { + "epoch": 0.7514825389852844, + "grad_norm": 1.832167148590088, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7063091993331909, + "num_tokens": 175735061.0, + "step": 6843 + }, + { + "epoch": 0.7515923566878981, + "grad_norm": 2.2876830101013184, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7015178203582764, + "num_tokens": 175759539.0, + "step": 6844 + }, + { + "epoch": 0.7517021743905118, + "grad_norm": 2.171098470687866, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7091394066810608, + "num_tokens": 175786014.0, + "step": 6845 + }, + { + "epoch": 0.7518119920931254, + "grad_norm": 1.7615705728530884, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7082309126853943, + "num_tokens": 175821926.0, + "step": 6846 + }, + { + "epoch": 0.7519218097957391, + "grad_norm": 2.047288417816162, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6957377791404724, + "num_tokens": 175850950.0, + "step": 6847 + }, + { + "epoch": 0.7520316274983527, + "grad_norm": 2.0544002056121826, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7020565271377563, + "num_tokens": 175878718.0, + "step": 6848 + }, + { + "epoch": 0.7521414452009664, + "grad_norm": 1.912805438041687, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.6827844977378845, + "num_tokens": 175910726.0, + "step": 6849 + }, + { + "epoch": 0.75225126290358, + "grad_norm": 1.998695731163025, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7031059861183167, + "num_tokens": 175939364.0, + "step": 6850 + }, + { + "epoch": 0.7523610806061937, + "grad_norm": 2.013648271560669, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6853058934211731, + "num_tokens": 175968737.0, + "step": 6851 + }, + { + "epoch": 0.7524708983088074, + "grad_norm": 2.1975300312042236, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6906040906906128, + "num_tokens": 175994895.0, + "step": 6852 + }, + { + "epoch": 0.7525807160114211, + "grad_norm": 2.3504042625427246, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.7097675800323486, + "num_tokens": 176019764.0, + "step": 6853 + }, + { + "epoch": 0.7526905337140347, + "grad_norm": 2.125082492828369, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7027307748794556, + "num_tokens": 176045263.0, + "step": 6854 + }, + { + "epoch": 0.7528003514166484, + "grad_norm": 2.0470328330993652, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.723838210105896, + "num_tokens": 176071894.0, + "step": 6855 + }, + { + "epoch": 0.752910169119262, + "grad_norm": 2.0118653774261475, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7034505009651184, + "num_tokens": 176103288.0, + "step": 6856 + }, + { + "epoch": 0.7530199868218757, + "grad_norm": 2.1135523319244385, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6930937767028809, + "num_tokens": 176128777.0, + "step": 6857 + }, + { + "epoch": 0.7531298045244893, + "grad_norm": 2.167414426803589, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6769276857376099, + "num_tokens": 176155920.0, + "step": 6858 + }, + { + "epoch": 0.7532396222271031, + "grad_norm": 1.9210246801376343, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7076256275177002, + "num_tokens": 176186054.0, + "step": 6859 + }, + { + "epoch": 0.7533494399297167, + "grad_norm": 2.2698333263397217, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7099315524101257, + "num_tokens": 176208928.0, + "step": 6860 + }, + { + "epoch": 0.7534592576323303, + "grad_norm": 2.3126118183135986, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6932588815689087, + "num_tokens": 176234168.0, + "step": 6861 + }, + { + "epoch": 0.753569075334944, + "grad_norm": 2.09328293800354, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7110927700996399, + "num_tokens": 176262091.0, + "step": 6862 + }, + { + "epoch": 0.7536788930375576, + "grad_norm": 1.9211719036102295, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7226179838180542, + "num_tokens": 176293150.0, + "step": 6863 + }, + { + "epoch": 0.7537887107401713, + "grad_norm": 2.16091251373291, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6970117688179016, + "num_tokens": 176318993.0, + "step": 6864 + }, + { + "epoch": 0.7538985284427849, + "grad_norm": 2.098776340484619, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.7097927927970886, + "num_tokens": 176345867.0, + "step": 6865 + }, + { + "epoch": 0.7540083461453987, + "grad_norm": 2.095320224761963, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7191305160522461, + "num_tokens": 176373152.0, + "step": 6866 + }, + { + "epoch": 0.7541181638480123, + "grad_norm": 2.437891721725464, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7009491920471191, + "num_tokens": 176394455.0, + "step": 6867 + }, + { + "epoch": 0.754227981550626, + "grad_norm": 2.148782968521118, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7257373332977295, + "num_tokens": 176419042.0, + "step": 6868 + }, + { + "epoch": 0.7543377992532396, + "grad_norm": 2.0219297409057617, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7036280632019043, + "num_tokens": 176448843.0, + "step": 6869 + }, + { + "epoch": 0.7544476169558533, + "grad_norm": 1.9635907411575317, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7040300369262695, + "num_tokens": 176477752.0, + "step": 6870 + }, + { + "epoch": 0.7545574346584669, + "grad_norm": 2.4424755573272705, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7118185758590698, + "num_tokens": 176498015.0, + "step": 6871 + }, + { + "epoch": 0.7546672523610806, + "grad_norm": 2.242555618286133, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7107601761817932, + "num_tokens": 176523987.0, + "step": 6872 + }, + { + "epoch": 0.7547770700636943, + "grad_norm": 2.156658172607422, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7035787105560303, + "num_tokens": 176548970.0, + "step": 6873 + }, + { + "epoch": 0.754886887766308, + "grad_norm": 1.9939008951187134, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7141821384429932, + "num_tokens": 176579033.0, + "step": 6874 + }, + { + "epoch": 0.7549967054689216, + "grad_norm": 2.273632764816284, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7021932005882263, + "num_tokens": 176602589.0, + "step": 6875 + }, + { + "epoch": 0.7551065231715353, + "grad_norm": 2.293365955352783, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6985325813293457, + "num_tokens": 176628271.0, + "step": 6876 + }, + { + "epoch": 0.7552163408741489, + "grad_norm": 2.434386968612671, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7016503214836121, + "num_tokens": 176648933.0, + "step": 6877 + }, + { + "epoch": 0.7553261585767626, + "grad_norm": 2.282792091369629, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6823059320449829, + "num_tokens": 176673540.0, + "step": 6878 + }, + { + "epoch": 0.7554359762793762, + "grad_norm": 1.9018306732177734, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7184338569641113, + "num_tokens": 176707195.0, + "step": 6879 + }, + { + "epoch": 0.7555457939819898, + "grad_norm": 2.7216365337371826, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7146337032318115, + "num_tokens": 176723725.0, + "step": 6880 + }, + { + "epoch": 0.7556556116846036, + "grad_norm": 2.1370460987091064, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7025716304779053, + "num_tokens": 176751382.0, + "step": 6881 + }, + { + "epoch": 0.7557654293872172, + "grad_norm": 2.0164616107940674, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7150380611419678, + "num_tokens": 176778691.0, + "step": 6882 + }, + { + "epoch": 0.7558752470898309, + "grad_norm": 2.2030792236328125, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7039254903793335, + "num_tokens": 176804982.0, + "step": 6883 + }, + { + "epoch": 0.7559850647924445, + "grad_norm": 2.0879878997802734, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7021051645278931, + "num_tokens": 176831708.0, + "step": 6884 + }, + { + "epoch": 0.7560948824950582, + "grad_norm": 2.4681246280670166, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7160261869430542, + "num_tokens": 176851736.0, + "step": 6885 + }, + { + "epoch": 0.7562047001976718, + "grad_norm": 1.9187610149383545, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6961950659751892, + "num_tokens": 176884224.0, + "step": 6886 + }, + { + "epoch": 0.7563145179002855, + "grad_norm": 2.209481716156006, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7201510071754456, + "num_tokens": 176907464.0, + "step": 6887 + }, + { + "epoch": 0.7564243356028992, + "grad_norm": 2.040412664413452, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6982118487358093, + "num_tokens": 176937261.0, + "step": 6888 + }, + { + "epoch": 0.7565341533055129, + "grad_norm": 2.04630446434021, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6922202110290527, + "num_tokens": 176965441.0, + "step": 6889 + }, + { + "epoch": 0.7566439710081265, + "grad_norm": 2.3350324630737305, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7077428102493286, + "num_tokens": 176989873.0, + "step": 6890 + }, + { + "epoch": 0.7567537887107402, + "grad_norm": 1.9476733207702637, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7299104928970337, + "num_tokens": 177018420.0, + "step": 6891 + }, + { + "epoch": 0.7568636064133538, + "grad_norm": 2.4340579509735107, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7070644497871399, + "num_tokens": 177038243.0, + "step": 6892 + }, + { + "epoch": 0.7569734241159675, + "grad_norm": 2.309972047805786, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6969940066337585, + "num_tokens": 177062449.0, + "step": 6893 + }, + { + "epoch": 0.7570832418185811, + "grad_norm": 2.477264642715454, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7206586599349976, + "num_tokens": 177081225.0, + "step": 6894 + }, + { + "epoch": 0.7571930595211949, + "grad_norm": 2.169949531555176, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7208001613616943, + "num_tokens": 177104954.0, + "step": 6895 + }, + { + "epoch": 0.7573028772238085, + "grad_norm": 2.405975818634033, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7217923402786255, + "num_tokens": 177126153.0, + "step": 6896 + }, + { + "epoch": 0.7574126949264222, + "grad_norm": 2.2006614208221436, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6995759606361389, + "num_tokens": 177150321.0, + "step": 6897 + }, + { + "epoch": 0.7575225126290358, + "grad_norm": 2.3203392028808594, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7159916162490845, + "num_tokens": 177171917.0, + "step": 6898 + }, + { + "epoch": 0.7576323303316495, + "grad_norm": 1.9222840070724487, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6872932314872742, + "num_tokens": 177204025.0, + "step": 6899 + }, + { + "epoch": 0.7577421480342631, + "grad_norm": 2.093463182449341, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6976751089096069, + "num_tokens": 177231411.0, + "step": 6900 + }, + { + "epoch": 0.7578519657368767, + "grad_norm": 1.9269344806671143, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.708624005317688, + "num_tokens": 177264415.0, + "step": 6901 + }, + { + "epoch": 0.7579617834394905, + "grad_norm": 2.474802255630493, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.6992676258087158, + "num_tokens": 177283351.0, + "step": 6902 + }, + { + "epoch": 0.7580716011421041, + "grad_norm": 2.065873384475708, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7110244035720825, + "num_tokens": 177309398.0, + "step": 6903 + }, + { + "epoch": 0.7581814188447178, + "grad_norm": 2.0836644172668457, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7094050645828247, + "num_tokens": 177336516.0, + "step": 6904 + }, + { + "epoch": 0.7582912365473314, + "grad_norm": 1.9908385276794434, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7164726853370667, + "num_tokens": 177365093.0, + "step": 6905 + }, + { + "epoch": 0.7584010542499451, + "grad_norm": 2.099514961242676, + "learning_rate": 1e-06, + "loss": 1.0949, + "mean_token_accuracy": 0.6718912124633789, + "num_tokens": 177391863.0, + "step": 6906 + }, + { + "epoch": 0.7585108719525587, + "grad_norm": 2.389052629470825, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7362055778503418, + "num_tokens": 177413099.0, + "step": 6907 + }, + { + "epoch": 0.7586206896551724, + "grad_norm": 2.119396448135376, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6934981346130371, + "num_tokens": 177440815.0, + "step": 6908 + }, + { + "epoch": 0.758730507357786, + "grad_norm": 2.1718132495880127, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7117267847061157, + "num_tokens": 177465427.0, + "step": 6909 + }, + { + "epoch": 0.7588403250603998, + "grad_norm": 2.24884295463562, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7263303399085999, + "num_tokens": 177488387.0, + "step": 6910 + }, + { + "epoch": 0.7589501427630134, + "grad_norm": 1.96678626537323, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7399686574935913, + "num_tokens": 177514964.0, + "step": 6911 + }, + { + "epoch": 0.7590599604656271, + "grad_norm": 1.9706250429153442, + "learning_rate": 1e-06, + "loss": 1.1035, + "mean_token_accuracy": 0.6687518358230591, + "num_tokens": 177548528.0, + "step": 6912 + }, + { + "epoch": 0.7591697781682407, + "grad_norm": 2.5340967178344727, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7127602100372314, + "num_tokens": 177569660.0, + "step": 6913 + }, + { + "epoch": 0.7592795958708544, + "grad_norm": 2.086778163909912, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7089153528213501, + "num_tokens": 177596557.0, + "step": 6914 + }, + { + "epoch": 0.759389413573468, + "grad_norm": 2.2042529582977295, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7060580253601074, + "num_tokens": 177622142.0, + "step": 6915 + }, + { + "epoch": 0.7594992312760817, + "grad_norm": 2.47879958152771, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6975669860839844, + "num_tokens": 177645382.0, + "step": 6916 + }, + { + "epoch": 0.7596090489786954, + "grad_norm": 1.9613616466522217, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7012937068939209, + "num_tokens": 177676379.0, + "step": 6917 + }, + { + "epoch": 0.7597188666813091, + "grad_norm": 2.110788106918335, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6894996762275696, + "num_tokens": 177703913.0, + "step": 6918 + }, + { + "epoch": 0.7598286843839227, + "grad_norm": 2.163154363632202, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7303782105445862, + "num_tokens": 177729119.0, + "step": 6919 + }, + { + "epoch": 0.7599385020865363, + "grad_norm": 2.066053628921509, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7027137279510498, + "num_tokens": 177756535.0, + "step": 6920 + }, + { + "epoch": 0.76004831978915, + "grad_norm": 1.9486808776855469, + "learning_rate": 1e-06, + "loss": 1.0979, + "mean_token_accuracy": 0.6711419224739075, + "num_tokens": 177789824.0, + "step": 6921 + }, + { + "epoch": 0.7601581374917636, + "grad_norm": 2.249603271484375, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7324461936950684, + "num_tokens": 177814577.0, + "step": 6922 + }, + { + "epoch": 0.7602679551943773, + "grad_norm": 2.218339204788208, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7317871451377869, + "num_tokens": 177838176.0, + "step": 6923 + }, + { + "epoch": 0.760377772896991, + "grad_norm": 2.2201027870178223, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6938998699188232, + "num_tokens": 177865364.0, + "step": 6924 + }, + { + "epoch": 0.7604875905996047, + "grad_norm": 2.7178916931152344, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7177456617355347, + "num_tokens": 177884506.0, + "step": 6925 + }, + { + "epoch": 0.7605974083022183, + "grad_norm": 2.445910930633545, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7240002155303955, + "num_tokens": 177904818.0, + "step": 6926 + }, + { + "epoch": 0.760707226004832, + "grad_norm": 2.0887539386749268, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7099874019622803, + "num_tokens": 177931614.0, + "step": 6927 + }, + { + "epoch": 0.7608170437074456, + "grad_norm": 1.8838632106781006, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6948995590209961, + "num_tokens": 177965341.0, + "step": 6928 + }, + { + "epoch": 0.7609268614100593, + "grad_norm": 2.6103341579437256, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7204862236976624, + "num_tokens": 177982642.0, + "step": 6929 + }, + { + "epoch": 0.7610366791126729, + "grad_norm": 2.495802402496338, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6854570508003235, + "num_tokens": 178005105.0, + "step": 6930 + }, + { + "epoch": 0.7611464968152867, + "grad_norm": 2.2887117862701416, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7098366022109985, + "num_tokens": 178029274.0, + "step": 6931 + }, + { + "epoch": 0.7612563145179003, + "grad_norm": 2.169935941696167, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6938701868057251, + "num_tokens": 178055580.0, + "step": 6932 + }, + { + "epoch": 0.761366132220514, + "grad_norm": 2.1419477462768555, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7097955346107483, + "num_tokens": 178080586.0, + "step": 6933 + }, + { + "epoch": 0.7614759499231276, + "grad_norm": 2.1666886806488037, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7222299575805664, + "num_tokens": 178104693.0, + "step": 6934 + }, + { + "epoch": 0.7615857676257413, + "grad_norm": 2.182598352432251, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.707895040512085, + "num_tokens": 178129886.0, + "step": 6935 + }, + { + "epoch": 0.7616955853283549, + "grad_norm": 1.8066775798797607, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6958484649658203, + "num_tokens": 178165699.0, + "step": 6936 + }, + { + "epoch": 0.7618054030309686, + "grad_norm": 2.4762301445007324, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7142561674118042, + "num_tokens": 178186204.0, + "step": 6937 + }, + { + "epoch": 0.7619152207335822, + "grad_norm": 2.3124239444732666, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7317367792129517, + "num_tokens": 178207813.0, + "step": 6938 + }, + { + "epoch": 0.762025038436196, + "grad_norm": 2.057826042175293, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7266843318939209, + "num_tokens": 178235543.0, + "step": 6939 + }, + { + "epoch": 0.7621348561388096, + "grad_norm": 2.196808099746704, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7058233022689819, + "num_tokens": 178261522.0, + "step": 6940 + }, + { + "epoch": 0.7622446738414232, + "grad_norm": 2.3336541652679443, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7264597415924072, + "num_tokens": 178283292.0, + "step": 6941 + }, + { + "epoch": 0.7623544915440369, + "grad_norm": 2.0988097190856934, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7356590032577515, + "num_tokens": 178309287.0, + "step": 6942 + }, + { + "epoch": 0.7624643092466505, + "grad_norm": 2.1858460903167725, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7442587614059448, + "num_tokens": 178334848.0, + "step": 6943 + }, + { + "epoch": 0.7625741269492642, + "grad_norm": 2.256587266921997, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.693166971206665, + "num_tokens": 178359122.0, + "step": 6944 + }, + { + "epoch": 0.7626839446518778, + "grad_norm": 2.0737833976745605, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7125818729400635, + "num_tokens": 178385421.0, + "step": 6945 + }, + { + "epoch": 0.7627937623544916, + "grad_norm": 2.09367299079895, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7118435502052307, + "num_tokens": 178413695.0, + "step": 6946 + }, + { + "epoch": 0.7629035800571052, + "grad_norm": 2.5804107189178467, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7119537591934204, + "num_tokens": 178432250.0, + "step": 6947 + }, + { + "epoch": 0.7630133977597189, + "grad_norm": 2.1541247367858887, + "learning_rate": 1e-06, + "loss": 1.0936, + "mean_token_accuracy": 0.6761828064918518, + "num_tokens": 178460837.0, + "step": 6948 + }, + { + "epoch": 0.7631232154623325, + "grad_norm": 2.2234323024749756, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7254012227058411, + "num_tokens": 178484077.0, + "step": 6949 + }, + { + "epoch": 0.7632330331649462, + "grad_norm": 2.2860231399536133, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7129116058349609, + "num_tokens": 178506721.0, + "step": 6950 + }, + { + "epoch": 0.7633428508675598, + "grad_norm": 2.0102198123931885, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6901706457138062, + "num_tokens": 178538076.0, + "step": 6951 + }, + { + "epoch": 0.7634526685701735, + "grad_norm": 2.1048381328582764, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.693391740322113, + "num_tokens": 178565956.0, + "step": 6952 + }, + { + "epoch": 0.7635624862727872, + "grad_norm": 2.2235536575317383, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7335612773895264, + "num_tokens": 178590070.0, + "step": 6953 + }, + { + "epoch": 0.7636723039754009, + "grad_norm": 2.2027981281280518, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6888037919998169, + "num_tokens": 178615784.0, + "step": 6954 + }, + { + "epoch": 0.7637821216780145, + "grad_norm": 2.0338008403778076, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7374151945114136, + "num_tokens": 178641101.0, + "step": 6955 + }, + { + "epoch": 0.7638919393806282, + "grad_norm": 1.9978805780410767, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.698601484298706, + "num_tokens": 178669180.0, + "step": 6956 + }, + { + "epoch": 0.7640017570832418, + "grad_norm": 2.42337965965271, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7015126943588257, + "num_tokens": 178690065.0, + "step": 6957 + }, + { + "epoch": 0.7641115747858555, + "grad_norm": 2.0807645320892334, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6932243704795837, + "num_tokens": 178717830.0, + "step": 6958 + }, + { + "epoch": 0.7642213924884691, + "grad_norm": 2.1765308380126953, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7258832454681396, + "num_tokens": 178741824.0, + "step": 6959 + }, + { + "epoch": 0.7643312101910829, + "grad_norm": 2.3137526512145996, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7199660539627075, + "num_tokens": 178762830.0, + "step": 6960 + }, + { + "epoch": 0.7644410278936965, + "grad_norm": 2.7589187622070312, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7142625451087952, + "num_tokens": 178780837.0, + "step": 6961 + }, + { + "epoch": 0.7645508455963101, + "grad_norm": 2.1065313816070557, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6930162906646729, + "num_tokens": 178807584.0, + "step": 6962 + }, + { + "epoch": 0.7646606632989238, + "grad_norm": 1.8606013059616089, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.7000147104263306, + "num_tokens": 178841359.0, + "step": 6963 + }, + { + "epoch": 0.7647704810015374, + "grad_norm": 2.1664621829986572, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6990708112716675, + "num_tokens": 178866994.0, + "step": 6964 + }, + { + "epoch": 0.7648802987041511, + "grad_norm": 2.0932717323303223, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7226076126098633, + "num_tokens": 178893526.0, + "step": 6965 + }, + { + "epoch": 0.7649901164067647, + "grad_norm": 2.1292107105255127, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7059219479560852, + "num_tokens": 178923166.0, + "step": 6966 + }, + { + "epoch": 0.7650999341093784, + "grad_norm": 2.4393537044525146, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7295593023300171, + "num_tokens": 178944469.0, + "step": 6967 + }, + { + "epoch": 0.7652097518119921, + "grad_norm": 2.2011215686798096, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.697848379611969, + "num_tokens": 178970437.0, + "step": 6968 + }, + { + "epoch": 0.7653195695146058, + "grad_norm": 2.1048028469085693, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7181267142295837, + "num_tokens": 178996890.0, + "step": 6969 + }, + { + "epoch": 0.7654293872172194, + "grad_norm": 1.9422502517700195, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6879357695579529, + "num_tokens": 179026915.0, + "step": 6970 + }, + { + "epoch": 0.7655392049198331, + "grad_norm": 2.2832534313201904, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6991114616394043, + "num_tokens": 179050574.0, + "step": 6971 + }, + { + "epoch": 0.7656490226224467, + "grad_norm": 2.2094218730926514, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7159236669540405, + "num_tokens": 179074465.0, + "step": 6972 + }, + { + "epoch": 0.7657588403250604, + "grad_norm": 2.271913766860962, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.726410448551178, + "num_tokens": 179096528.0, + "step": 6973 + }, + { + "epoch": 0.765868658027674, + "grad_norm": 2.35732102394104, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7038024663925171, + "num_tokens": 179119398.0, + "step": 6974 + }, + { + "epoch": 0.7659784757302878, + "grad_norm": 1.9927546977996826, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7108581066131592, + "num_tokens": 179150472.0, + "step": 6975 + }, + { + "epoch": 0.7660882934329014, + "grad_norm": 2.2251954078674316, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6956707239151001, + "num_tokens": 179174916.0, + "step": 6976 + }, + { + "epoch": 0.7661981111355151, + "grad_norm": 2.2534093856811523, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7050329446792603, + "num_tokens": 179198419.0, + "step": 6977 + }, + { + "epoch": 0.7663079288381287, + "grad_norm": 2.4559645652770996, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7244126200675964, + "num_tokens": 179217081.0, + "step": 6978 + }, + { + "epoch": 0.7664177465407424, + "grad_norm": 2.460395336151123, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6984003782272339, + "num_tokens": 179237687.0, + "step": 6979 + }, + { + "epoch": 0.766527564243356, + "grad_norm": 2.04268217086792, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6861295700073242, + "num_tokens": 179266573.0, + "step": 6980 + }, + { + "epoch": 0.7666373819459696, + "grad_norm": 2.073063850402832, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7164058685302734, + "num_tokens": 179293572.0, + "step": 6981 + }, + { + "epoch": 0.7667471996485834, + "grad_norm": 2.123875617980957, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.674620509147644, + "num_tokens": 179321584.0, + "step": 6982 + }, + { + "epoch": 0.766857017351197, + "grad_norm": 2.473806858062744, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6900467872619629, + "num_tokens": 179342621.0, + "step": 6983 + }, + { + "epoch": 0.7669668350538107, + "grad_norm": 2.0075151920318604, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7060198783874512, + "num_tokens": 179372712.0, + "step": 6984 + }, + { + "epoch": 0.7670766527564243, + "grad_norm": 2.074348211288452, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7310969829559326, + "num_tokens": 179398845.0, + "step": 6985 + }, + { + "epoch": 0.767186470459038, + "grad_norm": 2.3406169414520264, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7152679562568665, + "num_tokens": 179420271.0, + "step": 6986 + }, + { + "epoch": 0.7672962881616516, + "grad_norm": 2.3555338382720947, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7090915441513062, + "num_tokens": 179442673.0, + "step": 6987 + }, + { + "epoch": 0.7674061058642653, + "grad_norm": 1.9247080087661743, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6953991651535034, + "num_tokens": 179475098.0, + "step": 6988 + }, + { + "epoch": 0.767515923566879, + "grad_norm": 2.1856515407562256, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7047258615493774, + "num_tokens": 179499343.0, + "step": 6989 + }, + { + "epoch": 0.7676257412694927, + "grad_norm": 2.0606136322021484, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.7023236751556396, + "num_tokens": 179530552.0, + "step": 6990 + }, + { + "epoch": 0.7677355589721063, + "grad_norm": 2.2151222229003906, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7015333771705627, + "num_tokens": 179554938.0, + "step": 6991 + }, + { + "epoch": 0.76784537667472, + "grad_norm": 1.9744795560836792, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.682785153388977, + "num_tokens": 179588793.0, + "step": 6992 + }, + { + "epoch": 0.7679551943773336, + "grad_norm": 2.054659843444824, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7140205502510071, + "num_tokens": 179616782.0, + "step": 6993 + }, + { + "epoch": 0.7680650120799473, + "grad_norm": 2.242168426513672, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7331365942955017, + "num_tokens": 179639264.0, + "step": 6994 + }, + { + "epoch": 0.7681748297825609, + "grad_norm": 2.0925989151000977, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7047516703605652, + "num_tokens": 179665925.0, + "step": 6995 + }, + { + "epoch": 0.7682846474851747, + "grad_norm": 2.0121614933013916, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6918919086456299, + "num_tokens": 179695690.0, + "step": 6996 + }, + { + "epoch": 0.7683944651877883, + "grad_norm": 2.2770068645477295, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6967010498046875, + "num_tokens": 179719752.0, + "step": 6997 + }, + { + "epoch": 0.768504282890402, + "grad_norm": 2.2236194610595703, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6885460615158081, + "num_tokens": 179744900.0, + "step": 6998 + }, + { + "epoch": 0.7686141005930156, + "grad_norm": 2.113976001739502, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.713712215423584, + "num_tokens": 179773006.0, + "step": 6999 + }, + { + "epoch": 0.7687239182956292, + "grad_norm": 2.538119077682495, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.727997899055481, + "num_tokens": 179794354.0, + "step": 7000 + }, + { + "epoch": 0.7688337359982429, + "grad_norm": 2.2708702087402344, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7121087312698364, + "num_tokens": 179816399.0, + "step": 7001 + }, + { + "epoch": 0.7689435537008565, + "grad_norm": 2.130349636077881, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6995772123336792, + "num_tokens": 179843561.0, + "step": 7002 + }, + { + "epoch": 0.7690533714034702, + "grad_norm": 2.0526533126831055, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7057830095291138, + "num_tokens": 179870242.0, + "step": 7003 + }, + { + "epoch": 0.7691631891060839, + "grad_norm": 2.0453498363494873, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7272244691848755, + "num_tokens": 179898702.0, + "step": 7004 + }, + { + "epoch": 0.7692730068086976, + "grad_norm": 2.3325729370117188, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7134815454483032, + "num_tokens": 179919978.0, + "step": 7005 + }, + { + "epoch": 0.7693828245113112, + "grad_norm": 2.436417579650879, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7162994146347046, + "num_tokens": 179940238.0, + "step": 7006 + }, + { + "epoch": 0.7694926422139249, + "grad_norm": 2.0617330074310303, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7222955226898193, + "num_tokens": 179967118.0, + "step": 7007 + }, + { + "epoch": 0.7696024599165385, + "grad_norm": 2.241764783859253, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7022028565406799, + "num_tokens": 179992315.0, + "step": 7008 + }, + { + "epoch": 0.7697122776191522, + "grad_norm": 2.193488597869873, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7189911603927612, + "num_tokens": 180016860.0, + "step": 7009 + }, + { + "epoch": 0.7698220953217658, + "grad_norm": 1.9444732666015625, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6943241357803345, + "num_tokens": 180047599.0, + "step": 7010 + }, + { + "epoch": 0.7699319130243796, + "grad_norm": 2.1741795539855957, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7037456631660461, + "num_tokens": 180072521.0, + "step": 7011 + }, + { + "epoch": 0.7700417307269932, + "grad_norm": 2.21909236907959, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.702580451965332, + "num_tokens": 180097919.0, + "step": 7012 + }, + { + "epoch": 0.7701515484296069, + "grad_norm": 2.44333553314209, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7291601896286011, + "num_tokens": 180117450.0, + "step": 7013 + }, + { + "epoch": 0.7702613661322205, + "grad_norm": 2.415288209915161, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7155638933181763, + "num_tokens": 180138737.0, + "step": 7014 + }, + { + "epoch": 0.7703711838348342, + "grad_norm": 2.0467936992645264, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6939023733139038, + "num_tokens": 180166160.0, + "step": 7015 + }, + { + "epoch": 0.7704810015374478, + "grad_norm": 2.115898609161377, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.71058189868927, + "num_tokens": 180193139.0, + "step": 7016 + }, + { + "epoch": 0.7705908192400615, + "grad_norm": 2.376810073852539, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7104450464248657, + "num_tokens": 180215169.0, + "step": 7017 + }, + { + "epoch": 0.7707006369426752, + "grad_norm": 2.3891820907592773, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7116265892982483, + "num_tokens": 180235132.0, + "step": 7018 + }, + { + "epoch": 0.7708104546452889, + "grad_norm": 2.0846235752105713, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6915227174758911, + "num_tokens": 180263345.0, + "step": 7019 + }, + { + "epoch": 0.7709202723479025, + "grad_norm": 2.028080701828003, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6848102807998657, + "num_tokens": 180292680.0, + "step": 7020 + }, + { + "epoch": 0.7710300900505161, + "grad_norm": 2.0428500175476074, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7191364169120789, + "num_tokens": 180317860.0, + "step": 7021 + }, + { + "epoch": 0.7711399077531298, + "grad_norm": 2.1749963760375977, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6866263151168823, + "num_tokens": 180345450.0, + "step": 7022 + }, + { + "epoch": 0.7712497254557434, + "grad_norm": 2.372990846633911, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.7115525007247925, + "num_tokens": 180367018.0, + "step": 7023 + }, + { + "epoch": 0.7713595431583571, + "grad_norm": 2.294480085372925, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7184755802154541, + "num_tokens": 180391290.0, + "step": 7024 + }, + { + "epoch": 0.7714693608609708, + "grad_norm": 2.116114854812622, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6974615454673767, + "num_tokens": 180419692.0, + "step": 7025 + }, + { + "epoch": 0.7715791785635845, + "grad_norm": 2.204155206680298, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7013026475906372, + "num_tokens": 180444748.0, + "step": 7026 + }, + { + "epoch": 0.7716889962661981, + "grad_norm": 2.3385159969329834, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7105637192726135, + "num_tokens": 180470640.0, + "step": 7027 + }, + { + "epoch": 0.7717988139688118, + "grad_norm": 2.2969846725463867, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.710425853729248, + "num_tokens": 180493982.0, + "step": 7028 + }, + { + "epoch": 0.7719086316714254, + "grad_norm": 2.398068428039551, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.722568154335022, + "num_tokens": 180518594.0, + "step": 7029 + }, + { + "epoch": 0.7720184493740391, + "grad_norm": 2.539196014404297, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7098287343978882, + "num_tokens": 180539077.0, + "step": 7030 + }, + { + "epoch": 0.7721282670766527, + "grad_norm": 2.2345662117004395, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7278902530670166, + "num_tokens": 180563239.0, + "step": 7031 + }, + { + "epoch": 0.7722380847792664, + "grad_norm": 2.3263304233551025, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6897259950637817, + "num_tokens": 180587505.0, + "step": 7032 + }, + { + "epoch": 0.7723479024818801, + "grad_norm": 2.414517641067505, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7075116634368896, + "num_tokens": 180609885.0, + "step": 7033 + }, + { + "epoch": 0.7724577201844938, + "grad_norm": 2.158609628677368, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6913522481918335, + "num_tokens": 180636321.0, + "step": 7034 + }, + { + "epoch": 0.7725675378871074, + "grad_norm": 2.414803981781006, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7275519967079163, + "num_tokens": 180657273.0, + "step": 7035 + }, + { + "epoch": 0.7726773555897211, + "grad_norm": 2.0083260536193848, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7233738899230957, + "num_tokens": 180684748.0, + "step": 7036 + }, + { + "epoch": 0.7727871732923347, + "grad_norm": 2.2557270526885986, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6960247755050659, + "num_tokens": 180711689.0, + "step": 7037 + }, + { + "epoch": 0.7728969909949484, + "grad_norm": 2.514261245727539, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7109847068786621, + "num_tokens": 180733363.0, + "step": 7038 + }, + { + "epoch": 0.773006808697562, + "grad_norm": 2.2674930095672607, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.705628514289856, + "num_tokens": 180756890.0, + "step": 7039 + }, + { + "epoch": 0.7731166264001758, + "grad_norm": 2.331106662750244, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6964571475982666, + "num_tokens": 180780872.0, + "step": 7040 + }, + { + "epoch": 0.7732264441027894, + "grad_norm": 2.266200542449951, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6993985176086426, + "num_tokens": 180805975.0, + "step": 7041 + }, + { + "epoch": 0.773336261805403, + "grad_norm": 2.033280611038208, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.713580846786499, + "num_tokens": 180833916.0, + "step": 7042 + }, + { + "epoch": 0.7734460795080167, + "grad_norm": 2.2225048542022705, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7202212810516357, + "num_tokens": 180856937.0, + "step": 7043 + }, + { + "epoch": 0.7735558972106303, + "grad_norm": 2.1908299922943115, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7188321948051453, + "num_tokens": 180883391.0, + "step": 7044 + }, + { + "epoch": 0.773665714913244, + "grad_norm": 2.0461394786834717, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.68758624792099, + "num_tokens": 180913911.0, + "step": 7045 + }, + { + "epoch": 0.7737755326158576, + "grad_norm": 2.0413060188293457, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7023707032203674, + "num_tokens": 180940510.0, + "step": 7046 + }, + { + "epoch": 0.7738853503184714, + "grad_norm": 2.082453489303589, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7139235734939575, + "num_tokens": 180968075.0, + "step": 7047 + }, + { + "epoch": 0.773995168021085, + "grad_norm": 2.280463695526123, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6859899163246155, + "num_tokens": 180993796.0, + "step": 7048 + }, + { + "epoch": 0.7741049857236987, + "grad_norm": 2.239042043685913, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7114037871360779, + "num_tokens": 181018536.0, + "step": 7049 + }, + { + "epoch": 0.7742148034263123, + "grad_norm": 2.186948776245117, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7400092482566833, + "num_tokens": 181042277.0, + "step": 7050 + }, + { + "epoch": 0.774324621128926, + "grad_norm": 2.2482712268829346, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.710631251335144, + "num_tokens": 181066804.0, + "step": 7051 + }, + { + "epoch": 0.7744344388315396, + "grad_norm": 2.208361864089966, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7184268832206726, + "num_tokens": 181091839.0, + "step": 7052 + }, + { + "epoch": 0.7745442565341533, + "grad_norm": 2.6619374752044678, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7157785296440125, + "num_tokens": 181110458.0, + "step": 7053 + }, + { + "epoch": 0.774654074236767, + "grad_norm": 2.2587108612060547, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7123937606811523, + "num_tokens": 181135499.0, + "step": 7054 + }, + { + "epoch": 0.7747638919393807, + "grad_norm": 2.198166608810425, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7090596556663513, + "num_tokens": 181162095.0, + "step": 7055 + }, + { + "epoch": 0.7748737096419943, + "grad_norm": 2.231024742126465, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7105998992919922, + "num_tokens": 181187636.0, + "step": 7056 + }, + { + "epoch": 0.774983527344608, + "grad_norm": 2.3676862716674805, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7113340497016907, + "num_tokens": 181209307.0, + "step": 7057 + }, + { + "epoch": 0.7750933450472216, + "grad_norm": 2.2368099689483643, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7041491866111755, + "num_tokens": 181234752.0, + "step": 7058 + }, + { + "epoch": 0.7752031627498353, + "grad_norm": 2.000511407852173, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6950838565826416, + "num_tokens": 181265289.0, + "step": 7059 + }, + { + "epoch": 0.7753129804524489, + "grad_norm": 2.5690033435821533, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7124095559120178, + "num_tokens": 181285313.0, + "step": 7060 + }, + { + "epoch": 0.7754227981550625, + "grad_norm": 2.7864553928375244, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7285856008529663, + "num_tokens": 181303220.0, + "step": 7061 + }, + { + "epoch": 0.7755326158576763, + "grad_norm": 2.3203022480010986, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.721051037311554, + "num_tokens": 181325402.0, + "step": 7062 + }, + { + "epoch": 0.77564243356029, + "grad_norm": 2.4972243309020996, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.730506420135498, + "num_tokens": 181344020.0, + "step": 7063 + }, + { + "epoch": 0.7757522512629036, + "grad_norm": 2.0295681953430176, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7087036967277527, + "num_tokens": 181372013.0, + "step": 7064 + }, + { + "epoch": 0.7758620689655172, + "grad_norm": 2.1944684982299805, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7040345668792725, + "num_tokens": 181396118.0, + "step": 7065 + }, + { + "epoch": 0.7759718866681309, + "grad_norm": 2.151498556137085, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7003715634346008, + "num_tokens": 181420405.0, + "step": 7066 + }, + { + "epoch": 0.7760817043707445, + "grad_norm": 2.3454995155334473, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7336851358413696, + "num_tokens": 181442415.0, + "step": 7067 + }, + { + "epoch": 0.7761915220733582, + "grad_norm": 1.9629504680633545, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6999914050102234, + "num_tokens": 181472231.0, + "step": 7068 + }, + { + "epoch": 0.7763013397759719, + "grad_norm": 2.197117328643799, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7081066370010376, + "num_tokens": 181496797.0, + "step": 7069 + }, + { + "epoch": 0.7764111574785856, + "grad_norm": 2.4214730262756348, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7207208871841431, + "num_tokens": 181517024.0, + "step": 7070 + }, + { + "epoch": 0.7765209751811992, + "grad_norm": 2.1954758167266846, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7120587229728699, + "num_tokens": 181541846.0, + "step": 7071 + }, + { + "epoch": 0.7766307928838129, + "grad_norm": 2.575556755065918, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7079514861106873, + "num_tokens": 181562459.0, + "step": 7072 + }, + { + "epoch": 0.7767406105864265, + "grad_norm": 2.375054121017456, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6903367042541504, + "num_tokens": 181585826.0, + "step": 7073 + }, + { + "epoch": 0.7768504282890402, + "grad_norm": 2.1662349700927734, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7196468710899353, + "num_tokens": 181610137.0, + "step": 7074 + }, + { + "epoch": 0.7769602459916538, + "grad_norm": 2.3509607315063477, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7166508436203003, + "num_tokens": 181632672.0, + "step": 7075 + }, + { + "epoch": 0.7770700636942676, + "grad_norm": 2.0127668380737305, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.72304368019104, + "num_tokens": 181660639.0, + "step": 7076 + }, + { + "epoch": 0.7771798813968812, + "grad_norm": 2.099269151687622, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6955938339233398, + "num_tokens": 181686228.0, + "step": 7077 + }, + { + "epoch": 0.7772896990994949, + "grad_norm": 1.9447317123413086, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6889916658401489, + "num_tokens": 181718780.0, + "step": 7078 + }, + { + "epoch": 0.7773995168021085, + "grad_norm": 2.234633445739746, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7241836190223694, + "num_tokens": 181741264.0, + "step": 7079 + }, + { + "epoch": 0.7775093345047221, + "grad_norm": 2.0001943111419678, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6922035813331604, + "num_tokens": 181771468.0, + "step": 7080 + }, + { + "epoch": 0.7776191522073358, + "grad_norm": 2.052663803100586, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7103860378265381, + "num_tokens": 181798837.0, + "step": 7081 + }, + { + "epoch": 0.7777289699099494, + "grad_norm": 2.441572666168213, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7033857107162476, + "num_tokens": 181819932.0, + "step": 7082 + }, + { + "epoch": 0.7778387876125632, + "grad_norm": 1.8866355419158936, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7087937593460083, + "num_tokens": 181851346.0, + "step": 7083 + }, + { + "epoch": 0.7779486053151768, + "grad_norm": 2.155780076980591, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6929336786270142, + "num_tokens": 181878596.0, + "step": 7084 + }, + { + "epoch": 0.7780584230177905, + "grad_norm": 2.736370801925659, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7194343209266663, + "num_tokens": 181895386.0, + "step": 7085 + }, + { + "epoch": 0.7781682407204041, + "grad_norm": 2.2603871822357178, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7102447152137756, + "num_tokens": 181919327.0, + "step": 7086 + }, + { + "epoch": 0.7782780584230178, + "grad_norm": 2.3072783946990967, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7019660472869873, + "num_tokens": 181943554.0, + "step": 7087 + }, + { + "epoch": 0.7783878761256314, + "grad_norm": 2.0291898250579834, + "learning_rate": 1e-06, + "loss": 1.094, + "mean_token_accuracy": 0.6799911260604858, + "num_tokens": 181974522.0, + "step": 7088 + }, + { + "epoch": 0.7784976938282451, + "grad_norm": 2.4067933559417725, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7108528017997742, + "num_tokens": 181996162.0, + "step": 7089 + }, + { + "epoch": 0.7786075115308587, + "grad_norm": 2.448856830596924, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7313245534896851, + "num_tokens": 182014579.0, + "step": 7090 + }, + { + "epoch": 0.7787173292334725, + "grad_norm": 2.441174268722534, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7008448243141174, + "num_tokens": 182037071.0, + "step": 7091 + }, + { + "epoch": 0.7788271469360861, + "grad_norm": 2.1403121948242188, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6800771355628967, + "num_tokens": 182064661.0, + "step": 7092 + }, + { + "epoch": 0.7789369646386998, + "grad_norm": 1.9268954992294312, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.6823339462280273, + "num_tokens": 182098654.0, + "step": 7093 + }, + { + "epoch": 0.7790467823413134, + "grad_norm": 2.4391732215881348, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.687837541103363, + "num_tokens": 182119716.0, + "step": 7094 + }, + { + "epoch": 0.7791566000439271, + "grad_norm": 2.394669532775879, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7182738184928894, + "num_tokens": 182141446.0, + "step": 7095 + }, + { + "epoch": 0.7792664177465407, + "grad_norm": 2.3489437103271484, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6945919990539551, + "num_tokens": 182166633.0, + "step": 7096 + }, + { + "epoch": 0.7793762354491544, + "grad_norm": 2.271658420562744, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7071282863616943, + "num_tokens": 182188542.0, + "step": 7097 + }, + { + "epoch": 0.7794860531517681, + "grad_norm": 2.278665542602539, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7188997864723206, + "num_tokens": 182212657.0, + "step": 7098 + }, + { + "epoch": 0.7795958708543818, + "grad_norm": 2.039933919906616, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7027699947357178, + "num_tokens": 182243172.0, + "step": 7099 + }, + { + "epoch": 0.7797056885569954, + "grad_norm": 2.140599250793457, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7131596803665161, + "num_tokens": 182270055.0, + "step": 7100 + }, + { + "epoch": 0.779815506259609, + "grad_norm": 2.2018520832061768, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.690811276435852, + "num_tokens": 182296989.0, + "step": 7101 + }, + { + "epoch": 0.7799253239622227, + "grad_norm": 2.1016197204589844, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7133709788322449, + "num_tokens": 182324953.0, + "step": 7102 + }, + { + "epoch": 0.7800351416648363, + "grad_norm": 2.29728627204895, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7117935419082642, + "num_tokens": 182347817.0, + "step": 7103 + }, + { + "epoch": 0.78014495936745, + "grad_norm": 2.1255295276641846, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7023488283157349, + "num_tokens": 182373961.0, + "step": 7104 + }, + { + "epoch": 0.7802547770700637, + "grad_norm": 2.0603842735290527, + "learning_rate": 1e-06, + "loss": 1.0773, + "mean_token_accuracy": 0.6755459308624268, + "num_tokens": 182404275.0, + "step": 7105 + }, + { + "epoch": 0.7803645947726774, + "grad_norm": 1.8917449712753296, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.705204963684082, + "num_tokens": 182435329.0, + "step": 7106 + }, + { + "epoch": 0.780474412475291, + "grad_norm": 2.2159180641174316, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.702768087387085, + "num_tokens": 182460316.0, + "step": 7107 + }, + { + "epoch": 0.7805842301779047, + "grad_norm": 1.9775904417037964, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7134345769882202, + "num_tokens": 182485683.0, + "step": 7108 + }, + { + "epoch": 0.7806940478805183, + "grad_norm": 2.011414051055908, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7158036231994629, + "num_tokens": 182512098.0, + "step": 7109 + }, + { + "epoch": 0.780803865583132, + "grad_norm": 2.2834346294403076, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7144780158996582, + "num_tokens": 182534844.0, + "step": 7110 + }, + { + "epoch": 0.7809136832857456, + "grad_norm": 2.1709790229797363, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6909010410308838, + "num_tokens": 182560450.0, + "step": 7111 + }, + { + "epoch": 0.7810235009883594, + "grad_norm": 1.9575492143630981, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7019357681274414, + "num_tokens": 182590355.0, + "step": 7112 + }, + { + "epoch": 0.781133318690973, + "grad_norm": 2.1497550010681152, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7174577116966248, + "num_tokens": 182616358.0, + "step": 7113 + }, + { + "epoch": 0.7812431363935867, + "grad_norm": 2.2526469230651855, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7192248702049255, + "num_tokens": 182639391.0, + "step": 7114 + }, + { + "epoch": 0.7813529540962003, + "grad_norm": 2.3086447715759277, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7101924419403076, + "num_tokens": 182662759.0, + "step": 7115 + }, + { + "epoch": 0.781462771798814, + "grad_norm": 2.1823341846466064, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7097642421722412, + "num_tokens": 182687417.0, + "step": 7116 + }, + { + "epoch": 0.7815725895014276, + "grad_norm": 2.2231571674346924, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6992389559745789, + "num_tokens": 182714550.0, + "step": 7117 + }, + { + "epoch": 0.7816824072040413, + "grad_norm": 2.4023423194885254, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7181922197341919, + "num_tokens": 182735791.0, + "step": 7118 + }, + { + "epoch": 0.7817922249066549, + "grad_norm": 2.090127468109131, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7059517502784729, + "num_tokens": 182764338.0, + "step": 7119 + }, + { + "epoch": 0.7819020426092687, + "grad_norm": 2.1727230548858643, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7050325870513916, + "num_tokens": 182790533.0, + "step": 7120 + }, + { + "epoch": 0.7820118603118823, + "grad_norm": 2.2157130241394043, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7003215551376343, + "num_tokens": 182815738.0, + "step": 7121 + }, + { + "epoch": 0.782121678014496, + "grad_norm": 2.3580546379089355, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6981744170188904, + "num_tokens": 182839294.0, + "step": 7122 + }, + { + "epoch": 0.7822314957171096, + "grad_norm": 2.0398364067077637, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7012127637863159, + "num_tokens": 182869795.0, + "step": 7123 + }, + { + "epoch": 0.7823413134197232, + "grad_norm": 2.069831132888794, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7027112245559692, + "num_tokens": 182898014.0, + "step": 7124 + }, + { + "epoch": 0.7824511311223369, + "grad_norm": 2.0328376293182373, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6865032315254211, + "num_tokens": 182928035.0, + "step": 7125 + }, + { + "epoch": 0.7825609488249505, + "grad_norm": 1.9808156490325928, + "learning_rate": 1e-06, + "loss": 1.0775, + "mean_token_accuracy": 0.6729825139045715, + "num_tokens": 182956779.0, + "step": 7126 + }, + { + "epoch": 0.7826707665275643, + "grad_norm": 2.2918238639831543, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7302958369255066, + "num_tokens": 182977815.0, + "step": 7127 + }, + { + "epoch": 0.7827805842301779, + "grad_norm": 2.12811279296875, + "learning_rate": 1e-06, + "loss": 1.0813, + "mean_token_accuracy": 0.6737217903137207, + "num_tokens": 183005001.0, + "step": 7128 + }, + { + "epoch": 0.7828904019327916, + "grad_norm": 2.197779893875122, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6958423852920532, + "num_tokens": 183030194.0, + "step": 7129 + }, + { + "epoch": 0.7830002196354052, + "grad_norm": 2.2166740894317627, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.735040009021759, + "num_tokens": 183052016.0, + "step": 7130 + }, + { + "epoch": 0.7831100373380189, + "grad_norm": 2.1365840435028076, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6920524835586548, + "num_tokens": 183079359.0, + "step": 7131 + }, + { + "epoch": 0.7832198550406325, + "grad_norm": 2.1898305416107178, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7064282894134521, + "num_tokens": 183105771.0, + "step": 7132 + }, + { + "epoch": 0.7833296727432462, + "grad_norm": 2.042835235595703, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6984661817550659, + "num_tokens": 183134369.0, + "step": 7133 + }, + { + "epoch": 0.7834394904458599, + "grad_norm": 2.2656631469726562, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7089829444885254, + "num_tokens": 183159998.0, + "step": 7134 + }, + { + "epoch": 0.7835493081484736, + "grad_norm": 2.365039348602295, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7157471179962158, + "num_tokens": 183180993.0, + "step": 7135 + }, + { + "epoch": 0.7836591258510872, + "grad_norm": 2.3465285301208496, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6871782541275024, + "num_tokens": 183202367.0, + "step": 7136 + }, + { + "epoch": 0.7837689435537009, + "grad_norm": 2.0362985134124756, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6883801221847534, + "num_tokens": 183232524.0, + "step": 7137 + }, + { + "epoch": 0.7838787612563145, + "grad_norm": 2.588500499725342, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7078986167907715, + "num_tokens": 183255171.0, + "step": 7138 + }, + { + "epoch": 0.7839885789589282, + "grad_norm": 2.3695244789123535, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.734646201133728, + "num_tokens": 183275119.0, + "step": 7139 + }, + { + "epoch": 0.7840983966615418, + "grad_norm": 2.107684373855591, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7302529811859131, + "num_tokens": 183300644.0, + "step": 7140 + }, + { + "epoch": 0.7842082143641556, + "grad_norm": 1.9357093572616577, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6902010440826416, + "num_tokens": 183333139.0, + "step": 7141 + }, + { + "epoch": 0.7843180320667692, + "grad_norm": 2.086397409439087, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7063904404640198, + "num_tokens": 183359209.0, + "step": 7142 + }, + { + "epoch": 0.7844278497693828, + "grad_norm": 2.1642913818359375, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6945548057556152, + "num_tokens": 183386857.0, + "step": 7143 + }, + { + "epoch": 0.7845376674719965, + "grad_norm": 2.178614377975464, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7098766565322876, + "num_tokens": 183411548.0, + "step": 7144 + }, + { + "epoch": 0.7846474851746101, + "grad_norm": 2.0117785930633545, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7051969766616821, + "num_tokens": 183438876.0, + "step": 7145 + }, + { + "epoch": 0.7847573028772238, + "grad_norm": 2.366391658782959, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7033006548881531, + "num_tokens": 183461572.0, + "step": 7146 + }, + { + "epoch": 0.7848671205798374, + "grad_norm": 2.399909496307373, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7110835909843445, + "num_tokens": 183486248.0, + "step": 7147 + }, + { + "epoch": 0.7849769382824512, + "grad_norm": 2.28149151802063, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7095941305160522, + "num_tokens": 183510310.0, + "step": 7148 + }, + { + "epoch": 0.7850867559850648, + "grad_norm": 1.831184983253479, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7035138010978699, + "num_tokens": 183542430.0, + "step": 7149 + }, + { + "epoch": 0.7851965736876785, + "grad_norm": 2.137181043624878, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7017152905464172, + "num_tokens": 183569966.0, + "step": 7150 + }, + { + "epoch": 0.7853063913902921, + "grad_norm": 2.1941208839416504, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.68491530418396, + "num_tokens": 183595690.0, + "step": 7151 + }, + { + "epoch": 0.7854162090929058, + "grad_norm": 2.3447799682617188, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7208800315856934, + "num_tokens": 183617804.0, + "step": 7152 + }, + { + "epoch": 0.7855260267955194, + "grad_norm": 1.9330430030822754, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.699640154838562, + "num_tokens": 183649669.0, + "step": 7153 + }, + { + "epoch": 0.7856358444981331, + "grad_norm": 2.136906385421753, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7039932012557983, + "num_tokens": 183677511.0, + "step": 7154 + }, + { + "epoch": 0.7857456622007467, + "grad_norm": 2.4055793285369873, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.692149817943573, + "num_tokens": 183700097.0, + "step": 7155 + }, + { + "epoch": 0.7858554799033605, + "grad_norm": 2.22523832321167, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7144308090209961, + "num_tokens": 183724990.0, + "step": 7156 + }, + { + "epoch": 0.7859652976059741, + "grad_norm": 2.203587770462036, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.701862096786499, + "num_tokens": 183751286.0, + "step": 7157 + }, + { + "epoch": 0.7860751153085878, + "grad_norm": 2.159592390060425, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7418504953384399, + "num_tokens": 183774142.0, + "step": 7158 + }, + { + "epoch": 0.7861849330112014, + "grad_norm": 2.619448184967041, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7018743753433228, + "num_tokens": 183794078.0, + "step": 7159 + }, + { + "epoch": 0.786294750713815, + "grad_norm": 2.383009672164917, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7226960062980652, + "num_tokens": 183816685.0, + "step": 7160 + }, + { + "epoch": 0.7864045684164287, + "grad_norm": 2.153740167617798, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6973332166671753, + "num_tokens": 183842199.0, + "step": 7161 + }, + { + "epoch": 0.7865143861190423, + "grad_norm": 2.1308646202087402, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.734483003616333, + "num_tokens": 183866165.0, + "step": 7162 + }, + { + "epoch": 0.7866242038216561, + "grad_norm": 2.3031256198883057, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6838864088058472, + "num_tokens": 183891590.0, + "step": 7163 + }, + { + "epoch": 0.7867340215242697, + "grad_norm": 2.088702440261841, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6984646320343018, + "num_tokens": 183921484.0, + "step": 7164 + }, + { + "epoch": 0.7868438392268834, + "grad_norm": 2.062760829925537, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7110884189605713, + "num_tokens": 183949160.0, + "step": 7165 + }, + { + "epoch": 0.786953656929497, + "grad_norm": 2.0403566360473633, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7241719961166382, + "num_tokens": 183974097.0, + "step": 7166 + }, + { + "epoch": 0.7870634746321107, + "grad_norm": 2.446458339691162, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7056067585945129, + "num_tokens": 183997427.0, + "step": 7167 + }, + { + "epoch": 0.7871732923347243, + "grad_norm": 2.376152753829956, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7287865877151489, + "num_tokens": 184018670.0, + "step": 7168 + }, + { + "epoch": 0.787283110037338, + "grad_norm": 1.8770225048065186, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7028132081031799, + "num_tokens": 184050575.0, + "step": 7169 + }, + { + "epoch": 0.7873929277399517, + "grad_norm": 2.1181352138519287, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.708014726638794, + "num_tokens": 184077497.0, + "step": 7170 + }, + { + "epoch": 0.7875027454425654, + "grad_norm": 2.243532419204712, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7126114368438721, + "num_tokens": 184100526.0, + "step": 7171 + }, + { + "epoch": 0.787612563145179, + "grad_norm": 2.3921008110046387, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7110947370529175, + "num_tokens": 184121580.0, + "step": 7172 + }, + { + "epoch": 0.7877223808477927, + "grad_norm": 2.043471336364746, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7031956911087036, + "num_tokens": 184149469.0, + "step": 7173 + }, + { + "epoch": 0.7878321985504063, + "grad_norm": 2.259331226348877, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6979607343673706, + "num_tokens": 184174583.0, + "step": 7174 + }, + { + "epoch": 0.78794201625302, + "grad_norm": 2.594247817993164, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.699937105178833, + "num_tokens": 184195692.0, + "step": 7175 + }, + { + "epoch": 0.7880518339556336, + "grad_norm": 2.4058871269226074, + "learning_rate": 1e-06, + "loss": 1.0867, + "mean_token_accuracy": 0.670862078666687, + "num_tokens": 184219781.0, + "step": 7176 + }, + { + "epoch": 0.7881616516582474, + "grad_norm": 2.2504589557647705, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.7004817724227905, + "num_tokens": 184243763.0, + "step": 7177 + }, + { + "epoch": 0.788271469360861, + "grad_norm": 2.166551113128662, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7127617597579956, + "num_tokens": 184269467.0, + "step": 7178 + }, + { + "epoch": 0.7883812870634747, + "grad_norm": 2.24350905418396, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6961319446563721, + "num_tokens": 184296635.0, + "step": 7179 + }, + { + "epoch": 0.7884911047660883, + "grad_norm": 2.371992826461792, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7096410989761353, + "num_tokens": 184319517.0, + "step": 7180 + }, + { + "epoch": 0.788600922468702, + "grad_norm": 2.775667667388916, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7288601398468018, + "num_tokens": 184344163.0, + "step": 7181 + }, + { + "epoch": 0.7887107401713156, + "grad_norm": 2.2224769592285156, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7125982046127319, + "num_tokens": 184368512.0, + "step": 7182 + }, + { + "epoch": 0.7888205578739292, + "grad_norm": 2.3221702575683594, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7021580934524536, + "num_tokens": 184391096.0, + "step": 7183 + }, + { + "epoch": 0.7889303755765429, + "grad_norm": 2.4758384227752686, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7140902280807495, + "num_tokens": 184413447.0, + "step": 7184 + }, + { + "epoch": 0.7890401932791566, + "grad_norm": 2.32419753074646, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.6950818300247192, + "num_tokens": 184437946.0, + "step": 7185 + }, + { + "epoch": 0.7891500109817703, + "grad_norm": 1.8956060409545898, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7009680271148682, + "num_tokens": 184469195.0, + "step": 7186 + }, + { + "epoch": 0.7892598286843839, + "grad_norm": 2.0515973567962646, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7119762897491455, + "num_tokens": 184496433.0, + "step": 7187 + }, + { + "epoch": 0.7893696463869976, + "grad_norm": 2.4769248962402344, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7143168449401855, + "num_tokens": 184517022.0, + "step": 7188 + }, + { + "epoch": 0.7894794640896112, + "grad_norm": 2.3221583366394043, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7155513763427734, + "num_tokens": 184540390.0, + "step": 7189 + }, + { + "epoch": 0.7895892817922249, + "grad_norm": 2.2233312129974365, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6941176056861877, + "num_tokens": 184565612.0, + "step": 7190 + }, + { + "epoch": 0.7896990994948385, + "grad_norm": 2.3776962757110596, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.706524133682251, + "num_tokens": 184588051.0, + "step": 7191 + }, + { + "epoch": 0.7898089171974523, + "grad_norm": 1.9793105125427246, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7083138227462769, + "num_tokens": 184618824.0, + "step": 7192 + }, + { + "epoch": 0.7899187349000659, + "grad_norm": 1.986221194267273, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7047348022460938, + "num_tokens": 184647534.0, + "step": 7193 + }, + { + "epoch": 0.7900285526026796, + "grad_norm": 2.1703526973724365, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.726351261138916, + "num_tokens": 184671464.0, + "step": 7194 + }, + { + "epoch": 0.7901383703052932, + "grad_norm": 1.965908169746399, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7008489370346069, + "num_tokens": 184703081.0, + "step": 7195 + }, + { + "epoch": 0.7902481880079069, + "grad_norm": 2.419854164123535, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7246443629264832, + "num_tokens": 184724555.0, + "step": 7196 + }, + { + "epoch": 0.7903580057105205, + "grad_norm": 1.9511544704437256, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7113506197929382, + "num_tokens": 184755499.0, + "step": 7197 + }, + { + "epoch": 0.7904678234131342, + "grad_norm": 2.2819178104400635, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7177609801292419, + "num_tokens": 184777923.0, + "step": 7198 + }, + { + "epoch": 0.7905776411157479, + "grad_norm": 2.1572909355163574, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7132796049118042, + "num_tokens": 184802444.0, + "step": 7199 + }, + { + "epoch": 0.7906874588183616, + "grad_norm": 2.2247204780578613, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7252084016799927, + "num_tokens": 184826373.0, + "step": 7200 + }, + { + "epoch": 0.7907972765209752, + "grad_norm": 2.332397222518921, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7281480431556702, + "num_tokens": 184849409.0, + "step": 7201 + }, + { + "epoch": 0.7909070942235888, + "grad_norm": 2.0761284828186035, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7128461599349976, + "num_tokens": 184874939.0, + "step": 7202 + }, + { + "epoch": 0.7910169119262025, + "grad_norm": 2.0600953102111816, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7191794514656067, + "num_tokens": 184899836.0, + "step": 7203 + }, + { + "epoch": 0.7911267296288161, + "grad_norm": 2.2879438400268555, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7111030220985413, + "num_tokens": 184923592.0, + "step": 7204 + }, + { + "epoch": 0.7912365473314298, + "grad_norm": 2.304297685623169, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6904134154319763, + "num_tokens": 184947063.0, + "step": 7205 + }, + { + "epoch": 0.7913463650340435, + "grad_norm": 2.055509090423584, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7228392362594604, + "num_tokens": 184972225.0, + "step": 7206 + }, + { + "epoch": 0.7914561827366572, + "grad_norm": 2.2018890380859375, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6880704164505005, + "num_tokens": 184998503.0, + "step": 7207 + }, + { + "epoch": 0.7915660004392708, + "grad_norm": 2.2565090656280518, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.713695764541626, + "num_tokens": 185023515.0, + "step": 7208 + }, + { + "epoch": 0.7916758181418845, + "grad_norm": 2.2504420280456543, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6922464370727539, + "num_tokens": 185048927.0, + "step": 7209 + }, + { + "epoch": 0.7917856358444981, + "grad_norm": 2.306514263153076, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7259345650672913, + "num_tokens": 185070244.0, + "step": 7210 + }, + { + "epoch": 0.7918954535471118, + "grad_norm": 2.222917079925537, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.721305251121521, + "num_tokens": 185095007.0, + "step": 7211 + }, + { + "epoch": 0.7920052712497254, + "grad_norm": 2.2480201721191406, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7184518575668335, + "num_tokens": 185118328.0, + "step": 7212 + }, + { + "epoch": 0.7921150889523391, + "grad_norm": 2.00350022315979, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.719199538230896, + "num_tokens": 185145466.0, + "step": 7213 + }, + { + "epoch": 0.7922249066549528, + "grad_norm": 2.0686254501342773, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7062579393386841, + "num_tokens": 185171748.0, + "step": 7214 + }, + { + "epoch": 0.7923347243575665, + "grad_norm": 2.423914909362793, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.71375972032547, + "num_tokens": 185192502.0, + "step": 7215 + }, + { + "epoch": 0.7924445420601801, + "grad_norm": 2.3653178215026855, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6898308992385864, + "num_tokens": 185215078.0, + "step": 7216 + }, + { + "epoch": 0.7925543597627938, + "grad_norm": 1.9359304904937744, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7178080081939697, + "num_tokens": 185243513.0, + "step": 7217 + }, + { + "epoch": 0.7926641774654074, + "grad_norm": 2.2072670459747314, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6964320540428162, + "num_tokens": 185267746.0, + "step": 7218 + }, + { + "epoch": 0.792773995168021, + "grad_norm": 1.8861325979232788, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6928232908248901, + "num_tokens": 185302559.0, + "step": 7219 + }, + { + "epoch": 0.7928838128706347, + "grad_norm": 2.0536599159240723, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6938857436180115, + "num_tokens": 185330419.0, + "step": 7220 + }, + { + "epoch": 0.7929936305732485, + "grad_norm": 1.9752315282821655, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.711259126663208, + "num_tokens": 185360016.0, + "step": 7221 + }, + { + "epoch": 0.7931034482758621, + "grad_norm": 2.0203819274902344, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7037997841835022, + "num_tokens": 185390679.0, + "step": 7222 + }, + { + "epoch": 0.7932132659784757, + "grad_norm": 2.5722594261169434, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7172542810440063, + "num_tokens": 185411074.0, + "step": 7223 + }, + { + "epoch": 0.7933230836810894, + "grad_norm": 2.0549473762512207, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6861129999160767, + "num_tokens": 185442263.0, + "step": 7224 + }, + { + "epoch": 0.793432901383703, + "grad_norm": 2.282271385192871, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7166258096694946, + "num_tokens": 185464538.0, + "step": 7225 + }, + { + "epoch": 0.7935427190863167, + "grad_norm": 2.5237526893615723, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7144769430160522, + "num_tokens": 185484206.0, + "step": 7226 + }, + { + "epoch": 0.7936525367889303, + "grad_norm": 2.249706983566284, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7018377780914307, + "num_tokens": 185509350.0, + "step": 7227 + }, + { + "epoch": 0.7937623544915441, + "grad_norm": 2.26920485496521, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7058421969413757, + "num_tokens": 185535191.0, + "step": 7228 + }, + { + "epoch": 0.7938721721941577, + "grad_norm": 2.181746244430542, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7099862694740295, + "num_tokens": 185558798.0, + "step": 7229 + }, + { + "epoch": 0.7939819898967714, + "grad_norm": 2.1827268600463867, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6926267147064209, + "num_tokens": 185584663.0, + "step": 7230 + }, + { + "epoch": 0.794091807599385, + "grad_norm": 2.4652445316314697, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7285904884338379, + "num_tokens": 185605352.0, + "step": 7231 + }, + { + "epoch": 0.7942016253019987, + "grad_norm": 2.025432586669922, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.7059134244918823, + "num_tokens": 185634570.0, + "step": 7232 + }, + { + "epoch": 0.7943114430046123, + "grad_norm": 2.2537014484405518, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7252629995346069, + "num_tokens": 185658732.0, + "step": 7233 + }, + { + "epoch": 0.794421260707226, + "grad_norm": 2.125898838043213, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.6993355751037598, + "num_tokens": 185684484.0, + "step": 7234 + }, + { + "epoch": 0.7945310784098397, + "grad_norm": 2.1679446697235107, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6968046426773071, + "num_tokens": 185710951.0, + "step": 7235 + }, + { + "epoch": 0.7946408961124534, + "grad_norm": 2.164321184158325, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7229979038238525, + "num_tokens": 185739295.0, + "step": 7236 + }, + { + "epoch": 0.794750713815067, + "grad_norm": 2.485531806945801, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7168940305709839, + "num_tokens": 185762504.0, + "step": 7237 + }, + { + "epoch": 0.7948605315176807, + "grad_norm": 1.9545177221298218, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7037161588668823, + "num_tokens": 185792949.0, + "step": 7238 + }, + { + "epoch": 0.7949703492202943, + "grad_norm": 2.0160739421844482, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7184840440750122, + "num_tokens": 185822606.0, + "step": 7239 + }, + { + "epoch": 0.795080166922908, + "grad_norm": 2.2135820388793945, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7094398736953735, + "num_tokens": 185850088.0, + "step": 7240 + }, + { + "epoch": 0.7951899846255216, + "grad_norm": 2.617645025253296, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7355877161026001, + "num_tokens": 185867663.0, + "step": 7241 + }, + { + "epoch": 0.7952998023281352, + "grad_norm": 2.2605860233306885, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7224811315536499, + "num_tokens": 185889529.0, + "step": 7242 + }, + { + "epoch": 0.795409620030749, + "grad_norm": 2.2817509174346924, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7000442743301392, + "num_tokens": 185913027.0, + "step": 7243 + }, + { + "epoch": 0.7955194377333626, + "grad_norm": 2.3946564197540283, + "learning_rate": 1e-06, + "loss": 1.0572, + "mean_token_accuracy": 0.674708366394043, + "num_tokens": 185935864.0, + "step": 7244 + }, + { + "epoch": 0.7956292554359763, + "grad_norm": 2.210198163986206, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7158248424530029, + "num_tokens": 185961462.0, + "step": 7245 + }, + { + "epoch": 0.7957390731385899, + "grad_norm": 2.1439733505249023, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7119128704071045, + "num_tokens": 185986280.0, + "step": 7246 + }, + { + "epoch": 0.7958488908412036, + "grad_norm": 2.012803316116333, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6987085342407227, + "num_tokens": 186015033.0, + "step": 7247 + }, + { + "epoch": 0.7959587085438172, + "grad_norm": 2.1344199180603027, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7273051738739014, + "num_tokens": 186039365.0, + "step": 7248 + }, + { + "epoch": 0.7960685262464309, + "grad_norm": 2.2833151817321777, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7131514549255371, + "num_tokens": 186062236.0, + "step": 7249 + }, + { + "epoch": 0.7961783439490446, + "grad_norm": 2.0760650634765625, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7006853818893433, + "num_tokens": 186089140.0, + "step": 7250 + }, + { + "epoch": 0.7962881616516583, + "grad_norm": 2.0049853324890137, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6962600946426392, + "num_tokens": 186119556.0, + "step": 7251 + }, + { + "epoch": 0.7963979793542719, + "grad_norm": 2.146467924118042, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6992305517196655, + "num_tokens": 186146735.0, + "step": 7252 + }, + { + "epoch": 0.7965077970568856, + "grad_norm": 1.9561389684677124, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7019784450531006, + "num_tokens": 186176698.0, + "step": 7253 + }, + { + "epoch": 0.7966176147594992, + "grad_norm": 2.231394052505493, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6850401759147644, + "num_tokens": 186201697.0, + "step": 7254 + }, + { + "epoch": 0.7967274324621129, + "grad_norm": 2.58341383934021, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7076908946037292, + "num_tokens": 186220939.0, + "step": 7255 + }, + { + "epoch": 0.7968372501647265, + "grad_norm": 2.086371421813965, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7019863128662109, + "num_tokens": 186248408.0, + "step": 7256 + }, + { + "epoch": 0.7969470678673403, + "grad_norm": 2.14473032951355, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7402899861335754, + "num_tokens": 186272004.0, + "step": 7257 + }, + { + "epoch": 0.7970568855699539, + "grad_norm": 1.8047219514846802, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6924466490745544, + "num_tokens": 186307592.0, + "step": 7258 + }, + { + "epoch": 0.7971667032725676, + "grad_norm": 2.0410525798797607, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6956788301467896, + "num_tokens": 186336072.0, + "step": 7259 + }, + { + "epoch": 0.7972765209751812, + "grad_norm": 2.1743452548980713, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7072904706001282, + "num_tokens": 186361876.0, + "step": 7260 + }, + { + "epoch": 0.7973863386777948, + "grad_norm": 2.198903799057007, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7307925224304199, + "num_tokens": 186387123.0, + "step": 7261 + }, + { + "epoch": 0.7974961563804085, + "grad_norm": 2.17708683013916, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6976264715194702, + "num_tokens": 186413246.0, + "step": 7262 + }, + { + "epoch": 0.7976059740830221, + "grad_norm": 2.1748671531677246, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6780783534049988, + "num_tokens": 186440486.0, + "step": 7263 + }, + { + "epoch": 0.7977157917856359, + "grad_norm": 1.9505928754806519, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7094809412956238, + "num_tokens": 186470889.0, + "step": 7264 + }, + { + "epoch": 0.7978256094882495, + "grad_norm": 1.9539992809295654, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7143106460571289, + "num_tokens": 186500229.0, + "step": 7265 + }, + { + "epoch": 0.7979354271908632, + "grad_norm": 2.3321640491485596, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.73198401927948, + "num_tokens": 186521621.0, + "step": 7266 + }, + { + "epoch": 0.7980452448934768, + "grad_norm": 2.1247801780700684, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.697678804397583, + "num_tokens": 186548284.0, + "step": 7267 + }, + { + "epoch": 0.7981550625960905, + "grad_norm": 2.22440242767334, + "learning_rate": 1e-06, + "loss": 1.1046, + "mean_token_accuracy": 0.6644991636276245, + "num_tokens": 186575231.0, + "step": 7268 + }, + { + "epoch": 0.7982648802987041, + "grad_norm": 2.627493381500244, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7158089876174927, + "num_tokens": 186593947.0, + "step": 7269 + }, + { + "epoch": 0.7983746980013178, + "grad_norm": 2.112881898880005, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6974471807479858, + "num_tokens": 186619868.0, + "step": 7270 + }, + { + "epoch": 0.7984845157039314, + "grad_norm": 2.5385615825653076, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7241933941841125, + "num_tokens": 186637930.0, + "step": 7271 + }, + { + "epoch": 0.7985943334065452, + "grad_norm": 2.5033397674560547, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7364833354949951, + "num_tokens": 186657209.0, + "step": 7272 + }, + { + "epoch": 0.7987041511091588, + "grad_norm": 2.1217761039733887, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7188819050788879, + "num_tokens": 186684392.0, + "step": 7273 + }, + { + "epoch": 0.7988139688117725, + "grad_norm": 2.315654754638672, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7273563742637634, + "num_tokens": 186705734.0, + "step": 7274 + }, + { + "epoch": 0.7989237865143861, + "grad_norm": 2.4079532623291016, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7279213070869446, + "num_tokens": 186726216.0, + "step": 7275 + }, + { + "epoch": 0.7990336042169998, + "grad_norm": 2.0765562057495117, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6839233636856079, + "num_tokens": 186755685.0, + "step": 7276 + }, + { + "epoch": 0.7991434219196134, + "grad_norm": 2.0374605655670166, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7122926712036133, + "num_tokens": 186782000.0, + "step": 7277 + }, + { + "epoch": 0.799253239622227, + "grad_norm": 2.238445520401001, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7241134643554688, + "num_tokens": 186805716.0, + "step": 7278 + }, + { + "epoch": 0.7993630573248408, + "grad_norm": 2.181361198425293, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6926087141036987, + "num_tokens": 186832331.0, + "step": 7279 + }, + { + "epoch": 0.7994728750274545, + "grad_norm": 2.2521543502807617, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7015252113342285, + "num_tokens": 186855013.0, + "step": 7280 + }, + { + "epoch": 0.7995826927300681, + "grad_norm": 2.2389135360717773, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.724926233291626, + "num_tokens": 186877042.0, + "step": 7281 + }, + { + "epoch": 0.7996925104326817, + "grad_norm": 2.0760321617126465, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7311760187149048, + "num_tokens": 186903746.0, + "step": 7282 + }, + { + "epoch": 0.7998023281352954, + "grad_norm": 2.0860300064086914, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7186011075973511, + "num_tokens": 186931406.0, + "step": 7283 + }, + { + "epoch": 0.799912145837909, + "grad_norm": 2.1307902336120605, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7062183618545532, + "num_tokens": 186958058.0, + "step": 7284 + }, + { + "epoch": 0.8000219635405227, + "grad_norm": 1.9949378967285156, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.695808470249176, + "num_tokens": 186987477.0, + "step": 7285 + }, + { + "epoch": 0.8001317812431364, + "grad_norm": 2.2721526622772217, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6984661817550659, + "num_tokens": 187011321.0, + "step": 7286 + }, + { + "epoch": 0.8002415989457501, + "grad_norm": 2.1680245399475098, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6998865604400635, + "num_tokens": 187036346.0, + "step": 7287 + }, + { + "epoch": 0.8003514166483637, + "grad_norm": 2.525167226791382, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7109559178352356, + "num_tokens": 187055568.0, + "step": 7288 + }, + { + "epoch": 0.8004612343509774, + "grad_norm": 2.1570065021514893, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7299758791923523, + "num_tokens": 187079946.0, + "step": 7289 + }, + { + "epoch": 0.800571052053591, + "grad_norm": 2.1923301219940186, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6997846364974976, + "num_tokens": 187104462.0, + "step": 7290 + }, + { + "epoch": 0.8006808697562047, + "grad_norm": 2.1167960166931152, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.716099739074707, + "num_tokens": 187131479.0, + "step": 7291 + }, + { + "epoch": 0.8007906874588183, + "grad_norm": 1.9154777526855469, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6915696263313293, + "num_tokens": 187163734.0, + "step": 7292 + }, + { + "epoch": 0.8009005051614321, + "grad_norm": 2.2812373638153076, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7165708541870117, + "num_tokens": 187186684.0, + "step": 7293 + }, + { + "epoch": 0.8010103228640457, + "grad_norm": 1.958120584487915, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7043125629425049, + "num_tokens": 187219320.0, + "step": 7294 + }, + { + "epoch": 0.8011201405666594, + "grad_norm": 1.8648773431777954, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6985153555870056, + "num_tokens": 187253170.0, + "step": 7295 + }, + { + "epoch": 0.801229958269273, + "grad_norm": 2.033662796020508, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7368971109390259, + "num_tokens": 187280078.0, + "step": 7296 + }, + { + "epoch": 0.8013397759718867, + "grad_norm": 2.0956482887268066, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6850250959396362, + "num_tokens": 187308956.0, + "step": 7297 + }, + { + "epoch": 0.8014495936745003, + "grad_norm": 2.4027538299560547, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7206987142562866, + "num_tokens": 187330254.0, + "step": 7298 + }, + { + "epoch": 0.801559411377114, + "grad_norm": 2.0935819149017334, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7046313285827637, + "num_tokens": 187355730.0, + "step": 7299 + }, + { + "epoch": 0.8016692290797276, + "grad_norm": 1.9290589094161987, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6880533695220947, + "num_tokens": 187387802.0, + "step": 7300 + }, + { + "epoch": 0.8017790467823414, + "grad_norm": 2.367328643798828, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6983488202095032, + "num_tokens": 187416773.0, + "step": 7301 + }, + { + "epoch": 0.801888864484955, + "grad_norm": 2.158475875854492, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6805617213249207, + "num_tokens": 187443977.0, + "step": 7302 + }, + { + "epoch": 0.8019986821875686, + "grad_norm": 2.338388442993164, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7051770687103271, + "num_tokens": 187467790.0, + "step": 7303 + }, + { + "epoch": 0.8021084998901823, + "grad_norm": 2.279390573501587, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6812568306922913, + "num_tokens": 187491467.0, + "step": 7304 + }, + { + "epoch": 0.8022183175927959, + "grad_norm": 2.535707473754883, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7310059666633606, + "num_tokens": 187510053.0, + "step": 7305 + }, + { + "epoch": 0.8023281352954096, + "grad_norm": 2.2396111488342285, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6880571246147156, + "num_tokens": 187534105.0, + "step": 7306 + }, + { + "epoch": 0.8024379529980232, + "grad_norm": 2.317564010620117, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7309006452560425, + "num_tokens": 187553682.0, + "step": 7307 + }, + { + "epoch": 0.802547770700637, + "grad_norm": 2.144282817840576, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7115074396133423, + "num_tokens": 187577951.0, + "step": 7308 + }, + { + "epoch": 0.8026575884032506, + "grad_norm": 2.5708253383636475, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7096309661865234, + "num_tokens": 187597435.0, + "step": 7309 + }, + { + "epoch": 0.8027674061058643, + "grad_norm": 2.3986971378326416, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7226994037628174, + "num_tokens": 187617543.0, + "step": 7310 + }, + { + "epoch": 0.8028772238084779, + "grad_norm": 2.297712564468384, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.717902660369873, + "num_tokens": 187640812.0, + "step": 7311 + }, + { + "epoch": 0.8029870415110916, + "grad_norm": 2.0473668575286865, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7161829471588135, + "num_tokens": 187668475.0, + "step": 7312 + }, + { + "epoch": 0.8030968592137052, + "grad_norm": 2.2057034969329834, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7101888656616211, + "num_tokens": 187691992.0, + "step": 7313 + }, + { + "epoch": 0.8032066769163189, + "grad_norm": 2.3790502548217773, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7131471037864685, + "num_tokens": 187713268.0, + "step": 7314 + }, + { + "epoch": 0.8033164946189326, + "grad_norm": 2.0788395404815674, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7347698211669922, + "num_tokens": 187738945.0, + "step": 7315 + }, + { + "epoch": 0.8034263123215463, + "grad_norm": 2.3849008083343506, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7160048484802246, + "num_tokens": 187758688.0, + "step": 7316 + }, + { + "epoch": 0.8035361300241599, + "grad_norm": 2.1228039264678955, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6927374005317688, + "num_tokens": 187784661.0, + "step": 7317 + }, + { + "epoch": 0.8036459477267736, + "grad_norm": 2.186077833175659, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7209863662719727, + "num_tokens": 187808524.0, + "step": 7318 + }, + { + "epoch": 0.8037557654293872, + "grad_norm": 2.0462045669555664, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7172878980636597, + "num_tokens": 187835120.0, + "step": 7319 + }, + { + "epoch": 0.8038655831320008, + "grad_norm": 2.42869234085083, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7004814147949219, + "num_tokens": 187856512.0, + "step": 7320 + }, + { + "epoch": 0.8039754008346145, + "grad_norm": 2.199206829071045, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7313861846923828, + "num_tokens": 187880340.0, + "step": 7321 + }, + { + "epoch": 0.8040852185372283, + "grad_norm": 2.00142240524292, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7048348188400269, + "num_tokens": 187910001.0, + "step": 7322 + }, + { + "epoch": 0.8041950362398419, + "grad_norm": 2.1975197792053223, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7299166917800903, + "num_tokens": 187934177.0, + "step": 7323 + }, + { + "epoch": 0.8043048539424555, + "grad_norm": 2.1970207691192627, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6992936730384827, + "num_tokens": 187959846.0, + "step": 7324 + }, + { + "epoch": 0.8044146716450692, + "grad_norm": 1.934692621231079, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6937941908836365, + "num_tokens": 187994602.0, + "step": 7325 + }, + { + "epoch": 0.8045244893476828, + "grad_norm": 2.038769006729126, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.723486065864563, + "num_tokens": 188022662.0, + "step": 7326 + }, + { + "epoch": 0.8046343070502965, + "grad_norm": 1.986042857170105, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7059409618377686, + "num_tokens": 188050287.0, + "step": 7327 + }, + { + "epoch": 0.8047441247529101, + "grad_norm": 2.0686891078948975, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.7051184177398682, + "num_tokens": 188076730.0, + "step": 7328 + }, + { + "epoch": 0.8048539424555239, + "grad_norm": 2.14308500289917, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6998500823974609, + "num_tokens": 188103410.0, + "step": 7329 + }, + { + "epoch": 0.8049637601581375, + "grad_norm": 2.485060214996338, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7216160297393799, + "num_tokens": 188123490.0, + "step": 7330 + }, + { + "epoch": 0.8050735778607512, + "grad_norm": 2.132021188735962, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7064613103866577, + "num_tokens": 188150708.0, + "step": 7331 + }, + { + "epoch": 0.8051833955633648, + "grad_norm": 2.2122514247894287, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7125141620635986, + "num_tokens": 188176517.0, + "step": 7332 + }, + { + "epoch": 0.8052932132659785, + "grad_norm": 1.921919822692871, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7060109972953796, + "num_tokens": 188207612.0, + "step": 7333 + }, + { + "epoch": 0.8054030309685921, + "grad_norm": 2.1242711544036865, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7025219798088074, + "num_tokens": 188235969.0, + "step": 7334 + }, + { + "epoch": 0.8055128486712058, + "grad_norm": 1.9835617542266846, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6821624636650085, + "num_tokens": 188267124.0, + "step": 7335 + }, + { + "epoch": 0.8056226663738194, + "grad_norm": 1.9513969421386719, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7010945677757263, + "num_tokens": 188295675.0, + "step": 7336 + }, + { + "epoch": 0.8057324840764332, + "grad_norm": 2.4331774711608887, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.734493613243103, + "num_tokens": 188314944.0, + "step": 7337 + }, + { + "epoch": 0.8058423017790468, + "grad_norm": 1.8823113441467285, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7001357078552246, + "num_tokens": 188345454.0, + "step": 7338 + }, + { + "epoch": 0.8059521194816605, + "grad_norm": 2.1748929023742676, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7166663408279419, + "num_tokens": 188369405.0, + "step": 7339 + }, + { + "epoch": 0.8060619371842741, + "grad_norm": 2.224597692489624, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.7021711468696594, + "num_tokens": 188393371.0, + "step": 7340 + }, + { + "epoch": 0.8061717548868877, + "grad_norm": 2.2352190017700195, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7061325311660767, + "num_tokens": 188418573.0, + "step": 7341 + }, + { + "epoch": 0.8062815725895014, + "grad_norm": 2.126704216003418, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7134310007095337, + "num_tokens": 188445064.0, + "step": 7342 + }, + { + "epoch": 0.806391390292115, + "grad_norm": 2.150378942489624, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.709441065788269, + "num_tokens": 188469903.0, + "step": 7343 + }, + { + "epoch": 0.8065012079947288, + "grad_norm": 2.3018651008605957, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7174121141433716, + "num_tokens": 188492653.0, + "step": 7344 + }, + { + "epoch": 0.8066110256973424, + "grad_norm": 2.2358462810516357, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6858961582183838, + "num_tokens": 188519189.0, + "step": 7345 + }, + { + "epoch": 0.8067208433999561, + "grad_norm": 2.0596487522125244, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7164151668548584, + "num_tokens": 188546330.0, + "step": 7346 + }, + { + "epoch": 0.8068306611025697, + "grad_norm": 2.093822479248047, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6949291229248047, + "num_tokens": 188574829.0, + "step": 7347 + }, + { + "epoch": 0.8069404788051834, + "grad_norm": 1.878160834312439, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6891703009605408, + "num_tokens": 188608661.0, + "step": 7348 + }, + { + "epoch": 0.807050296507797, + "grad_norm": 2.283886671066284, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7027562856674194, + "num_tokens": 188632113.0, + "step": 7349 + }, + { + "epoch": 0.8071601142104107, + "grad_norm": 1.9494292736053467, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6998002529144287, + "num_tokens": 188660728.0, + "step": 7350 + }, + { + "epoch": 0.8072699319130244, + "grad_norm": 2.2793259620666504, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6877858638763428, + "num_tokens": 188686084.0, + "step": 7351 + }, + { + "epoch": 0.8073797496156381, + "grad_norm": 2.14571213722229, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6919930577278137, + "num_tokens": 188713231.0, + "step": 7352 + }, + { + "epoch": 0.8074895673182517, + "grad_norm": 2.0987331867218018, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7178448438644409, + "num_tokens": 188739581.0, + "step": 7353 + }, + { + "epoch": 0.8075993850208654, + "grad_norm": 2.218998908996582, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6843709945678711, + "num_tokens": 188764778.0, + "step": 7354 + }, + { + "epoch": 0.807709202723479, + "grad_norm": 2.2483465671539307, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7183213829994202, + "num_tokens": 188787688.0, + "step": 7355 + }, + { + "epoch": 0.8078190204260927, + "grad_norm": 2.0961310863494873, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6919312477111816, + "num_tokens": 188815831.0, + "step": 7356 + }, + { + "epoch": 0.8079288381287063, + "grad_norm": 2.010814905166626, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7290821075439453, + "num_tokens": 188842169.0, + "step": 7357 + }, + { + "epoch": 0.8080386558313201, + "grad_norm": 2.211148738861084, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7061219811439514, + "num_tokens": 188865243.0, + "step": 7358 + }, + { + "epoch": 0.8081484735339337, + "grad_norm": 2.4213762283325195, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7146602869033813, + "num_tokens": 188886858.0, + "step": 7359 + }, + { + "epoch": 0.8082582912365474, + "grad_norm": 2.210798501968384, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7290501594543457, + "num_tokens": 188909697.0, + "step": 7360 + }, + { + "epoch": 0.808368108939161, + "grad_norm": 1.9454118013381958, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7083914279937744, + "num_tokens": 188940669.0, + "step": 7361 + }, + { + "epoch": 0.8084779266417746, + "grad_norm": 2.3138275146484375, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7098096013069153, + "num_tokens": 188962716.0, + "step": 7362 + }, + { + "epoch": 0.8085877443443883, + "grad_norm": 2.197580099105835, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7072518467903137, + "num_tokens": 188987766.0, + "step": 7363 + }, + { + "epoch": 0.8086975620470019, + "grad_norm": 2.294841766357422, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7068641185760498, + "num_tokens": 189011530.0, + "step": 7364 + }, + { + "epoch": 0.8088073797496156, + "grad_norm": 2.2209815979003906, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7206263542175293, + "num_tokens": 189035320.0, + "step": 7365 + }, + { + "epoch": 0.8089171974522293, + "grad_norm": 1.981562614440918, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6905505657196045, + "num_tokens": 189066618.0, + "step": 7366 + }, + { + "epoch": 0.809027015154843, + "grad_norm": 2.380643129348755, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7229741811752319, + "num_tokens": 189088361.0, + "step": 7367 + }, + { + "epoch": 0.8091368328574566, + "grad_norm": 2.4408984184265137, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7040874361991882, + "num_tokens": 189110648.0, + "step": 7368 + }, + { + "epoch": 0.8092466505600703, + "grad_norm": 1.9304198026657104, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7016245126724243, + "num_tokens": 189142137.0, + "step": 7369 + }, + { + "epoch": 0.8093564682626839, + "grad_norm": 2.307375907897949, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7105681896209717, + "num_tokens": 189167204.0, + "step": 7370 + }, + { + "epoch": 0.8094662859652976, + "grad_norm": 2.176593542098999, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6922580003738403, + "num_tokens": 189195322.0, + "step": 7371 + }, + { + "epoch": 0.8095761036679112, + "grad_norm": 2.207225799560547, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7240316867828369, + "num_tokens": 189220651.0, + "step": 7372 + }, + { + "epoch": 0.809685921370525, + "grad_norm": 2.198434591293335, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6944845914840698, + "num_tokens": 189247710.0, + "step": 7373 + }, + { + "epoch": 0.8097957390731386, + "grad_norm": 2.3992583751678467, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7094568014144897, + "num_tokens": 189269894.0, + "step": 7374 + }, + { + "epoch": 0.8099055567757523, + "grad_norm": 2.055708885192871, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6957238912582397, + "num_tokens": 189298574.0, + "step": 7375 + }, + { + "epoch": 0.8100153744783659, + "grad_norm": 2.1433537006378174, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7162685394287109, + "num_tokens": 189324338.0, + "step": 7376 + }, + { + "epoch": 0.8101251921809796, + "grad_norm": 2.0663869380950928, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.7071899175643921, + "num_tokens": 189350947.0, + "step": 7377 + }, + { + "epoch": 0.8102350098835932, + "grad_norm": 2.232236862182617, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7093642354011536, + "num_tokens": 189374633.0, + "step": 7378 + }, + { + "epoch": 0.8103448275862069, + "grad_norm": 2.227403402328491, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6928713321685791, + "num_tokens": 189399317.0, + "step": 7379 + }, + { + "epoch": 0.8104546452888206, + "grad_norm": 2.2430286407470703, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6976399421691895, + "num_tokens": 189422917.0, + "step": 7380 + }, + { + "epoch": 0.8105644629914343, + "grad_norm": 2.1005802154541016, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6937166452407837, + "num_tokens": 189449006.0, + "step": 7381 + }, + { + "epoch": 0.8106742806940479, + "grad_norm": 2.2242913246154785, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6957210898399353, + "num_tokens": 189473321.0, + "step": 7382 + }, + { + "epoch": 0.8107840983966615, + "grad_norm": 2.4948413372039795, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.718207597732544, + "num_tokens": 189493054.0, + "step": 7383 + }, + { + "epoch": 0.8108939160992752, + "grad_norm": 2.082714319229126, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7079404592514038, + "num_tokens": 189520089.0, + "step": 7384 + }, + { + "epoch": 0.8110037338018888, + "grad_norm": 2.2481563091278076, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7452484369277954, + "num_tokens": 189544262.0, + "step": 7385 + }, + { + "epoch": 0.8111135515045025, + "grad_norm": 2.0533812046051025, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7165220975875854, + "num_tokens": 189572179.0, + "step": 7386 + }, + { + "epoch": 0.8112233692071162, + "grad_norm": 2.2280187606811523, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6876537203788757, + "num_tokens": 189598601.0, + "step": 7387 + }, + { + "epoch": 0.8113331869097299, + "grad_norm": 2.086294651031494, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7118436098098755, + "num_tokens": 189625002.0, + "step": 7388 + }, + { + "epoch": 0.8114430046123435, + "grad_norm": 2.0237176418304443, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7071805596351624, + "num_tokens": 189655351.0, + "step": 7389 + }, + { + "epoch": 0.8115528223149572, + "grad_norm": 2.0569403171539307, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7001075744628906, + "num_tokens": 189683793.0, + "step": 7390 + }, + { + "epoch": 0.8116626400175708, + "grad_norm": 2.2994203567504883, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7162090539932251, + "num_tokens": 189705687.0, + "step": 7391 + }, + { + "epoch": 0.8117724577201845, + "grad_norm": 2.01835298538208, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7032641768455505, + "num_tokens": 189734424.0, + "step": 7392 + }, + { + "epoch": 0.8118822754227981, + "grad_norm": 2.1952362060546875, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7142404317855835, + "num_tokens": 189758368.0, + "step": 7393 + }, + { + "epoch": 0.8119920931254118, + "grad_norm": 2.5241689682006836, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7197033166885376, + "num_tokens": 189777489.0, + "step": 7394 + }, + { + "epoch": 0.8121019108280255, + "grad_norm": 2.065567970275879, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.707522988319397, + "num_tokens": 189804790.0, + "step": 7395 + }, + { + "epoch": 0.8122117285306392, + "grad_norm": 2.2226734161376953, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7114826440811157, + "num_tokens": 189827316.0, + "step": 7396 + }, + { + "epoch": 0.8123215462332528, + "grad_norm": 2.244661331176758, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7148774266242981, + "num_tokens": 189850832.0, + "step": 7397 + }, + { + "epoch": 0.8124313639358665, + "grad_norm": 2.086714267730713, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7039383053779602, + "num_tokens": 189877608.0, + "step": 7398 + }, + { + "epoch": 0.8125411816384801, + "grad_norm": 2.4239137172698975, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7307893633842468, + "num_tokens": 189898136.0, + "step": 7399 + }, + { + "epoch": 0.8126509993410937, + "grad_norm": 2.4993274211883545, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6878061294555664, + "num_tokens": 189919352.0, + "step": 7400 + }, + { + "epoch": 0.8127608170437074, + "grad_norm": 2.443025827407837, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.726814329624176, + "num_tokens": 189939375.0, + "step": 7401 + }, + { + "epoch": 0.8128706347463212, + "grad_norm": 2.164572238922119, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7153656482696533, + "num_tokens": 189965359.0, + "step": 7402 + }, + { + "epoch": 0.8129804524489348, + "grad_norm": 2.131706714630127, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7130407094955444, + "num_tokens": 189992067.0, + "step": 7403 + }, + { + "epoch": 0.8130902701515484, + "grad_norm": 2.248196601867676, + "learning_rate": 1e-06, + "loss": 1.092, + "mean_token_accuracy": 0.6841680407524109, + "num_tokens": 190016992.0, + "step": 7404 + }, + { + "epoch": 0.8132000878541621, + "grad_norm": 1.9506858587265015, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7111544609069824, + "num_tokens": 190047388.0, + "step": 7405 + }, + { + "epoch": 0.8133099055567757, + "grad_norm": 2.034627914428711, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7092825174331665, + "num_tokens": 190074527.0, + "step": 7406 + }, + { + "epoch": 0.8134197232593894, + "grad_norm": 1.9831050634384155, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7026472091674805, + "num_tokens": 190106554.0, + "step": 7407 + }, + { + "epoch": 0.813529540962003, + "grad_norm": 2.288205862045288, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6829192638397217, + "num_tokens": 190131712.0, + "step": 7408 + }, + { + "epoch": 0.8136393586646168, + "grad_norm": 2.19823956489563, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7051745057106018, + "num_tokens": 190157111.0, + "step": 7409 + }, + { + "epoch": 0.8137491763672304, + "grad_norm": 2.1190364360809326, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7229762077331543, + "num_tokens": 190183044.0, + "step": 7410 + }, + { + "epoch": 0.8138589940698441, + "grad_norm": 2.4683403968811035, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7130540609359741, + "num_tokens": 190203460.0, + "step": 7411 + }, + { + "epoch": 0.8139688117724577, + "grad_norm": 2.215623140335083, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.723690390586853, + "num_tokens": 190227737.0, + "step": 7412 + }, + { + "epoch": 0.8140786294750714, + "grad_norm": 2.2710015773773193, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7023065090179443, + "num_tokens": 190253279.0, + "step": 7413 + }, + { + "epoch": 0.814188447177685, + "grad_norm": 2.141195297241211, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7246390581130981, + "num_tokens": 190278729.0, + "step": 7414 + }, + { + "epoch": 0.8142982648802987, + "grad_norm": 2.3531882762908936, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7039051651954651, + "num_tokens": 190301999.0, + "step": 7415 + }, + { + "epoch": 0.8144080825829124, + "grad_norm": 2.224754810333252, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7147612571716309, + "num_tokens": 190325157.0, + "step": 7416 + }, + { + "epoch": 0.8145179002855261, + "grad_norm": 2.476210832595825, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6958907246589661, + "num_tokens": 190346772.0, + "step": 7417 + }, + { + "epoch": 0.8146277179881397, + "grad_norm": 2.5994560718536377, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7065548300743103, + "num_tokens": 190367147.0, + "step": 7418 + }, + { + "epoch": 0.8147375356907534, + "grad_norm": 2.1669840812683105, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.739972710609436, + "num_tokens": 190390602.0, + "step": 7419 + }, + { + "epoch": 0.814847353393367, + "grad_norm": 1.9495824575424194, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6900518536567688, + "num_tokens": 190422541.0, + "step": 7420 + }, + { + "epoch": 0.8149571710959806, + "grad_norm": 2.389313220977783, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7362214922904968, + "num_tokens": 190444305.0, + "step": 7421 + }, + { + "epoch": 0.8150669887985943, + "grad_norm": 2.055177688598633, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6859886050224304, + "num_tokens": 190474076.0, + "step": 7422 + }, + { + "epoch": 0.8151768065012079, + "grad_norm": 2.0422897338867188, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6778838634490967, + "num_tokens": 190503993.0, + "step": 7423 + }, + { + "epoch": 0.8152866242038217, + "grad_norm": 2.4337575435638428, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7265012264251709, + "num_tokens": 190523355.0, + "step": 7424 + }, + { + "epoch": 0.8153964419064353, + "grad_norm": 1.909812569618225, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6887538433074951, + "num_tokens": 190554649.0, + "step": 7425 + }, + { + "epoch": 0.815506259609049, + "grad_norm": 2.5523877143859863, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7300916314125061, + "num_tokens": 190573536.0, + "step": 7426 + }, + { + "epoch": 0.8156160773116626, + "grad_norm": 2.1364824771881104, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7034591436386108, + "num_tokens": 190601028.0, + "step": 7427 + }, + { + "epoch": 0.8157258950142763, + "grad_norm": 2.0110700130462646, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6965124607086182, + "num_tokens": 190633234.0, + "step": 7428 + }, + { + "epoch": 0.8158357127168899, + "grad_norm": 2.573634386062622, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7193338871002197, + "num_tokens": 190652050.0, + "step": 7429 + }, + { + "epoch": 0.8159455304195036, + "grad_norm": 2.195627212524414, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6929311752319336, + "num_tokens": 190676992.0, + "step": 7430 + }, + { + "epoch": 0.8160553481221173, + "grad_norm": 2.104433298110962, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7128300666809082, + "num_tokens": 190703786.0, + "step": 7431 + }, + { + "epoch": 0.816165165824731, + "grad_norm": 2.1652467250823975, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7097675204277039, + "num_tokens": 190729112.0, + "step": 7432 + }, + { + "epoch": 0.8162749835273446, + "grad_norm": 2.395805835723877, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7145015001296997, + "num_tokens": 190750877.0, + "step": 7433 + }, + { + "epoch": 0.8163848012299583, + "grad_norm": 2.0807878971099854, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6838723421096802, + "num_tokens": 190778509.0, + "step": 7434 + }, + { + "epoch": 0.8164946189325719, + "grad_norm": 2.0451838970184326, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7125916481018066, + "num_tokens": 190808877.0, + "step": 7435 + }, + { + "epoch": 0.8166044366351856, + "grad_norm": 1.9572898149490356, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6963546276092529, + "num_tokens": 190841261.0, + "step": 7436 + }, + { + "epoch": 0.8167142543377992, + "grad_norm": 2.1079888343811035, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7067161202430725, + "num_tokens": 190865516.0, + "step": 7437 + }, + { + "epoch": 0.816824072040413, + "grad_norm": 1.995668649673462, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6928691864013672, + "num_tokens": 190895462.0, + "step": 7438 + }, + { + "epoch": 0.8169338897430266, + "grad_norm": 2.382282018661499, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7235383987426758, + "num_tokens": 190916192.0, + "step": 7439 + }, + { + "epoch": 0.8170437074456403, + "grad_norm": 2.2048444747924805, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.710543692111969, + "num_tokens": 190942090.0, + "step": 7440 + }, + { + "epoch": 0.8171535251482539, + "grad_norm": 1.9412779808044434, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6970024108886719, + "num_tokens": 190973688.0, + "step": 7441 + }, + { + "epoch": 0.8172633428508675, + "grad_norm": 2.1143667697906494, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7274006605148315, + "num_tokens": 190999595.0, + "step": 7442 + }, + { + "epoch": 0.8173731605534812, + "grad_norm": 2.3676562309265137, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7066226601600647, + "num_tokens": 191021132.0, + "step": 7443 + }, + { + "epoch": 0.8174829782560948, + "grad_norm": 2.5076003074645996, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7121363878250122, + "num_tokens": 191042078.0, + "step": 7444 + }, + { + "epoch": 0.8175927959587086, + "grad_norm": 2.0590615272521973, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6988669633865356, + "num_tokens": 191071228.0, + "step": 7445 + }, + { + "epoch": 0.8177026136613222, + "grad_norm": 2.268933057785034, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7094323039054871, + "num_tokens": 191095345.0, + "step": 7446 + }, + { + "epoch": 0.8178124313639359, + "grad_norm": 2.145672559738159, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7325196266174316, + "num_tokens": 191121498.0, + "step": 7447 + }, + { + "epoch": 0.8179222490665495, + "grad_norm": 2.221008777618408, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.6811730861663818, + "num_tokens": 191146547.0, + "step": 7448 + }, + { + "epoch": 0.8180320667691632, + "grad_norm": 2.0989255905151367, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7076353430747986, + "num_tokens": 191175935.0, + "step": 7449 + }, + { + "epoch": 0.8181418844717768, + "grad_norm": 2.3945467472076416, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6910528540611267, + "num_tokens": 191199794.0, + "step": 7450 + }, + { + "epoch": 0.8182517021743905, + "grad_norm": 2.1866397857666016, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7266086339950562, + "num_tokens": 191224155.0, + "step": 7451 + }, + { + "epoch": 0.8183615198770041, + "grad_norm": 1.912137746810913, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6959041357040405, + "num_tokens": 191255545.0, + "step": 7452 + }, + { + "epoch": 0.8184713375796179, + "grad_norm": 2.250192165374756, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.697394609451294, + "num_tokens": 191282188.0, + "step": 7453 + }, + { + "epoch": 0.8185811552822315, + "grad_norm": 2.3485615253448486, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7345538139343262, + "num_tokens": 191305775.0, + "step": 7454 + }, + { + "epoch": 0.8186909729848452, + "grad_norm": 2.3447694778442383, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7016571164131165, + "num_tokens": 191327822.0, + "step": 7455 + }, + { + "epoch": 0.8188007906874588, + "grad_norm": 2.021510601043701, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7072010636329651, + "num_tokens": 191355508.0, + "step": 7456 + }, + { + "epoch": 0.8189106083900725, + "grad_norm": 2.0092029571533203, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6854271292686462, + "num_tokens": 191386097.0, + "step": 7457 + }, + { + "epoch": 0.8190204260926861, + "grad_norm": 2.2029361724853516, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7034040689468384, + "num_tokens": 191413556.0, + "step": 7458 + }, + { + "epoch": 0.8191302437952998, + "grad_norm": 2.007981538772583, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7140949964523315, + "num_tokens": 191441544.0, + "step": 7459 + }, + { + "epoch": 0.8192400614979135, + "grad_norm": 2.259589195251465, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7220037579536438, + "num_tokens": 191462880.0, + "step": 7460 + }, + { + "epoch": 0.8193498792005272, + "grad_norm": 2.1925089359283447, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6898547410964966, + "num_tokens": 191487627.0, + "step": 7461 + }, + { + "epoch": 0.8194596969031408, + "grad_norm": 2.2567336559295654, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6881522536277771, + "num_tokens": 191512827.0, + "step": 7462 + }, + { + "epoch": 0.8195695146057544, + "grad_norm": 1.97969388961792, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7002342939376831, + "num_tokens": 191541874.0, + "step": 7463 + }, + { + "epoch": 0.8196793323083681, + "grad_norm": 2.2512059211730957, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6751030087471008, + "num_tokens": 191568015.0, + "step": 7464 + }, + { + "epoch": 0.8197891500109817, + "grad_norm": 2.2001922130584717, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7126285433769226, + "num_tokens": 191590802.0, + "step": 7465 + }, + { + "epoch": 0.8198989677135954, + "grad_norm": 2.216496467590332, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7075302004814148, + "num_tokens": 191614046.0, + "step": 7466 + }, + { + "epoch": 0.8200087854162091, + "grad_norm": 2.39546275138855, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7103064656257629, + "num_tokens": 191636363.0, + "step": 7467 + }, + { + "epoch": 0.8201186031188228, + "grad_norm": 2.226358413696289, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7168818712234497, + "num_tokens": 191660502.0, + "step": 7468 + }, + { + "epoch": 0.8202284208214364, + "grad_norm": 2.05715012550354, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6949803829193115, + "num_tokens": 191689684.0, + "step": 7469 + }, + { + "epoch": 0.8203382385240501, + "grad_norm": 2.0810258388519287, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.6971349120140076, + "num_tokens": 191718171.0, + "step": 7470 + }, + { + "epoch": 0.8204480562266637, + "grad_norm": 2.3720715045928955, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7126927375793457, + "num_tokens": 191741271.0, + "step": 7471 + }, + { + "epoch": 0.8205578739292774, + "grad_norm": 2.144836664199829, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7207773923873901, + "num_tokens": 191766781.0, + "step": 7472 + }, + { + "epoch": 0.820667691631891, + "grad_norm": 2.255185604095459, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7195128202438354, + "num_tokens": 191789921.0, + "step": 7473 + }, + { + "epoch": 0.8207775093345048, + "grad_norm": 2.3464882373809814, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7064913511276245, + "num_tokens": 191812137.0, + "step": 7474 + }, + { + "epoch": 0.8208873270371184, + "grad_norm": 2.477689266204834, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7340071797370911, + "num_tokens": 191832795.0, + "step": 7475 + }, + { + "epoch": 0.8209971447397321, + "grad_norm": 2.155501365661621, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7302656173706055, + "num_tokens": 191857779.0, + "step": 7476 + }, + { + "epoch": 0.8211069624423457, + "grad_norm": 2.248997688293457, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7031179666519165, + "num_tokens": 191882636.0, + "step": 7477 + }, + { + "epoch": 0.8212167801449594, + "grad_norm": 2.276278018951416, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7007812261581421, + "num_tokens": 191906450.0, + "step": 7478 + }, + { + "epoch": 0.821326597847573, + "grad_norm": 2.08849835395813, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7030566334724426, + "num_tokens": 191934434.0, + "step": 7479 + }, + { + "epoch": 0.8214364155501866, + "grad_norm": 2.1225249767303467, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6835595965385437, + "num_tokens": 191964153.0, + "step": 7480 + }, + { + "epoch": 0.8215462332528004, + "grad_norm": 1.9900519847869873, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6996179819107056, + "num_tokens": 191993515.0, + "step": 7481 + }, + { + "epoch": 0.821656050955414, + "grad_norm": 2.0699715614318848, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7101045846939087, + "num_tokens": 192020925.0, + "step": 7482 + }, + { + "epoch": 0.8217658686580277, + "grad_norm": 2.1487882137298584, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6806025505065918, + "num_tokens": 192050535.0, + "step": 7483 + }, + { + "epoch": 0.8218756863606413, + "grad_norm": 2.06595516204834, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7086585760116577, + "num_tokens": 192078124.0, + "step": 7484 + }, + { + "epoch": 0.821985504063255, + "grad_norm": 2.1218485832214355, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.707582950592041, + "num_tokens": 192105023.0, + "step": 7485 + }, + { + "epoch": 0.8220953217658686, + "grad_norm": 2.110865592956543, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.694230318069458, + "num_tokens": 192130639.0, + "step": 7486 + }, + { + "epoch": 0.8222051394684823, + "grad_norm": 2.492259979248047, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7188421487808228, + "num_tokens": 192151097.0, + "step": 7487 + }, + { + "epoch": 0.8223149571710959, + "grad_norm": 2.2849044799804688, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7343342304229736, + "num_tokens": 192174296.0, + "step": 7488 + }, + { + "epoch": 0.8224247748737097, + "grad_norm": 2.293217897415161, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7003666758537292, + "num_tokens": 192199330.0, + "step": 7489 + }, + { + "epoch": 0.8225345925763233, + "grad_norm": 1.886555552482605, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6915674209594727, + "num_tokens": 192231759.0, + "step": 7490 + }, + { + "epoch": 0.822644410278937, + "grad_norm": 2.058624267578125, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7245322465896606, + "num_tokens": 192260635.0, + "step": 7491 + }, + { + "epoch": 0.8227542279815506, + "grad_norm": 2.255868673324585, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.7003641128540039, + "num_tokens": 192284157.0, + "step": 7492 + }, + { + "epoch": 0.8228640456841643, + "grad_norm": 2.07452130317688, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6950743198394775, + "num_tokens": 192313557.0, + "step": 7493 + }, + { + "epoch": 0.8229738633867779, + "grad_norm": 2.2226340770721436, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7041918635368347, + "num_tokens": 192336780.0, + "step": 7494 + }, + { + "epoch": 0.8230836810893916, + "grad_norm": 2.2759857177734375, + "learning_rate": 1e-06, + "loss": 1.0503, + "mean_token_accuracy": 0.6884710788726807, + "num_tokens": 192362230.0, + "step": 7495 + }, + { + "epoch": 0.8231934987920053, + "grad_norm": 2.1510045528411865, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7003313302993774, + "num_tokens": 192387588.0, + "step": 7496 + }, + { + "epoch": 0.823303316494619, + "grad_norm": 2.293088436126709, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7098573446273804, + "num_tokens": 192409447.0, + "step": 7497 + }, + { + "epoch": 0.8234131341972326, + "grad_norm": 2.1860616207122803, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7134600877761841, + "num_tokens": 192435228.0, + "step": 7498 + }, + { + "epoch": 0.8235229518998463, + "grad_norm": 2.2444286346435547, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.725521445274353, + "num_tokens": 192456301.0, + "step": 7499 + }, + { + "epoch": 0.8236327696024599, + "grad_norm": 2.091273069381714, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7102763652801514, + "num_tokens": 192483122.0, + "step": 7500 + }, + { + "epoch": 0.8237425873050735, + "grad_norm": 2.2461466789245605, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7232635021209717, + "num_tokens": 192505270.0, + "step": 7501 + }, + { + "epoch": 0.8238524050076872, + "grad_norm": 2.358682632446289, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7297006845474243, + "num_tokens": 192525658.0, + "step": 7502 + }, + { + "epoch": 0.823962222710301, + "grad_norm": 2.198883533477783, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7042331695556641, + "num_tokens": 192550572.0, + "step": 7503 + }, + { + "epoch": 0.8240720404129146, + "grad_norm": 2.4652769565582275, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7429614067077637, + "num_tokens": 192569606.0, + "step": 7504 + }, + { + "epoch": 0.8241818581155282, + "grad_norm": 2.2627615928649902, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7025688886642456, + "num_tokens": 192592216.0, + "step": 7505 + }, + { + "epoch": 0.8242916758181419, + "grad_norm": 2.146127939224243, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7270656824111938, + "num_tokens": 192616207.0, + "step": 7506 + }, + { + "epoch": 0.8244014935207555, + "grad_norm": 2.4307355880737305, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7180466055870056, + "num_tokens": 192636937.0, + "step": 7507 + }, + { + "epoch": 0.8245113112233692, + "grad_norm": 2.0938198566436768, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7330164909362793, + "num_tokens": 192663968.0, + "step": 7508 + }, + { + "epoch": 0.8246211289259828, + "grad_norm": 2.2736382484436035, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7071468830108643, + "num_tokens": 192689272.0, + "step": 7509 + }, + { + "epoch": 0.8247309466285966, + "grad_norm": 2.2834408283233643, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7098827958106995, + "num_tokens": 192712994.0, + "step": 7510 + }, + { + "epoch": 0.8248407643312102, + "grad_norm": 2.1771767139434814, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.694230854511261, + "num_tokens": 192740652.0, + "step": 7511 + }, + { + "epoch": 0.8249505820338239, + "grad_norm": 2.259521484375, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6941217184066772, + "num_tokens": 192765247.0, + "step": 7512 + }, + { + "epoch": 0.8250603997364375, + "grad_norm": 2.501502275466919, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7227382063865662, + "num_tokens": 192784580.0, + "step": 7513 + }, + { + "epoch": 0.8251702174390512, + "grad_norm": 2.1431875228881836, + "learning_rate": 1e-06, + "loss": 1.0757, + "mean_token_accuracy": 0.6726506948471069, + "num_tokens": 192812550.0, + "step": 7514 + }, + { + "epoch": 0.8252800351416648, + "grad_norm": 2.429513454437256, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7068368792533875, + "num_tokens": 192834218.0, + "step": 7515 + }, + { + "epoch": 0.8253898528442785, + "grad_norm": 2.157294750213623, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7094007730484009, + "num_tokens": 192863032.0, + "step": 7516 + }, + { + "epoch": 0.8254996705468921, + "grad_norm": 2.0360267162323, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7061911225318909, + "num_tokens": 192890763.0, + "step": 7517 + }, + { + "epoch": 0.8256094882495059, + "grad_norm": 2.172175645828247, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.696177065372467, + "num_tokens": 192915711.0, + "step": 7518 + }, + { + "epoch": 0.8257193059521195, + "grad_norm": 2.1353366374969482, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.693051815032959, + "num_tokens": 192942636.0, + "step": 7519 + }, + { + "epoch": 0.8258291236547332, + "grad_norm": 2.2157628536224365, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6939462423324585, + "num_tokens": 192966572.0, + "step": 7520 + }, + { + "epoch": 0.8259389413573468, + "grad_norm": 2.1611735820770264, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7036559581756592, + "num_tokens": 192992901.0, + "step": 7521 + }, + { + "epoch": 0.8260487590599604, + "grad_norm": 2.3045482635498047, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6937677264213562, + "num_tokens": 193016951.0, + "step": 7522 + }, + { + "epoch": 0.8261585767625741, + "grad_norm": 2.2490243911743164, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7078248262405396, + "num_tokens": 193040173.0, + "step": 7523 + }, + { + "epoch": 0.8262683944651877, + "grad_norm": 2.2080180644989014, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7287946939468384, + "num_tokens": 193063479.0, + "step": 7524 + }, + { + "epoch": 0.8263782121678015, + "grad_norm": 2.3578782081604004, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7281308174133301, + "num_tokens": 193084832.0, + "step": 7525 + }, + { + "epoch": 0.8264880298704151, + "grad_norm": 2.3162453174591064, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7098780870437622, + "num_tokens": 193110579.0, + "step": 7526 + }, + { + "epoch": 0.8265978475730288, + "grad_norm": 2.132020950317383, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7051078677177429, + "num_tokens": 193137784.0, + "step": 7527 + }, + { + "epoch": 0.8267076652756424, + "grad_norm": 2.318089246749878, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7307358384132385, + "num_tokens": 193162811.0, + "step": 7528 + }, + { + "epoch": 0.8268174829782561, + "grad_norm": 1.829047679901123, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6780024766921997, + "num_tokens": 193201701.0, + "step": 7529 + }, + { + "epoch": 0.8269273006808697, + "grad_norm": 2.0996956825256348, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.693954348564148, + "num_tokens": 193228744.0, + "step": 7530 + }, + { + "epoch": 0.8270371183834834, + "grad_norm": 1.9522074460983276, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7064040899276733, + "num_tokens": 193260154.0, + "step": 7531 + }, + { + "epoch": 0.8271469360860971, + "grad_norm": 2.0809638500213623, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7156320810317993, + "num_tokens": 193287060.0, + "step": 7532 + }, + { + "epoch": 0.8272567537887108, + "grad_norm": 2.1630914211273193, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6946245431900024, + "num_tokens": 193313555.0, + "step": 7533 + }, + { + "epoch": 0.8273665714913244, + "grad_norm": 2.174703598022461, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6965272426605225, + "num_tokens": 193338901.0, + "step": 7534 + }, + { + "epoch": 0.8274763891939381, + "grad_norm": 2.5372822284698486, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7213085889816284, + "num_tokens": 193358833.0, + "step": 7535 + }, + { + "epoch": 0.8275862068965517, + "grad_norm": 2.1031105518341064, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.715954065322876, + "num_tokens": 193384034.0, + "step": 7536 + }, + { + "epoch": 0.8276960245991654, + "grad_norm": 2.5362000465393066, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.719338595867157, + "num_tokens": 193405349.0, + "step": 7537 + }, + { + "epoch": 0.827805842301779, + "grad_norm": 2.038372039794922, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7141507863998413, + "num_tokens": 193435995.0, + "step": 7538 + }, + { + "epoch": 0.8279156600043928, + "grad_norm": 2.0361106395721436, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7185467481613159, + "num_tokens": 193462772.0, + "step": 7539 + }, + { + "epoch": 0.8280254777070064, + "grad_norm": 2.058418035507202, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7035157680511475, + "num_tokens": 193489184.0, + "step": 7540 + }, + { + "epoch": 0.82813529540962, + "grad_norm": 2.4303503036499023, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.726005494594574, + "num_tokens": 193509692.0, + "step": 7541 + }, + { + "epoch": 0.8282451131122337, + "grad_norm": 2.095372200012207, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7205653190612793, + "num_tokens": 193534674.0, + "step": 7542 + }, + { + "epoch": 0.8283549308148473, + "grad_norm": 2.1412763595581055, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7170944213867188, + "num_tokens": 193566113.0, + "step": 7543 + }, + { + "epoch": 0.828464748517461, + "grad_norm": 2.2576358318328857, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7048354744911194, + "num_tokens": 193590497.0, + "step": 7544 + }, + { + "epoch": 0.8285745662200746, + "grad_norm": 2.219640016555786, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7100753784179688, + "num_tokens": 193614430.0, + "step": 7545 + }, + { + "epoch": 0.8286843839226883, + "grad_norm": 2.1326093673706055, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6850687265396118, + "num_tokens": 193641710.0, + "step": 7546 + }, + { + "epoch": 0.828794201625302, + "grad_norm": 1.939257264137268, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7211795449256897, + "num_tokens": 193671213.0, + "step": 7547 + }, + { + "epoch": 0.8289040193279157, + "grad_norm": 2.1851439476013184, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6945266127586365, + "num_tokens": 193697239.0, + "step": 7548 + }, + { + "epoch": 0.8290138370305293, + "grad_norm": 1.959938645362854, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.6989074349403381, + "num_tokens": 193727519.0, + "step": 7549 + }, + { + "epoch": 0.829123654733143, + "grad_norm": 2.335411310195923, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7005249857902527, + "num_tokens": 193749733.0, + "step": 7550 + }, + { + "epoch": 0.8292334724357566, + "grad_norm": 2.2770209312438965, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7148796916007996, + "num_tokens": 193773864.0, + "step": 7551 + }, + { + "epoch": 0.8293432901383703, + "grad_norm": 2.256089448928833, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7069882750511169, + "num_tokens": 193798084.0, + "step": 7552 + }, + { + "epoch": 0.8294531078409839, + "grad_norm": 2.004343032836914, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7362295389175415, + "num_tokens": 193825829.0, + "step": 7553 + }, + { + "epoch": 0.8295629255435977, + "grad_norm": 1.9800463914871216, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7220181226730347, + "num_tokens": 193856607.0, + "step": 7554 + }, + { + "epoch": 0.8296727432462113, + "grad_norm": 2.0184414386749268, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.700761079788208, + "num_tokens": 193887060.0, + "step": 7555 + }, + { + "epoch": 0.829782560948825, + "grad_norm": 2.254655361175537, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.745090901851654, + "num_tokens": 193910009.0, + "step": 7556 + }, + { + "epoch": 0.8298923786514386, + "grad_norm": 1.92280912399292, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6958521604537964, + "num_tokens": 193942618.0, + "step": 7557 + }, + { + "epoch": 0.8300021963540523, + "grad_norm": 2.4468369483947754, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7265653610229492, + "num_tokens": 193962070.0, + "step": 7558 + }, + { + "epoch": 0.8301120140566659, + "grad_norm": 2.0810487270355225, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7013312578201294, + "num_tokens": 193989834.0, + "step": 7559 + }, + { + "epoch": 0.8302218317592795, + "grad_norm": 2.075024127960205, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7101583480834961, + "num_tokens": 194019737.0, + "step": 7560 + }, + { + "epoch": 0.8303316494618933, + "grad_norm": 2.355182409286499, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7142506837844849, + "num_tokens": 194042843.0, + "step": 7561 + }, + { + "epoch": 0.830441467164507, + "grad_norm": 2.3682985305786133, + "learning_rate": 1e-06, + "loss": 1.072, + "mean_token_accuracy": 0.6782498359680176, + "num_tokens": 194066709.0, + "step": 7562 + }, + { + "epoch": 0.8305512848671206, + "grad_norm": 2.0974440574645996, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6884779930114746, + "num_tokens": 194095134.0, + "step": 7563 + }, + { + "epoch": 0.8306611025697342, + "grad_norm": 2.224259853363037, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7085961103439331, + "num_tokens": 194117983.0, + "step": 7564 + }, + { + "epoch": 0.8307709202723479, + "grad_norm": 2.152073383331299, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7180113792419434, + "num_tokens": 194143547.0, + "step": 7565 + }, + { + "epoch": 0.8308807379749615, + "grad_norm": 2.2596774101257324, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7090893983840942, + "num_tokens": 194167318.0, + "step": 7566 + }, + { + "epoch": 0.8309905556775752, + "grad_norm": 1.9307669401168823, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6894640922546387, + "num_tokens": 194200226.0, + "step": 7567 + }, + { + "epoch": 0.8311003733801889, + "grad_norm": 2.2554006576538086, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7456721067428589, + "num_tokens": 194223158.0, + "step": 7568 + }, + { + "epoch": 0.8312101910828026, + "grad_norm": 2.1299872398376465, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6900749206542969, + "num_tokens": 194250170.0, + "step": 7569 + }, + { + "epoch": 0.8313200087854162, + "grad_norm": 2.572075605392456, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7148488163948059, + "num_tokens": 194267217.0, + "step": 7570 + }, + { + "epoch": 0.8314298264880299, + "grad_norm": 2.044870615005493, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6903432607650757, + "num_tokens": 194296731.0, + "step": 7571 + }, + { + "epoch": 0.8315396441906435, + "grad_norm": 2.1004486083984375, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6971625089645386, + "num_tokens": 194322929.0, + "step": 7572 + }, + { + "epoch": 0.8316494618932572, + "grad_norm": 2.3319523334503174, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6919460296630859, + "num_tokens": 194347492.0, + "step": 7573 + }, + { + "epoch": 0.8317592795958708, + "grad_norm": 2.221761703491211, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7178186178207397, + "num_tokens": 194372331.0, + "step": 7574 + }, + { + "epoch": 0.8318690972984845, + "grad_norm": 2.1709206104278564, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6837051510810852, + "num_tokens": 194399681.0, + "step": 7575 + }, + { + "epoch": 0.8319789150010982, + "grad_norm": 2.221489191055298, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.716914713382721, + "num_tokens": 194426226.0, + "step": 7576 + }, + { + "epoch": 0.8320887327037119, + "grad_norm": 2.2420198917388916, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6970925331115723, + "num_tokens": 194450336.0, + "step": 7577 + }, + { + "epoch": 0.8321985504063255, + "grad_norm": 2.182004451751709, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7050644159317017, + "num_tokens": 194476692.0, + "step": 7578 + }, + { + "epoch": 0.8323083681089392, + "grad_norm": 2.138275384902954, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7126854658126831, + "num_tokens": 194503084.0, + "step": 7579 + }, + { + "epoch": 0.8324181858115528, + "grad_norm": 1.9872066974639893, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7134093642234802, + "num_tokens": 194530688.0, + "step": 7580 + }, + { + "epoch": 0.8325280035141664, + "grad_norm": 2.3244686126708984, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7092599868774414, + "num_tokens": 194552308.0, + "step": 7581 + }, + { + "epoch": 0.8326378212167801, + "grad_norm": 2.2037858963012695, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7015516757965088, + "num_tokens": 194577847.0, + "step": 7582 + }, + { + "epoch": 0.8327476389193939, + "grad_norm": 2.34041428565979, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7279008030891418, + "num_tokens": 194599728.0, + "step": 7583 + }, + { + "epoch": 0.8328574566220075, + "grad_norm": 1.9846405982971191, + "learning_rate": 1e-06, + "loss": 1.1039, + "mean_token_accuracy": 0.667829155921936, + "num_tokens": 194631209.0, + "step": 7584 + }, + { + "epoch": 0.8329672743246211, + "grad_norm": 2.2995123863220215, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7212172746658325, + "num_tokens": 194654395.0, + "step": 7585 + }, + { + "epoch": 0.8330770920272348, + "grad_norm": 2.2014763355255127, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6878042221069336, + "num_tokens": 194678279.0, + "step": 7586 + }, + { + "epoch": 0.8331869097298484, + "grad_norm": 2.152700662612915, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6898804306983948, + "num_tokens": 194704526.0, + "step": 7587 + }, + { + "epoch": 0.8332967274324621, + "grad_norm": 2.0111470222473145, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6958173513412476, + "num_tokens": 194733005.0, + "step": 7588 + }, + { + "epoch": 0.8334065451350757, + "grad_norm": 2.0351688861846924, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6975314617156982, + "num_tokens": 194761343.0, + "step": 7589 + }, + { + "epoch": 0.8335163628376895, + "grad_norm": 2.0301270484924316, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7047477960586548, + "num_tokens": 194789781.0, + "step": 7590 + }, + { + "epoch": 0.8336261805403031, + "grad_norm": 2.2210440635681152, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6973788142204285, + "num_tokens": 194815594.0, + "step": 7591 + }, + { + "epoch": 0.8337359982429168, + "grad_norm": 2.0616419315338135, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7023647427558899, + "num_tokens": 194843760.0, + "step": 7592 + }, + { + "epoch": 0.8338458159455304, + "grad_norm": 2.231236219406128, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7100750803947449, + "num_tokens": 194869033.0, + "step": 7593 + }, + { + "epoch": 0.8339556336481441, + "grad_norm": 2.170097589492798, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7201952934265137, + "num_tokens": 194892553.0, + "step": 7594 + }, + { + "epoch": 0.8340654513507577, + "grad_norm": 2.2114734649658203, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7100750207901001, + "num_tokens": 194916591.0, + "step": 7595 + }, + { + "epoch": 0.8341752690533714, + "grad_norm": 2.002162456512451, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7105658054351807, + "num_tokens": 194945245.0, + "step": 7596 + }, + { + "epoch": 0.8342850867559851, + "grad_norm": 1.8342951536178589, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7084226608276367, + "num_tokens": 194977144.0, + "step": 7597 + }, + { + "epoch": 0.8343949044585988, + "grad_norm": 2.0282340049743652, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6960923671722412, + "num_tokens": 195005692.0, + "step": 7598 + }, + { + "epoch": 0.8345047221612124, + "grad_norm": 2.522763967514038, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7319532632827759, + "num_tokens": 195025657.0, + "step": 7599 + }, + { + "epoch": 0.834614539863826, + "grad_norm": 2.1342532634735107, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7064732313156128, + "num_tokens": 195052725.0, + "step": 7600 + }, + { + "epoch": 0.8347243575664397, + "grad_norm": 2.2362570762634277, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6884350776672363, + "num_tokens": 195077951.0, + "step": 7601 + }, + { + "epoch": 0.8348341752690533, + "grad_norm": 2.3138296604156494, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7206123471260071, + "num_tokens": 195100796.0, + "step": 7602 + }, + { + "epoch": 0.834943992971667, + "grad_norm": 2.245455741882324, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7258599400520325, + "num_tokens": 195124764.0, + "step": 7603 + }, + { + "epoch": 0.8350538106742806, + "grad_norm": 2.027376651763916, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6977822780609131, + "num_tokens": 195153748.0, + "step": 7604 + }, + { + "epoch": 0.8351636283768944, + "grad_norm": 2.2187583446502686, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7092716693878174, + "num_tokens": 195177779.0, + "step": 7605 + }, + { + "epoch": 0.835273446079508, + "grad_norm": 2.12919545173645, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7051866054534912, + "num_tokens": 195202917.0, + "step": 7606 + }, + { + "epoch": 0.8353832637821217, + "grad_norm": 2.189000368118286, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6876223087310791, + "num_tokens": 195231244.0, + "step": 7607 + }, + { + "epoch": 0.8354930814847353, + "grad_norm": 2.2657198905944824, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7024143934249878, + "num_tokens": 195256379.0, + "step": 7608 + }, + { + "epoch": 0.835602899187349, + "grad_norm": 2.178427219390869, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6924781799316406, + "num_tokens": 195281843.0, + "step": 7609 + }, + { + "epoch": 0.8357127168899626, + "grad_norm": 2.1406540870666504, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7089323997497559, + "num_tokens": 195308552.0, + "step": 7610 + }, + { + "epoch": 0.8358225345925763, + "grad_norm": 2.0130465030670166, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7130019068717957, + "num_tokens": 195335943.0, + "step": 7611 + }, + { + "epoch": 0.83593235229519, + "grad_norm": 2.442880630493164, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7013014554977417, + "num_tokens": 195359423.0, + "step": 7612 + }, + { + "epoch": 0.8360421699978037, + "grad_norm": 2.245865821838379, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6913044452667236, + "num_tokens": 195384708.0, + "step": 7613 + }, + { + "epoch": 0.8361519877004173, + "grad_norm": 2.5290603637695312, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7113180160522461, + "num_tokens": 195405595.0, + "step": 7614 + }, + { + "epoch": 0.836261805403031, + "grad_norm": 2.1146202087402344, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7074682712554932, + "num_tokens": 195433168.0, + "step": 7615 + }, + { + "epoch": 0.8363716231056446, + "grad_norm": 2.0070407390594482, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.717803955078125, + "num_tokens": 195462723.0, + "step": 7616 + }, + { + "epoch": 0.8364814408082583, + "grad_norm": 2.25799822807312, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7074048519134521, + "num_tokens": 195487603.0, + "step": 7617 + }, + { + "epoch": 0.8365912585108719, + "grad_norm": 2.2867438793182373, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7493338584899902, + "num_tokens": 195509805.0, + "step": 7618 + }, + { + "epoch": 0.8367010762134857, + "grad_norm": 2.2146666049957275, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7195618152618408, + "num_tokens": 195533076.0, + "step": 7619 + }, + { + "epoch": 0.8368108939160993, + "grad_norm": 2.074925184249878, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7122219800949097, + "num_tokens": 195560165.0, + "step": 7620 + }, + { + "epoch": 0.836920711618713, + "grad_norm": 2.4173948764801025, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7009108662605286, + "num_tokens": 195582983.0, + "step": 7621 + }, + { + "epoch": 0.8370305293213266, + "grad_norm": 2.160939931869507, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7163504958152771, + "num_tokens": 195608541.0, + "step": 7622 + }, + { + "epoch": 0.8371403470239402, + "grad_norm": 2.360398769378662, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6971036791801453, + "num_tokens": 195631677.0, + "step": 7623 + }, + { + "epoch": 0.8372501647265539, + "grad_norm": 2.5399580001831055, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7140718102455139, + "num_tokens": 195652951.0, + "step": 7624 + }, + { + "epoch": 0.8373599824291675, + "grad_norm": 2.3268165588378906, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7210604548454285, + "num_tokens": 195675690.0, + "step": 7625 + }, + { + "epoch": 0.8374698001317813, + "grad_norm": 2.2715325355529785, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7307964563369751, + "num_tokens": 195699250.0, + "step": 7626 + }, + { + "epoch": 0.8375796178343949, + "grad_norm": 2.378075361251831, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7359253764152527, + "num_tokens": 195719354.0, + "step": 7627 + }, + { + "epoch": 0.8376894355370086, + "grad_norm": 2.228058338165283, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6956287622451782, + "num_tokens": 195743855.0, + "step": 7628 + }, + { + "epoch": 0.8377992532396222, + "grad_norm": 1.9912508726119995, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7377206087112427, + "num_tokens": 195772270.0, + "step": 7629 + }, + { + "epoch": 0.8379090709422359, + "grad_norm": 2.3204777240753174, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7137116193771362, + "num_tokens": 195794886.0, + "step": 7630 + }, + { + "epoch": 0.8380188886448495, + "grad_norm": 2.190852403640747, + "learning_rate": 1e-06, + "loss": 1.0856, + "mean_token_accuracy": 0.6715496778488159, + "num_tokens": 195821366.0, + "step": 7631 + }, + { + "epoch": 0.8381287063474632, + "grad_norm": 2.2120745182037354, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7246765494346619, + "num_tokens": 195845207.0, + "step": 7632 + }, + { + "epoch": 0.8382385240500769, + "grad_norm": 2.0638082027435303, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7187801599502563, + "num_tokens": 195872668.0, + "step": 7633 + }, + { + "epoch": 0.8383483417526906, + "grad_norm": 2.35097599029541, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7085795402526855, + "num_tokens": 195896315.0, + "step": 7634 + }, + { + "epoch": 0.8384581594553042, + "grad_norm": 2.2179746627807617, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7064039707183838, + "num_tokens": 195922309.0, + "step": 7635 + }, + { + "epoch": 0.8385679771579179, + "grad_norm": 1.9846802949905396, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7074213027954102, + "num_tokens": 195953280.0, + "step": 7636 + }, + { + "epoch": 0.8386777948605315, + "grad_norm": 1.9916653633117676, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7067564725875854, + "num_tokens": 195983093.0, + "step": 7637 + }, + { + "epoch": 0.8387876125631452, + "grad_norm": 2.006746768951416, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7151136994361877, + "num_tokens": 196012589.0, + "step": 7638 + }, + { + "epoch": 0.8388974302657588, + "grad_norm": 2.1491994857788086, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7206552028656006, + "num_tokens": 196037802.0, + "step": 7639 + }, + { + "epoch": 0.8390072479683724, + "grad_norm": 2.31894588470459, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6914685368537903, + "num_tokens": 196060678.0, + "step": 7640 + }, + { + "epoch": 0.8391170656709862, + "grad_norm": 2.204556465148926, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7363656759262085, + "num_tokens": 196086610.0, + "step": 7641 + }, + { + "epoch": 0.8392268833735999, + "grad_norm": 2.1789209842681885, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7153804898262024, + "num_tokens": 196114529.0, + "step": 7642 + }, + { + "epoch": 0.8393367010762135, + "grad_norm": 2.224050283432007, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7066066861152649, + "num_tokens": 196138137.0, + "step": 7643 + }, + { + "epoch": 0.8394465187788271, + "grad_norm": 2.0838088989257812, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6989798545837402, + "num_tokens": 196164813.0, + "step": 7644 + }, + { + "epoch": 0.8395563364814408, + "grad_norm": 2.052569627761841, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.711783766746521, + "num_tokens": 196193582.0, + "step": 7645 + }, + { + "epoch": 0.8396661541840544, + "grad_norm": 2.2950801849365234, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7101818323135376, + "num_tokens": 196216470.0, + "step": 7646 + }, + { + "epoch": 0.8397759718866681, + "grad_norm": 2.0703868865966797, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6983656883239746, + "num_tokens": 196245181.0, + "step": 7647 + }, + { + "epoch": 0.8398857895892818, + "grad_norm": 2.1993496417999268, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7075707912445068, + "num_tokens": 196271718.0, + "step": 7648 + }, + { + "epoch": 0.8399956072918955, + "grad_norm": 1.9858897924423218, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6955692172050476, + "num_tokens": 196301748.0, + "step": 7649 + }, + { + "epoch": 0.8401054249945091, + "grad_norm": 2.3917362689971924, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7113025784492493, + "num_tokens": 196324384.0, + "step": 7650 + }, + { + "epoch": 0.8402152426971228, + "grad_norm": 2.20861554145813, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7145061492919922, + "num_tokens": 196350465.0, + "step": 7651 + }, + { + "epoch": 0.8403250603997364, + "grad_norm": 2.1641411781311035, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.692524790763855, + "num_tokens": 196377689.0, + "step": 7652 + }, + { + "epoch": 0.8404348781023501, + "grad_norm": 2.2628562450408936, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6988750100135803, + "num_tokens": 196400140.0, + "step": 7653 + }, + { + "epoch": 0.8405446958049637, + "grad_norm": 2.024353265762329, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6930088996887207, + "num_tokens": 196427764.0, + "step": 7654 + }, + { + "epoch": 0.8406545135075775, + "grad_norm": 2.2305452823638916, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7503397464752197, + "num_tokens": 196450620.0, + "step": 7655 + }, + { + "epoch": 0.8407643312101911, + "grad_norm": 2.534303665161133, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6969847679138184, + "num_tokens": 196471359.0, + "step": 7656 + }, + { + "epoch": 0.8408741489128048, + "grad_norm": 2.1849753856658936, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.6863129138946533, + "num_tokens": 196496844.0, + "step": 7657 + }, + { + "epoch": 0.8409839666154184, + "grad_norm": 2.2136263847351074, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.724068284034729, + "num_tokens": 196519687.0, + "step": 7658 + }, + { + "epoch": 0.8410937843180321, + "grad_norm": 2.280302047729492, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6889276504516602, + "num_tokens": 196546529.0, + "step": 7659 + }, + { + "epoch": 0.8412036020206457, + "grad_norm": 2.092607259750366, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6950141191482544, + "num_tokens": 196574631.0, + "step": 7660 + }, + { + "epoch": 0.8413134197232593, + "grad_norm": 2.1260673999786377, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6841719746589661, + "num_tokens": 196601355.0, + "step": 7661 + }, + { + "epoch": 0.8414232374258731, + "grad_norm": 2.6815085411071777, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.704397439956665, + "num_tokens": 196618866.0, + "step": 7662 + }, + { + "epoch": 0.8415330551284868, + "grad_norm": 2.1318464279174805, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7196834087371826, + "num_tokens": 196647060.0, + "step": 7663 + }, + { + "epoch": 0.8416428728311004, + "grad_norm": 2.287064552307129, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7107377052307129, + "num_tokens": 196669131.0, + "step": 7664 + }, + { + "epoch": 0.841752690533714, + "grad_norm": 2.100184679031372, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7071279883384705, + "num_tokens": 196695839.0, + "step": 7665 + }, + { + "epoch": 0.8418625082363277, + "grad_norm": 2.1489369869232178, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.695661187171936, + "num_tokens": 196721466.0, + "step": 7666 + }, + { + "epoch": 0.8419723259389413, + "grad_norm": 2.2077767848968506, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7197349667549133, + "num_tokens": 196745782.0, + "step": 7667 + }, + { + "epoch": 0.842082143641555, + "grad_norm": 2.4103879928588867, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6933320164680481, + "num_tokens": 196769142.0, + "step": 7668 + }, + { + "epoch": 0.8421919613441686, + "grad_norm": 2.010411500930786, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7012801170349121, + "num_tokens": 196797857.0, + "step": 7669 + }, + { + "epoch": 0.8423017790467824, + "grad_norm": 2.412076473236084, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7040414214134216, + "num_tokens": 196820348.0, + "step": 7670 + }, + { + "epoch": 0.842411596749396, + "grad_norm": 2.111534595489502, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7199275493621826, + "num_tokens": 196846800.0, + "step": 7671 + }, + { + "epoch": 0.8425214144520097, + "grad_norm": 2.327451705932617, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.697838306427002, + "num_tokens": 196871490.0, + "step": 7672 + }, + { + "epoch": 0.8426312321546233, + "grad_norm": 2.062237024307251, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6977204084396362, + "num_tokens": 196901480.0, + "step": 7673 + }, + { + "epoch": 0.842741049857237, + "grad_norm": 2.4090747833251953, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7205181121826172, + "num_tokens": 196924691.0, + "step": 7674 + }, + { + "epoch": 0.8428508675598506, + "grad_norm": 2.153320074081421, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7399721145629883, + "num_tokens": 196949382.0, + "step": 7675 + }, + { + "epoch": 0.8429606852624643, + "grad_norm": 2.2857823371887207, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6943468451499939, + "num_tokens": 196974998.0, + "step": 7676 + }, + { + "epoch": 0.843070502965078, + "grad_norm": 2.4761624336242676, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6898101568222046, + "num_tokens": 196996397.0, + "step": 7677 + }, + { + "epoch": 0.8431803206676917, + "grad_norm": 2.1570353507995605, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7174956798553467, + "num_tokens": 197020647.0, + "step": 7678 + }, + { + "epoch": 0.8432901383703053, + "grad_norm": 2.115985631942749, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7106565237045288, + "num_tokens": 197046417.0, + "step": 7679 + }, + { + "epoch": 0.843399956072919, + "grad_norm": 2.099973201751709, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7242411375045776, + "num_tokens": 197071680.0, + "step": 7680 + }, + { + "epoch": 0.8435097737755326, + "grad_norm": 2.38657283782959, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7079788446426392, + "num_tokens": 197091852.0, + "step": 7681 + }, + { + "epoch": 0.8436195914781462, + "grad_norm": 2.2638070583343506, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7373447418212891, + "num_tokens": 197113002.0, + "step": 7682 + }, + { + "epoch": 0.8437294091807599, + "grad_norm": 2.220111131668091, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7222058176994324, + "num_tokens": 197136251.0, + "step": 7683 + }, + { + "epoch": 0.8438392268833736, + "grad_norm": 1.9464985132217407, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7080720663070679, + "num_tokens": 197166715.0, + "step": 7684 + }, + { + "epoch": 0.8439490445859873, + "grad_norm": 2.3039186000823975, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6963784694671631, + "num_tokens": 197190896.0, + "step": 7685 + }, + { + "epoch": 0.8440588622886009, + "grad_norm": 2.4736883640289307, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7065972089767456, + "num_tokens": 197211353.0, + "step": 7686 + }, + { + "epoch": 0.8441686799912146, + "grad_norm": 2.0736207962036133, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6958563327789307, + "num_tokens": 197238861.0, + "step": 7687 + }, + { + "epoch": 0.8442784976938282, + "grad_norm": 2.281399726867676, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7099153399467468, + "num_tokens": 197262694.0, + "step": 7688 + }, + { + "epoch": 0.8443883153964419, + "grad_norm": 2.1787519454956055, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7325176000595093, + "num_tokens": 197288212.0, + "step": 7689 + }, + { + "epoch": 0.8444981330990555, + "grad_norm": 2.4100749492645264, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7110708951950073, + "num_tokens": 197308506.0, + "step": 7690 + }, + { + "epoch": 0.8446079508016693, + "grad_norm": 2.3532135486602783, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6806011199951172, + "num_tokens": 197332576.0, + "step": 7691 + }, + { + "epoch": 0.8447177685042829, + "grad_norm": 1.8782387971878052, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.693615734577179, + "num_tokens": 197365810.0, + "step": 7692 + }, + { + "epoch": 0.8448275862068966, + "grad_norm": 1.9955061674118042, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6980124115943909, + "num_tokens": 197395845.0, + "step": 7693 + }, + { + "epoch": 0.8449374039095102, + "grad_norm": 2.0568106174468994, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7227927446365356, + "num_tokens": 197423642.0, + "step": 7694 + }, + { + "epoch": 0.8450472216121239, + "grad_norm": 2.621652841567993, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.72739577293396, + "num_tokens": 197441063.0, + "step": 7695 + }, + { + "epoch": 0.8451570393147375, + "grad_norm": 2.106753349304199, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.731333315372467, + "num_tokens": 197466636.0, + "step": 7696 + }, + { + "epoch": 0.8452668570173512, + "grad_norm": 2.371948719024658, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7181498408317566, + "num_tokens": 197489263.0, + "step": 7697 + }, + { + "epoch": 0.8453766747199648, + "grad_norm": 2.1296310424804688, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7156558036804199, + "num_tokens": 197515337.0, + "step": 7698 + }, + { + "epoch": 0.8454864924225786, + "grad_norm": 2.0187504291534424, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6859096884727478, + "num_tokens": 197546326.0, + "step": 7699 + }, + { + "epoch": 0.8455963101251922, + "grad_norm": 2.046642780303955, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7093715667724609, + "num_tokens": 197573974.0, + "step": 7700 + }, + { + "epoch": 0.8457061278278059, + "grad_norm": 2.0092947483062744, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6998032331466675, + "num_tokens": 197602952.0, + "step": 7701 + }, + { + "epoch": 0.8458159455304195, + "grad_norm": 2.207969903945923, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7234336733818054, + "num_tokens": 197626765.0, + "step": 7702 + }, + { + "epoch": 0.8459257632330331, + "grad_norm": 2.082087755203247, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7127552032470703, + "num_tokens": 197654322.0, + "step": 7703 + }, + { + "epoch": 0.8460355809356468, + "grad_norm": 2.3172519207000732, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6988147497177124, + "num_tokens": 197677440.0, + "step": 7704 + }, + { + "epoch": 0.8461453986382604, + "grad_norm": 1.8599302768707275, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7057970762252808, + "num_tokens": 197711067.0, + "step": 7705 + }, + { + "epoch": 0.8462552163408742, + "grad_norm": 1.7742950916290283, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6915904879570007, + "num_tokens": 197748666.0, + "step": 7706 + }, + { + "epoch": 0.8463650340434878, + "grad_norm": 2.187671661376953, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7058364748954773, + "num_tokens": 197771832.0, + "step": 7707 + }, + { + "epoch": 0.8464748517461015, + "grad_norm": 1.9622001647949219, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7088668346405029, + "num_tokens": 197800777.0, + "step": 7708 + }, + { + "epoch": 0.8465846694487151, + "grad_norm": 2.2566399574279785, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7177157402038574, + "num_tokens": 197823456.0, + "step": 7709 + }, + { + "epoch": 0.8466944871513288, + "grad_norm": 2.1788172721862793, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7096973657608032, + "num_tokens": 197849404.0, + "step": 7710 + }, + { + "epoch": 0.8468043048539424, + "grad_norm": 2.201117992401123, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.716521143913269, + "num_tokens": 197874067.0, + "step": 7711 + }, + { + "epoch": 0.8469141225565561, + "grad_norm": 1.8437514305114746, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6979708671569824, + "num_tokens": 197909241.0, + "step": 7712 + }, + { + "epoch": 0.8470239402591698, + "grad_norm": 2.2361979484558105, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6997087001800537, + "num_tokens": 197934485.0, + "step": 7713 + }, + { + "epoch": 0.8471337579617835, + "grad_norm": 2.186669111251831, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7122069597244263, + "num_tokens": 197958829.0, + "step": 7714 + }, + { + "epoch": 0.8472435756643971, + "grad_norm": 2.2730987071990967, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6939646005630493, + "num_tokens": 197984362.0, + "step": 7715 + }, + { + "epoch": 0.8473533933670108, + "grad_norm": 1.9964344501495361, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7066349983215332, + "num_tokens": 198014010.0, + "step": 7716 + }, + { + "epoch": 0.8474632110696244, + "grad_norm": 2.350635051727295, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6981214880943298, + "num_tokens": 198039251.0, + "step": 7717 + }, + { + "epoch": 0.8475730287722381, + "grad_norm": 2.1114554405212402, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6886325478553772, + "num_tokens": 198065955.0, + "step": 7718 + }, + { + "epoch": 0.8476828464748517, + "grad_norm": 2.0218451023101807, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7241076827049255, + "num_tokens": 198094172.0, + "step": 7719 + }, + { + "epoch": 0.8477926641774655, + "grad_norm": 2.100648880004883, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7232087850570679, + "num_tokens": 198119988.0, + "step": 7720 + }, + { + "epoch": 0.8479024818800791, + "grad_norm": 2.0426716804504395, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6890978217124939, + "num_tokens": 198150679.0, + "step": 7721 + }, + { + "epoch": 0.8480122995826928, + "grad_norm": 2.2780144214630127, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7224406599998474, + "num_tokens": 198172014.0, + "step": 7722 + }, + { + "epoch": 0.8481221172853064, + "grad_norm": 2.212904453277588, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.693329393863678, + "num_tokens": 198198985.0, + "step": 7723 + }, + { + "epoch": 0.84823193498792, + "grad_norm": 2.085005283355713, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7019265294075012, + "num_tokens": 198224667.0, + "step": 7724 + }, + { + "epoch": 0.8483417526905337, + "grad_norm": 2.1483373641967773, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7025227546691895, + "num_tokens": 198251056.0, + "step": 7725 + }, + { + "epoch": 0.8484515703931473, + "grad_norm": 2.438502550125122, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7012988328933716, + "num_tokens": 198274127.0, + "step": 7726 + }, + { + "epoch": 0.848561388095761, + "grad_norm": 2.1514980792999268, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6796579360961914, + "num_tokens": 198302367.0, + "step": 7727 + }, + { + "epoch": 0.8486712057983747, + "grad_norm": 2.0430073738098145, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7028553485870361, + "num_tokens": 198332058.0, + "step": 7728 + }, + { + "epoch": 0.8487810235009884, + "grad_norm": 2.16115665435791, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.7002512216567993, + "num_tokens": 198358683.0, + "step": 7729 + }, + { + "epoch": 0.848890841203602, + "grad_norm": 2.008913993835449, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7083445191383362, + "num_tokens": 198388211.0, + "step": 7730 + }, + { + "epoch": 0.8490006589062157, + "grad_norm": 2.0354251861572266, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.687698245048523, + "num_tokens": 198417454.0, + "step": 7731 + }, + { + "epoch": 0.8491104766088293, + "grad_norm": 2.242527961730957, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7197387218475342, + "num_tokens": 198440141.0, + "step": 7732 + }, + { + "epoch": 0.849220294311443, + "grad_norm": 2.518096685409546, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7057042121887207, + "num_tokens": 198460371.0, + "step": 7733 + }, + { + "epoch": 0.8493301120140566, + "grad_norm": 2.476365566253662, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.715609073638916, + "num_tokens": 198479682.0, + "step": 7734 + }, + { + "epoch": 0.8494399297166704, + "grad_norm": 2.1593732833862305, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6813936233520508, + "num_tokens": 198507170.0, + "step": 7735 + }, + { + "epoch": 0.849549747419284, + "grad_norm": 2.11643123626709, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6965689659118652, + "num_tokens": 198534125.0, + "step": 7736 + }, + { + "epoch": 0.8496595651218977, + "grad_norm": 2.1790614128112793, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6938788890838623, + "num_tokens": 198559896.0, + "step": 7737 + }, + { + "epoch": 0.8497693828245113, + "grad_norm": 2.438807249069214, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7321953773498535, + "num_tokens": 198579324.0, + "step": 7738 + }, + { + "epoch": 0.849879200527125, + "grad_norm": 2.277249813079834, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7367902398109436, + "num_tokens": 198601076.0, + "step": 7739 + }, + { + "epoch": 0.8499890182297386, + "grad_norm": 1.9907598495483398, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6939348578453064, + "num_tokens": 198630532.0, + "step": 7740 + }, + { + "epoch": 0.8500988359323522, + "grad_norm": 2.3192076683044434, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7061835527420044, + "num_tokens": 198653698.0, + "step": 7741 + }, + { + "epoch": 0.850208653634966, + "grad_norm": 2.2197248935699463, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7309544682502747, + "num_tokens": 198675222.0, + "step": 7742 + }, + { + "epoch": 0.8503184713375797, + "grad_norm": 1.9817110300064087, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7190371751785278, + "num_tokens": 198704333.0, + "step": 7743 + }, + { + "epoch": 0.8504282890401933, + "grad_norm": 2.0237631797790527, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7256357073783875, + "num_tokens": 198735325.0, + "step": 7744 + }, + { + "epoch": 0.8505381067428069, + "grad_norm": 2.30378794670105, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6867769360542297, + "num_tokens": 198759279.0, + "step": 7745 + }, + { + "epoch": 0.8506479244454206, + "grad_norm": 1.8994501829147339, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6910115480422974, + "num_tokens": 198790627.0, + "step": 7746 + }, + { + "epoch": 0.8507577421480342, + "grad_norm": 2.449334144592285, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7061243057250977, + "num_tokens": 198811504.0, + "step": 7747 + }, + { + "epoch": 0.8508675598506479, + "grad_norm": 2.2104921340942383, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7272664308547974, + "num_tokens": 198833808.0, + "step": 7748 + }, + { + "epoch": 0.8509773775532616, + "grad_norm": 2.59452748298645, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7101485133171082, + "num_tokens": 198851992.0, + "step": 7749 + }, + { + "epoch": 0.8510871952558753, + "grad_norm": 1.9650346040725708, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7115999460220337, + "num_tokens": 198881699.0, + "step": 7750 + }, + { + "epoch": 0.8511970129584889, + "grad_norm": 2.212437868118286, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6925492286682129, + "num_tokens": 198909424.0, + "step": 7751 + }, + { + "epoch": 0.8513068306611026, + "grad_norm": 2.2816076278686523, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6996742486953735, + "num_tokens": 198933470.0, + "step": 7752 + }, + { + "epoch": 0.8514166483637162, + "grad_norm": 1.9280753135681152, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6804571151733398, + "num_tokens": 198966834.0, + "step": 7753 + }, + { + "epoch": 0.8515264660663299, + "grad_norm": 2.2172088623046875, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7015490531921387, + "num_tokens": 198992188.0, + "step": 7754 + }, + { + "epoch": 0.8516362837689435, + "grad_norm": 2.584743022918701, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7284136414527893, + "num_tokens": 199010834.0, + "step": 7755 + }, + { + "epoch": 0.8517461014715572, + "grad_norm": 2.361326217651367, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7231913805007935, + "num_tokens": 199031594.0, + "step": 7756 + }, + { + "epoch": 0.8518559191741709, + "grad_norm": 2.140475034713745, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7245094776153564, + "num_tokens": 199057385.0, + "step": 7757 + }, + { + "epoch": 0.8519657368767846, + "grad_norm": 2.1048409938812256, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7031029462814331, + "num_tokens": 199085584.0, + "step": 7758 + }, + { + "epoch": 0.8520755545793982, + "grad_norm": 2.511040687561035, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6983342170715332, + "num_tokens": 199107799.0, + "step": 7759 + }, + { + "epoch": 0.8521853722820119, + "grad_norm": 2.199402332305908, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7036131620407104, + "num_tokens": 199137540.0, + "step": 7760 + }, + { + "epoch": 0.8522951899846255, + "grad_norm": 2.1560544967651367, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6912118196487427, + "num_tokens": 199163930.0, + "step": 7761 + }, + { + "epoch": 0.8524050076872391, + "grad_norm": 2.2901363372802734, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6731762290000916, + "num_tokens": 199188912.0, + "step": 7762 + }, + { + "epoch": 0.8525148253898528, + "grad_norm": 2.1448051929473877, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7068867683410645, + "num_tokens": 199213470.0, + "step": 7763 + }, + { + "epoch": 0.8526246430924665, + "grad_norm": 2.2056093215942383, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7026147842407227, + "num_tokens": 199238053.0, + "step": 7764 + }, + { + "epoch": 0.8527344607950802, + "grad_norm": 2.1568143367767334, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7050192356109619, + "num_tokens": 199262883.0, + "step": 7765 + }, + { + "epoch": 0.8528442784976938, + "grad_norm": 2.298832893371582, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.67923504114151, + "num_tokens": 199287795.0, + "step": 7766 + }, + { + "epoch": 0.8529540962003075, + "grad_norm": 2.495851755142212, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6959710121154785, + "num_tokens": 199309103.0, + "step": 7767 + }, + { + "epoch": 0.8530639139029211, + "grad_norm": 2.2277159690856934, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6754143834114075, + "num_tokens": 199337457.0, + "step": 7768 + }, + { + "epoch": 0.8531737316055348, + "grad_norm": 2.4666824340820312, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7174191474914551, + "num_tokens": 199359119.0, + "step": 7769 + }, + { + "epoch": 0.8532835493081484, + "grad_norm": 2.1587319374084473, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7204069495201111, + "num_tokens": 199383271.0, + "step": 7770 + }, + { + "epoch": 0.8533933670107622, + "grad_norm": 2.3582308292388916, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7202090620994568, + "num_tokens": 199406318.0, + "step": 7771 + }, + { + "epoch": 0.8535031847133758, + "grad_norm": 2.058865547180176, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7089762687683105, + "num_tokens": 199436346.0, + "step": 7772 + }, + { + "epoch": 0.8536130024159895, + "grad_norm": 2.0763587951660156, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7327311635017395, + "num_tokens": 199463380.0, + "step": 7773 + }, + { + "epoch": 0.8537228201186031, + "grad_norm": 2.280642509460449, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7127187252044678, + "num_tokens": 199487495.0, + "step": 7774 + }, + { + "epoch": 0.8538326378212168, + "grad_norm": 2.113701343536377, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7225103378295898, + "num_tokens": 199516664.0, + "step": 7775 + }, + { + "epoch": 0.8539424555238304, + "grad_norm": 2.2956066131591797, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6975044012069702, + "num_tokens": 199540027.0, + "step": 7776 + }, + { + "epoch": 0.8540522732264441, + "grad_norm": 2.1787567138671875, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7140541076660156, + "num_tokens": 199565466.0, + "step": 7777 + }, + { + "epoch": 0.8541620909290578, + "grad_norm": 2.3331868648529053, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7464516162872314, + "num_tokens": 199587092.0, + "step": 7778 + }, + { + "epoch": 0.8542719086316715, + "grad_norm": 2.257059097290039, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.684215247631073, + "num_tokens": 199611705.0, + "step": 7779 + }, + { + "epoch": 0.8543817263342851, + "grad_norm": 1.9461456537246704, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7031139135360718, + "num_tokens": 199643539.0, + "step": 7780 + }, + { + "epoch": 0.8544915440368988, + "grad_norm": 2.0637335777282715, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6980798840522766, + "num_tokens": 199672664.0, + "step": 7781 + }, + { + "epoch": 0.8546013617395124, + "grad_norm": 1.9835268259048462, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.6991435289382935, + "num_tokens": 199699874.0, + "step": 7782 + }, + { + "epoch": 0.854711179442126, + "grad_norm": 1.9932994842529297, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7131084203720093, + "num_tokens": 199727146.0, + "step": 7783 + }, + { + "epoch": 0.8548209971447397, + "grad_norm": 2.0480313301086426, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6913458108901978, + "num_tokens": 199755041.0, + "step": 7784 + }, + { + "epoch": 0.8549308148473534, + "grad_norm": 2.325904607772827, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7220280170440674, + "num_tokens": 199775721.0, + "step": 7785 + }, + { + "epoch": 0.8550406325499671, + "grad_norm": 2.523148536682129, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.701275646686554, + "num_tokens": 199797126.0, + "step": 7786 + }, + { + "epoch": 0.8551504502525807, + "grad_norm": 2.104771375656128, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.7063651084899902, + "num_tokens": 199823281.0, + "step": 7787 + }, + { + "epoch": 0.8552602679551944, + "grad_norm": 2.009969711303711, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7064011096954346, + "num_tokens": 199853504.0, + "step": 7788 + }, + { + "epoch": 0.855370085657808, + "grad_norm": 2.118237018585205, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7109367847442627, + "num_tokens": 199879878.0, + "step": 7789 + }, + { + "epoch": 0.8554799033604217, + "grad_norm": 2.09053635597229, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7095532417297363, + "num_tokens": 199905466.0, + "step": 7790 + }, + { + "epoch": 0.8555897210630353, + "grad_norm": 2.2508084774017334, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7171563506126404, + "num_tokens": 199929155.0, + "step": 7791 + }, + { + "epoch": 0.855699538765649, + "grad_norm": 2.3714475631713867, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7201827764511108, + "num_tokens": 199951182.0, + "step": 7792 + }, + { + "epoch": 0.8558093564682627, + "grad_norm": 2.101325273513794, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7071913480758667, + "num_tokens": 199977843.0, + "step": 7793 + }, + { + "epoch": 0.8559191741708764, + "grad_norm": 2.1657962799072266, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.690758228302002, + "num_tokens": 200001517.0, + "step": 7794 + }, + { + "epoch": 0.85602899187349, + "grad_norm": 1.919632911682129, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.70917809009552, + "num_tokens": 200034059.0, + "step": 7795 + }, + { + "epoch": 0.8561388095761037, + "grad_norm": 2.3893327713012695, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7000752687454224, + "num_tokens": 200057394.0, + "step": 7796 + }, + { + "epoch": 0.8562486272787173, + "grad_norm": 2.4473705291748047, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6984052658081055, + "num_tokens": 200079254.0, + "step": 7797 + }, + { + "epoch": 0.856358444981331, + "grad_norm": 2.3940768241882324, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7197281122207642, + "num_tokens": 200099920.0, + "step": 7798 + }, + { + "epoch": 0.8564682626839446, + "grad_norm": 2.608912467956543, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7203161716461182, + "num_tokens": 200118256.0, + "step": 7799 + }, + { + "epoch": 0.8565780803865584, + "grad_norm": 2.5481114387512207, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7308205962181091, + "num_tokens": 200135693.0, + "step": 7800 + }, + { + "epoch": 0.856687898089172, + "grad_norm": 2.3598012924194336, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7242307662963867, + "num_tokens": 200158075.0, + "step": 7801 + }, + { + "epoch": 0.8567977157917857, + "grad_norm": 2.0706191062927246, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.681685209274292, + "num_tokens": 200186521.0, + "step": 7802 + }, + { + "epoch": 0.8569075334943993, + "grad_norm": 2.3573999404907227, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.709030032157898, + "num_tokens": 200208822.0, + "step": 7803 + }, + { + "epoch": 0.8570173511970129, + "grad_norm": 2.200315475463867, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7166669964790344, + "num_tokens": 200235488.0, + "step": 7804 + }, + { + "epoch": 0.8571271688996266, + "grad_norm": 2.2198874950408936, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6903319358825684, + "num_tokens": 200259823.0, + "step": 7805 + }, + { + "epoch": 0.8572369866022402, + "grad_norm": 1.9908267259597778, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7028753161430359, + "num_tokens": 200288446.0, + "step": 7806 + }, + { + "epoch": 0.857346804304854, + "grad_norm": 2.111359119415283, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7006836533546448, + "num_tokens": 200314362.0, + "step": 7807 + }, + { + "epoch": 0.8574566220074676, + "grad_norm": 2.1546385288238525, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.704125165939331, + "num_tokens": 200341799.0, + "step": 7808 + }, + { + "epoch": 0.8575664397100813, + "grad_norm": 1.962221622467041, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6952900886535645, + "num_tokens": 200372332.0, + "step": 7809 + }, + { + "epoch": 0.8576762574126949, + "grad_norm": 2.0968551635742188, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7242996692657471, + "num_tokens": 200398787.0, + "step": 7810 + }, + { + "epoch": 0.8577860751153086, + "grad_norm": 2.3629727363586426, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7154075503349304, + "num_tokens": 200420366.0, + "step": 7811 + }, + { + "epoch": 0.8578958928179222, + "grad_norm": 2.2516396045684814, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7124149799346924, + "num_tokens": 200444261.0, + "step": 7812 + }, + { + "epoch": 0.8580057105205359, + "grad_norm": 2.2683494091033936, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7227029204368591, + "num_tokens": 200467092.0, + "step": 7813 + }, + { + "epoch": 0.8581155282231496, + "grad_norm": 2.2505276203155518, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6876058578491211, + "num_tokens": 200495423.0, + "step": 7814 + }, + { + "epoch": 0.8582253459257633, + "grad_norm": 2.273733377456665, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6979275941848755, + "num_tokens": 200518694.0, + "step": 7815 + }, + { + "epoch": 0.8583351636283769, + "grad_norm": 2.016482353210449, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6984817385673523, + "num_tokens": 200549616.0, + "step": 7816 + }, + { + "epoch": 0.8584449813309906, + "grad_norm": 2.130229949951172, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.6991525292396545, + "num_tokens": 200574760.0, + "step": 7817 + }, + { + "epoch": 0.8585547990336042, + "grad_norm": 2.0936083793640137, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7458599805831909, + "num_tokens": 200599629.0, + "step": 7818 + }, + { + "epoch": 0.8586646167362179, + "grad_norm": 2.2236990928649902, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7142760753631592, + "num_tokens": 200625726.0, + "step": 7819 + }, + { + "epoch": 0.8587744344388315, + "grad_norm": 2.132816791534424, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7166086435317993, + "num_tokens": 200652561.0, + "step": 7820 + }, + { + "epoch": 0.8588842521414451, + "grad_norm": 2.2722227573394775, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7041178345680237, + "num_tokens": 200675905.0, + "step": 7821 + }, + { + "epoch": 0.8589940698440589, + "grad_norm": 2.1030781269073486, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6851882934570312, + "num_tokens": 200702772.0, + "step": 7822 + }, + { + "epoch": 0.8591038875466726, + "grad_norm": 2.1743075847625732, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7180003523826599, + "num_tokens": 200727785.0, + "step": 7823 + }, + { + "epoch": 0.8592137052492862, + "grad_norm": 2.2762372493743896, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7010917663574219, + "num_tokens": 200750719.0, + "step": 7824 + }, + { + "epoch": 0.8593235229518998, + "grad_norm": 2.1523008346557617, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7055662870407104, + "num_tokens": 200776087.0, + "step": 7825 + }, + { + "epoch": 0.8594333406545135, + "grad_norm": 2.2606379985809326, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7004441022872925, + "num_tokens": 200801056.0, + "step": 7826 + }, + { + "epoch": 0.8595431583571271, + "grad_norm": 1.9823468923568726, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7157869338989258, + "num_tokens": 200828439.0, + "step": 7827 + }, + { + "epoch": 0.8596529760597408, + "grad_norm": 2.066671848297119, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7098629474639893, + "num_tokens": 200854507.0, + "step": 7828 + }, + { + "epoch": 0.8597627937623545, + "grad_norm": 2.3481366634368896, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7472659349441528, + "num_tokens": 200874875.0, + "step": 7829 + }, + { + "epoch": 0.8598726114649682, + "grad_norm": 2.300718307495117, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7037847638130188, + "num_tokens": 200899505.0, + "step": 7830 + }, + { + "epoch": 0.8599824291675818, + "grad_norm": 2.123946189880371, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7237281799316406, + "num_tokens": 200925304.0, + "step": 7831 + }, + { + "epoch": 0.8600922468701955, + "grad_norm": 2.028545618057251, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6915807127952576, + "num_tokens": 200955237.0, + "step": 7832 + }, + { + "epoch": 0.8602020645728091, + "grad_norm": 2.1149821281433105, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6908798217773438, + "num_tokens": 200982285.0, + "step": 7833 + }, + { + "epoch": 0.8603118822754228, + "grad_norm": 2.3086647987365723, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7068676352500916, + "num_tokens": 201005737.0, + "step": 7834 + }, + { + "epoch": 0.8604216999780364, + "grad_norm": 2.214083194732666, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7013092041015625, + "num_tokens": 201029367.0, + "step": 7835 + }, + { + "epoch": 0.8605315176806502, + "grad_norm": 2.128518581390381, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7074459195137024, + "num_tokens": 201055443.0, + "step": 7836 + }, + { + "epoch": 0.8606413353832638, + "grad_norm": 2.1746487617492676, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7035139799118042, + "num_tokens": 201081091.0, + "step": 7837 + }, + { + "epoch": 0.8607511530858775, + "grad_norm": 2.169206142425537, + "learning_rate": 1e-06, + "loss": 1.0654, + "mean_token_accuracy": 0.6765414476394653, + "num_tokens": 201110467.0, + "step": 7838 + }, + { + "epoch": 0.8608609707884911, + "grad_norm": 1.9559626579284668, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7018627524375916, + "num_tokens": 201140964.0, + "step": 7839 + }, + { + "epoch": 0.8609707884911048, + "grad_norm": 1.9651973247528076, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6778239011764526, + "num_tokens": 201173554.0, + "step": 7840 + }, + { + "epoch": 0.8610806061937184, + "grad_norm": 2.0787806510925293, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6945176124572754, + "num_tokens": 201202177.0, + "step": 7841 + }, + { + "epoch": 0.861190423896332, + "grad_norm": 2.162094831466675, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7109768390655518, + "num_tokens": 201229674.0, + "step": 7842 + }, + { + "epoch": 0.8613002415989458, + "grad_norm": 2.3513541221618652, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7398862838745117, + "num_tokens": 201250495.0, + "step": 7843 + }, + { + "epoch": 0.8614100593015594, + "grad_norm": 2.0083675384521484, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7013804316520691, + "num_tokens": 201278709.0, + "step": 7844 + }, + { + "epoch": 0.8615198770041731, + "grad_norm": 2.308493137359619, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7106406092643738, + "num_tokens": 201301005.0, + "step": 7845 + }, + { + "epoch": 0.8616296947067867, + "grad_norm": 2.3890151977539062, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7202334403991699, + "num_tokens": 201322332.0, + "step": 7846 + }, + { + "epoch": 0.8617395124094004, + "grad_norm": 2.449204206466675, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7130764722824097, + "num_tokens": 201342762.0, + "step": 7847 + }, + { + "epoch": 0.861849330112014, + "grad_norm": 2.1458117961883545, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.699640154838562, + "num_tokens": 201370402.0, + "step": 7848 + }, + { + "epoch": 0.8619591478146277, + "grad_norm": 2.3793959617614746, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.708607017993927, + "num_tokens": 201392227.0, + "step": 7849 + }, + { + "epoch": 0.8620689655172413, + "grad_norm": 1.9472622871398926, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.7010822892189026, + "num_tokens": 201423500.0, + "step": 7850 + }, + { + "epoch": 0.8621787832198551, + "grad_norm": 1.9148298501968384, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6972178220748901, + "num_tokens": 201454413.0, + "step": 7851 + }, + { + "epoch": 0.8622886009224687, + "grad_norm": 2.3468410968780518, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7204674482345581, + "num_tokens": 201476754.0, + "step": 7852 + }, + { + "epoch": 0.8623984186250824, + "grad_norm": 2.2510030269622803, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7206329107284546, + "num_tokens": 201500173.0, + "step": 7853 + }, + { + "epoch": 0.862508236327696, + "grad_norm": 2.216864824295044, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7067452669143677, + "num_tokens": 201525251.0, + "step": 7854 + }, + { + "epoch": 0.8626180540303097, + "grad_norm": 2.321136951446533, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7242436408996582, + "num_tokens": 201547161.0, + "step": 7855 + }, + { + "epoch": 0.8627278717329233, + "grad_norm": 2.2991464138031006, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.71529620885849, + "num_tokens": 201569663.0, + "step": 7856 + }, + { + "epoch": 0.862837689435537, + "grad_norm": 2.1608333587646484, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7102938890457153, + "num_tokens": 201595726.0, + "step": 7857 + }, + { + "epoch": 0.8629475071381507, + "grad_norm": 2.186082124710083, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7106091976165771, + "num_tokens": 201619482.0, + "step": 7858 + }, + { + "epoch": 0.8630573248407644, + "grad_norm": 2.4999709129333496, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7086894512176514, + "num_tokens": 201641805.0, + "step": 7859 + }, + { + "epoch": 0.863167142543378, + "grad_norm": 1.9549015760421753, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7161108255386353, + "num_tokens": 201669278.0, + "step": 7860 + }, + { + "epoch": 0.8632769602459917, + "grad_norm": 2.4365200996398926, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7286749482154846, + "num_tokens": 201689776.0, + "step": 7861 + }, + { + "epoch": 0.8633867779486053, + "grad_norm": 2.537128448486328, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6937798261642456, + "num_tokens": 201710119.0, + "step": 7862 + }, + { + "epoch": 0.863496595651219, + "grad_norm": 2.406392812728882, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7256046533584595, + "num_tokens": 201732402.0, + "step": 7863 + }, + { + "epoch": 0.8636064133538326, + "grad_norm": 2.3264195919036865, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7293092608451843, + "num_tokens": 201754235.0, + "step": 7864 + }, + { + "epoch": 0.8637162310564463, + "grad_norm": 2.162950277328491, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6949542760848999, + "num_tokens": 201780234.0, + "step": 7865 + }, + { + "epoch": 0.86382604875906, + "grad_norm": 2.155022621154785, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6942113637924194, + "num_tokens": 201806712.0, + "step": 7866 + }, + { + "epoch": 0.8639358664616736, + "grad_norm": 2.233196973800659, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7170323133468628, + "num_tokens": 201830211.0, + "step": 7867 + }, + { + "epoch": 0.8640456841642873, + "grad_norm": 2.6256861686706543, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6931084990501404, + "num_tokens": 201849159.0, + "step": 7868 + }, + { + "epoch": 0.8641555018669009, + "grad_norm": 2.190377712249756, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6946134567260742, + "num_tokens": 201876184.0, + "step": 7869 + }, + { + "epoch": 0.8642653195695146, + "grad_norm": 2.1480162143707275, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6909000277519226, + "num_tokens": 201902794.0, + "step": 7870 + }, + { + "epoch": 0.8643751372721282, + "grad_norm": 2.106497049331665, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.733445405960083, + "num_tokens": 201927998.0, + "step": 7871 + }, + { + "epoch": 0.864484954974742, + "grad_norm": 2.1779727935791016, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7472419738769531, + "num_tokens": 201950751.0, + "step": 7872 + }, + { + "epoch": 0.8645947726773556, + "grad_norm": 2.077172040939331, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6906297206878662, + "num_tokens": 201977633.0, + "step": 7873 + }, + { + "epoch": 0.8647045903799693, + "grad_norm": 2.0391790866851807, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7193597555160522, + "num_tokens": 202003446.0, + "step": 7874 + }, + { + "epoch": 0.8648144080825829, + "grad_norm": 2.4541332721710205, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6956847906112671, + "num_tokens": 202022916.0, + "step": 7875 + }, + { + "epoch": 0.8649242257851966, + "grad_norm": 2.288524866104126, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7126373052597046, + "num_tokens": 202046150.0, + "step": 7876 + }, + { + "epoch": 0.8650340434878102, + "grad_norm": 2.155514717102051, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7243444919586182, + "num_tokens": 202071060.0, + "step": 7877 + }, + { + "epoch": 0.8651438611904239, + "grad_norm": 2.2352542877197266, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7284855842590332, + "num_tokens": 202094232.0, + "step": 7878 + }, + { + "epoch": 0.8652536788930375, + "grad_norm": 2.210113525390625, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.6851220726966858, + "num_tokens": 202120815.0, + "step": 7879 + }, + { + "epoch": 0.8653634965956513, + "grad_norm": 2.020315170288086, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.692001461982727, + "num_tokens": 202151927.0, + "step": 7880 + }, + { + "epoch": 0.8654733142982649, + "grad_norm": 2.2624175548553467, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7032006978988647, + "num_tokens": 202176079.0, + "step": 7881 + }, + { + "epoch": 0.8655831320008786, + "grad_norm": 1.9856992959976196, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7013899087905884, + "num_tokens": 202204534.0, + "step": 7882 + }, + { + "epoch": 0.8656929497034922, + "grad_norm": 2.139521360397339, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7004490494728088, + "num_tokens": 202230889.0, + "step": 7883 + }, + { + "epoch": 0.8658027674061058, + "grad_norm": 2.441542625427246, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6802683472633362, + "num_tokens": 202252574.0, + "step": 7884 + }, + { + "epoch": 0.8659125851087195, + "grad_norm": 1.7739571332931519, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7073937058448792, + "num_tokens": 202287205.0, + "step": 7885 + }, + { + "epoch": 0.8660224028113331, + "grad_norm": 2.0446066856384277, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6959891319274902, + "num_tokens": 202316032.0, + "step": 7886 + }, + { + "epoch": 0.8661322205139469, + "grad_norm": 2.1099424362182617, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6798849105834961, + "num_tokens": 202345529.0, + "step": 7887 + }, + { + "epoch": 0.8662420382165605, + "grad_norm": 2.012951612472534, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6988121271133423, + "num_tokens": 202377743.0, + "step": 7888 + }, + { + "epoch": 0.8663518559191742, + "grad_norm": 1.967646837234497, + "learning_rate": 1e-06, + "loss": 1.087, + "mean_token_accuracy": 0.6789101362228394, + "num_tokens": 202408230.0, + "step": 7889 + }, + { + "epoch": 0.8664616736217878, + "grad_norm": 2.265174388885498, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7271479964256287, + "num_tokens": 202431069.0, + "step": 7890 + }, + { + "epoch": 0.8665714913244015, + "grad_norm": 2.2715983390808105, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6948050260543823, + "num_tokens": 202456057.0, + "step": 7891 + }, + { + "epoch": 0.8666813090270151, + "grad_norm": 1.9611603021621704, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7300209403038025, + "num_tokens": 202484012.0, + "step": 7892 + }, + { + "epoch": 0.8667911267296288, + "grad_norm": 2.202277183532715, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7103129625320435, + "num_tokens": 202507042.0, + "step": 7893 + }, + { + "epoch": 0.8669009444322425, + "grad_norm": 2.0405852794647217, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.7016758918762207, + "num_tokens": 202536950.0, + "step": 7894 + }, + { + "epoch": 0.8670107621348562, + "grad_norm": 2.245091676712036, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7241784334182739, + "num_tokens": 202561160.0, + "step": 7895 + }, + { + "epoch": 0.8671205798374698, + "grad_norm": 2.276959180831909, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6944701671600342, + "num_tokens": 202585789.0, + "step": 7896 + }, + { + "epoch": 0.8672303975400835, + "grad_norm": 1.9502102136611938, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6937101483345032, + "num_tokens": 202616244.0, + "step": 7897 + }, + { + "epoch": 0.8673402152426971, + "grad_norm": 1.981689214706421, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6960818767547607, + "num_tokens": 202650422.0, + "step": 7898 + }, + { + "epoch": 0.8674500329453108, + "grad_norm": 2.257399559020996, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6817653179168701, + "num_tokens": 202675270.0, + "step": 7899 + }, + { + "epoch": 0.8675598506479244, + "grad_norm": 2.1791725158691406, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7176932096481323, + "num_tokens": 202699629.0, + "step": 7900 + }, + { + "epoch": 0.8676696683505382, + "grad_norm": 2.256436824798584, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7205612659454346, + "num_tokens": 202723953.0, + "step": 7901 + }, + { + "epoch": 0.8677794860531518, + "grad_norm": 2.373175859451294, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6992285251617432, + "num_tokens": 202746677.0, + "step": 7902 + }, + { + "epoch": 0.8678893037557655, + "grad_norm": 2.269423007965088, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7233747839927673, + "num_tokens": 202770111.0, + "step": 7903 + }, + { + "epoch": 0.8679991214583791, + "grad_norm": 1.9562441110610962, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6951427459716797, + "num_tokens": 202801647.0, + "step": 7904 + }, + { + "epoch": 0.8681089391609927, + "grad_norm": 2.351844549179077, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6936026811599731, + "num_tokens": 202825479.0, + "step": 7905 + }, + { + "epoch": 0.8682187568636064, + "grad_norm": 2.3884284496307373, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6998310089111328, + "num_tokens": 202847927.0, + "step": 7906 + }, + { + "epoch": 0.86832857456622, + "grad_norm": 2.0272605419158936, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6878710985183716, + "num_tokens": 202878065.0, + "step": 7907 + }, + { + "epoch": 0.8684383922688337, + "grad_norm": 1.867433786392212, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7120394110679626, + "num_tokens": 202910158.0, + "step": 7908 + }, + { + "epoch": 0.8685482099714474, + "grad_norm": 2.127605438232422, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6910467743873596, + "num_tokens": 202936790.0, + "step": 7909 + }, + { + "epoch": 0.8686580276740611, + "grad_norm": 2.684065103530884, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7050939798355103, + "num_tokens": 202957337.0, + "step": 7910 + }, + { + "epoch": 0.8687678453766747, + "grad_norm": 2.0923948287963867, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.713905930519104, + "num_tokens": 202982973.0, + "step": 7911 + }, + { + "epoch": 0.8688776630792884, + "grad_norm": 2.4410862922668457, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7116845846176147, + "num_tokens": 203003602.0, + "step": 7912 + }, + { + "epoch": 0.868987480781902, + "grad_norm": 1.970959186553955, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7024638652801514, + "num_tokens": 203034010.0, + "step": 7913 + }, + { + "epoch": 0.8690972984845157, + "grad_norm": 2.2556493282318115, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.706578254699707, + "num_tokens": 203058796.0, + "step": 7914 + }, + { + "epoch": 0.8692071161871293, + "grad_norm": 2.286853313446045, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7226687669754028, + "num_tokens": 203083275.0, + "step": 7915 + }, + { + "epoch": 0.8693169338897431, + "grad_norm": 1.9832909107208252, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6745803952217102, + "num_tokens": 203114151.0, + "step": 7916 + }, + { + "epoch": 0.8694267515923567, + "grad_norm": 2.3975439071655273, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7030597925186157, + "num_tokens": 203139121.0, + "step": 7917 + }, + { + "epoch": 0.8695365692949704, + "grad_norm": 1.9542653560638428, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6978182792663574, + "num_tokens": 203170272.0, + "step": 7918 + }, + { + "epoch": 0.869646386997584, + "grad_norm": 2.1150128841400146, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7049229741096497, + "num_tokens": 203198102.0, + "step": 7919 + }, + { + "epoch": 0.8697562047001977, + "grad_norm": 2.1677539348602295, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6941237449645996, + "num_tokens": 203224668.0, + "step": 7920 + }, + { + "epoch": 0.8698660224028113, + "grad_norm": 2.0919289588928223, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7260205149650574, + "num_tokens": 203251142.0, + "step": 7921 + }, + { + "epoch": 0.869975840105425, + "grad_norm": 2.0702857971191406, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.7040422558784485, + "num_tokens": 203278421.0, + "step": 7922 + }, + { + "epoch": 0.8700856578080387, + "grad_norm": 2.2478384971618652, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7115283012390137, + "num_tokens": 203301646.0, + "step": 7923 + }, + { + "epoch": 0.8701954755106523, + "grad_norm": 2.278231620788574, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7236536741256714, + "num_tokens": 203324990.0, + "step": 7924 + }, + { + "epoch": 0.870305293213266, + "grad_norm": 2.12359356880188, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.6854338645935059, + "num_tokens": 203351801.0, + "step": 7925 + }, + { + "epoch": 0.8704151109158796, + "grad_norm": 2.350543260574341, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.726906418800354, + "num_tokens": 203375388.0, + "step": 7926 + }, + { + "epoch": 0.8705249286184933, + "grad_norm": 2.4178078174591064, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.7031792402267456, + "num_tokens": 203399508.0, + "step": 7927 + }, + { + "epoch": 0.8706347463211069, + "grad_norm": 2.0211341381073, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6973936557769775, + "num_tokens": 203428974.0, + "step": 7928 + }, + { + "epoch": 0.8707445640237206, + "grad_norm": 2.336500883102417, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7060056328773499, + "num_tokens": 203450417.0, + "step": 7929 + }, + { + "epoch": 0.8708543817263343, + "grad_norm": 2.349533796310425, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7206103801727295, + "num_tokens": 203473364.0, + "step": 7930 + }, + { + "epoch": 0.870964199428948, + "grad_norm": 2.0679314136505127, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.710918664932251, + "num_tokens": 203500758.0, + "step": 7931 + }, + { + "epoch": 0.8710740171315616, + "grad_norm": 2.4690072536468506, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7219817638397217, + "num_tokens": 203520774.0, + "step": 7932 + }, + { + "epoch": 0.8711838348341753, + "grad_norm": 1.9823167324066162, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7103492617607117, + "num_tokens": 203553594.0, + "step": 7933 + }, + { + "epoch": 0.8712936525367889, + "grad_norm": 2.421919822692871, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7262908220291138, + "num_tokens": 203574115.0, + "step": 7934 + }, + { + "epoch": 0.8714034702394026, + "grad_norm": 2.123060703277588, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7160454988479614, + "num_tokens": 203601238.0, + "step": 7935 + }, + { + "epoch": 0.8715132879420162, + "grad_norm": 2.053509473800659, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6970325708389282, + "num_tokens": 203631349.0, + "step": 7936 + }, + { + "epoch": 0.87162310564463, + "grad_norm": 2.3431427478790283, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7143545150756836, + "num_tokens": 203653469.0, + "step": 7937 + }, + { + "epoch": 0.8717329233472436, + "grad_norm": 2.6350088119506836, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7110718488693237, + "num_tokens": 203672167.0, + "step": 7938 + }, + { + "epoch": 0.8718427410498573, + "grad_norm": 2.0077600479125977, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7165588140487671, + "num_tokens": 203701870.0, + "step": 7939 + }, + { + "epoch": 0.8719525587524709, + "grad_norm": 2.1377811431884766, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7230890393257141, + "num_tokens": 203727249.0, + "step": 7940 + }, + { + "epoch": 0.8720623764550846, + "grad_norm": 2.2313246726989746, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6948900818824768, + "num_tokens": 203753024.0, + "step": 7941 + }, + { + "epoch": 0.8721721941576982, + "grad_norm": 2.296816349029541, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6954115033149719, + "num_tokens": 203778303.0, + "step": 7942 + }, + { + "epoch": 0.8722820118603118, + "grad_norm": 2.31711483001709, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7096763253211975, + "num_tokens": 203799333.0, + "step": 7943 + }, + { + "epoch": 0.8723918295629255, + "grad_norm": 2.0713348388671875, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.722868800163269, + "num_tokens": 203824592.0, + "step": 7944 + }, + { + "epoch": 0.8725016472655392, + "grad_norm": 2.064296007156372, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6978610157966614, + "num_tokens": 203853442.0, + "step": 7945 + }, + { + "epoch": 0.8726114649681529, + "grad_norm": 2.0276777744293213, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.6977962255477905, + "num_tokens": 203883138.0, + "step": 7946 + }, + { + "epoch": 0.8727212826707665, + "grad_norm": 2.1222009658813477, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6865204572677612, + "num_tokens": 203911283.0, + "step": 7947 + }, + { + "epoch": 0.8728311003733802, + "grad_norm": 2.318192958831787, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7156702280044556, + "num_tokens": 203933292.0, + "step": 7948 + }, + { + "epoch": 0.8729409180759938, + "grad_norm": 2.1548924446105957, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7132562398910522, + "num_tokens": 203959649.0, + "step": 7949 + }, + { + "epoch": 0.8730507357786075, + "grad_norm": 2.115194082260132, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6993362307548523, + "num_tokens": 203988697.0, + "step": 7950 + }, + { + "epoch": 0.8731605534812211, + "grad_norm": 2.4317288398742676, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7136850953102112, + "num_tokens": 204009832.0, + "step": 7951 + }, + { + "epoch": 0.8732703711838349, + "grad_norm": 1.9814016819000244, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7046705484390259, + "num_tokens": 204039407.0, + "step": 7952 + }, + { + "epoch": 0.8733801888864485, + "grad_norm": 1.989896535873413, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6994255781173706, + "num_tokens": 204070626.0, + "step": 7953 + }, + { + "epoch": 0.8734900065890622, + "grad_norm": 2.2801969051361084, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6922917366027832, + "num_tokens": 204095068.0, + "step": 7954 + }, + { + "epoch": 0.8735998242916758, + "grad_norm": 2.210905075073242, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7014460563659668, + "num_tokens": 204119963.0, + "step": 7955 + }, + { + "epoch": 0.8737096419942895, + "grad_norm": 2.022158145904541, + "learning_rate": 1e-06, + "loss": 1.0504, + "mean_token_accuracy": 0.6850137114524841, + "num_tokens": 204150693.0, + "step": 7956 + }, + { + "epoch": 0.8738194596969031, + "grad_norm": 2.35897159576416, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6915732622146606, + "num_tokens": 204174879.0, + "step": 7957 + }, + { + "epoch": 0.8739292773995168, + "grad_norm": 1.9946563243865967, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7146549224853516, + "num_tokens": 204202807.0, + "step": 7958 + }, + { + "epoch": 0.8740390951021305, + "grad_norm": 1.908065676689148, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7092697620391846, + "num_tokens": 204233749.0, + "step": 7959 + }, + { + "epoch": 0.8741489128047442, + "grad_norm": 2.083827018737793, + "learning_rate": 1e-06, + "loss": 1.0847, + "mean_token_accuracy": 0.6778532862663269, + "num_tokens": 204263617.0, + "step": 7960 + }, + { + "epoch": 0.8742587305073578, + "grad_norm": 2.2772717475891113, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7250525951385498, + "num_tokens": 204286958.0, + "step": 7961 + }, + { + "epoch": 0.8743685482099715, + "grad_norm": 2.0854368209838867, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7094532251358032, + "num_tokens": 204315083.0, + "step": 7962 + }, + { + "epoch": 0.8744783659125851, + "grad_norm": 2.0587894916534424, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.702414333820343, + "num_tokens": 204341947.0, + "step": 7963 + }, + { + "epoch": 0.8745881836151987, + "grad_norm": 2.619321584701538, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.719602644443512, + "num_tokens": 204362733.0, + "step": 7964 + }, + { + "epoch": 0.8746980013178124, + "grad_norm": 2.149141550064087, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7238514423370361, + "num_tokens": 204387571.0, + "step": 7965 + }, + { + "epoch": 0.8748078190204261, + "grad_norm": 2.080037832260132, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7121984958648682, + "num_tokens": 204414927.0, + "step": 7966 + }, + { + "epoch": 0.8749176367230398, + "grad_norm": 2.1616079807281494, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7035500407218933, + "num_tokens": 204439859.0, + "step": 7967 + }, + { + "epoch": 0.8750274544256534, + "grad_norm": 1.9658622741699219, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7132005095481873, + "num_tokens": 204472652.0, + "step": 7968 + }, + { + "epoch": 0.8751372721282671, + "grad_norm": 2.357778549194336, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7192503809928894, + "num_tokens": 204493805.0, + "step": 7969 + }, + { + "epoch": 0.8752470898308807, + "grad_norm": 2.4395158290863037, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7243051528930664, + "num_tokens": 204514084.0, + "step": 7970 + }, + { + "epoch": 0.8753569075334944, + "grad_norm": 2.1683921813964844, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.716925859451294, + "num_tokens": 204541107.0, + "step": 7971 + }, + { + "epoch": 0.875466725236108, + "grad_norm": 2.230135202407837, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6976258754730225, + "num_tokens": 204568605.0, + "step": 7972 + }, + { + "epoch": 0.8755765429387217, + "grad_norm": 2.0030980110168457, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7069292068481445, + "num_tokens": 204601206.0, + "step": 7973 + }, + { + "epoch": 0.8756863606413354, + "grad_norm": 2.1927602291107178, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7018280029296875, + "num_tokens": 204628586.0, + "step": 7974 + }, + { + "epoch": 0.8757961783439491, + "grad_norm": 2.2084150314331055, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6886420249938965, + "num_tokens": 204653413.0, + "step": 7975 + }, + { + "epoch": 0.8759059960465627, + "grad_norm": 2.06126070022583, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7220488786697388, + "num_tokens": 204681422.0, + "step": 7976 + }, + { + "epoch": 0.8760158137491764, + "grad_norm": 2.3099441528320312, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7093778848648071, + "num_tokens": 204705540.0, + "step": 7977 + }, + { + "epoch": 0.87612563145179, + "grad_norm": 2.4287617206573486, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7173053026199341, + "num_tokens": 204725742.0, + "step": 7978 + }, + { + "epoch": 0.8762354491544037, + "grad_norm": 2.2888364791870117, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6915486454963684, + "num_tokens": 204749791.0, + "step": 7979 + }, + { + "epoch": 0.8763452668570173, + "grad_norm": 2.3009302616119385, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7208008170127869, + "num_tokens": 204776526.0, + "step": 7980 + }, + { + "epoch": 0.8764550845596311, + "grad_norm": 2.0541086196899414, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6861138939857483, + "num_tokens": 204805486.0, + "step": 7981 + }, + { + "epoch": 0.8765649022622447, + "grad_norm": 2.196314811706543, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7247604727745056, + "num_tokens": 204829903.0, + "step": 7982 + }, + { + "epoch": 0.8766747199648584, + "grad_norm": 2.058185338973999, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7121132612228394, + "num_tokens": 204858879.0, + "step": 7983 + }, + { + "epoch": 0.876784537667472, + "grad_norm": 2.2862801551818848, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7004377245903015, + "num_tokens": 204881590.0, + "step": 7984 + }, + { + "epoch": 0.8768943553700856, + "grad_norm": 1.9763026237487793, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6930891871452332, + "num_tokens": 204912075.0, + "step": 7985 + }, + { + "epoch": 0.8770041730726993, + "grad_norm": 2.0869598388671875, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7066543102264404, + "num_tokens": 204937564.0, + "step": 7986 + }, + { + "epoch": 0.8771139907753129, + "grad_norm": 2.3880462646484375, + "learning_rate": 1e-06, + "loss": 1.0662, + "mean_token_accuracy": 0.6920490264892578, + "num_tokens": 204961594.0, + "step": 7987 + }, + { + "epoch": 0.8772238084779267, + "grad_norm": 1.9992963075637817, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6934962868690491, + "num_tokens": 204991641.0, + "step": 7988 + }, + { + "epoch": 0.8773336261805403, + "grad_norm": 1.9650946855545044, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6828591823577881, + "num_tokens": 205023052.0, + "step": 7989 + }, + { + "epoch": 0.877443443883154, + "grad_norm": 2.3076460361480713, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7163859605789185, + "num_tokens": 205045081.0, + "step": 7990 + }, + { + "epoch": 0.8775532615857676, + "grad_norm": 2.2705256938934326, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6971456408500671, + "num_tokens": 205067771.0, + "step": 7991 + }, + { + "epoch": 0.8776630792883813, + "grad_norm": 2.236100912094116, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7122327089309692, + "num_tokens": 205093169.0, + "step": 7992 + }, + { + "epoch": 0.8777728969909949, + "grad_norm": 2.123547315597534, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7271196842193604, + "num_tokens": 205117177.0, + "step": 7993 + }, + { + "epoch": 0.8778827146936086, + "grad_norm": 1.9446241855621338, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.67939293384552, + "num_tokens": 205152099.0, + "step": 7994 + }, + { + "epoch": 0.8779925323962223, + "grad_norm": 2.246462345123291, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.696350634098053, + "num_tokens": 205177162.0, + "step": 7995 + }, + { + "epoch": 0.878102350098836, + "grad_norm": 1.943043828010559, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6904306411743164, + "num_tokens": 205210981.0, + "step": 7996 + }, + { + "epoch": 0.8782121678014496, + "grad_norm": 2.2042429447174072, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.716513991355896, + "num_tokens": 205235182.0, + "step": 7997 + }, + { + "epoch": 0.8783219855040633, + "grad_norm": 2.1513917446136475, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6905355453491211, + "num_tokens": 205259985.0, + "step": 7998 + }, + { + "epoch": 0.8784318032066769, + "grad_norm": 2.0910756587982178, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6910033226013184, + "num_tokens": 205289232.0, + "step": 7999 + }, + { + "epoch": 0.8785416209092906, + "grad_norm": 2.3838212490081787, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.721872866153717, + "num_tokens": 205309332.0, + "step": 8000 + }, + { + "epoch": 0.8786514386119042, + "grad_norm": 2.019505500793457, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6972450017929077, + "num_tokens": 205339109.0, + "step": 8001 + }, + { + "epoch": 0.8787612563145178, + "grad_norm": 2.274437189102173, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7161060571670532, + "num_tokens": 205361956.0, + "step": 8002 + }, + { + "epoch": 0.8788710740171316, + "grad_norm": 2.509578227996826, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7137082815170288, + "num_tokens": 205381806.0, + "step": 8003 + }, + { + "epoch": 0.8789808917197452, + "grad_norm": 2.0620782375335693, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6846162676811218, + "num_tokens": 205411255.0, + "step": 8004 + }, + { + "epoch": 0.8790907094223589, + "grad_norm": 2.145266056060791, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7213670015335083, + "num_tokens": 205436064.0, + "step": 8005 + }, + { + "epoch": 0.8792005271249725, + "grad_norm": 2.5813114643096924, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7374969720840454, + "num_tokens": 205453876.0, + "step": 8006 + }, + { + "epoch": 0.8793103448275862, + "grad_norm": 2.280360460281372, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6963983774185181, + "num_tokens": 205477794.0, + "step": 8007 + }, + { + "epoch": 0.8794201625301998, + "grad_norm": 2.1116297245025635, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7123669385910034, + "num_tokens": 205503954.0, + "step": 8008 + }, + { + "epoch": 0.8795299802328135, + "grad_norm": 2.1277594566345215, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7380701899528503, + "num_tokens": 205528284.0, + "step": 8009 + }, + { + "epoch": 0.8796397979354272, + "grad_norm": 2.086820125579834, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7118127346038818, + "num_tokens": 205554124.0, + "step": 8010 + }, + { + "epoch": 0.8797496156380409, + "grad_norm": 2.311861515045166, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7339587211608887, + "num_tokens": 205577083.0, + "step": 8011 + }, + { + "epoch": 0.8798594333406545, + "grad_norm": 2.185218334197998, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7228816747665405, + "num_tokens": 205604068.0, + "step": 8012 + }, + { + "epoch": 0.8799692510432682, + "grad_norm": 2.2240469455718994, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7071549296379089, + "num_tokens": 205627960.0, + "step": 8013 + }, + { + "epoch": 0.8800790687458818, + "grad_norm": 2.129584789276123, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7087458372116089, + "num_tokens": 205656207.0, + "step": 8014 + }, + { + "epoch": 0.8801888864484955, + "grad_norm": 2.1892635822296143, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.7013278007507324, + "num_tokens": 205680872.0, + "step": 8015 + }, + { + "epoch": 0.8802987041511091, + "grad_norm": 2.0174481868743896, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7026631832122803, + "num_tokens": 205709531.0, + "step": 8016 + }, + { + "epoch": 0.8804085218537229, + "grad_norm": 2.5496675968170166, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7171427011489868, + "num_tokens": 205730541.0, + "step": 8017 + }, + { + "epoch": 0.8805183395563365, + "grad_norm": 2.160231828689575, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6935354471206665, + "num_tokens": 205755442.0, + "step": 8018 + }, + { + "epoch": 0.8806281572589502, + "grad_norm": 2.200995922088623, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7056608200073242, + "num_tokens": 205780824.0, + "step": 8019 + }, + { + "epoch": 0.8807379749615638, + "grad_norm": 2.0154943466186523, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7117273807525635, + "num_tokens": 205809950.0, + "step": 8020 + }, + { + "epoch": 0.8808477926641775, + "grad_norm": 2.1876332759857178, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7092798948287964, + "num_tokens": 205833344.0, + "step": 8021 + }, + { + "epoch": 0.8809576103667911, + "grad_norm": 2.4069554805755615, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7220321297645569, + "num_tokens": 205854927.0, + "step": 8022 + }, + { + "epoch": 0.8810674280694047, + "grad_norm": 2.115570306777954, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7079128623008728, + "num_tokens": 205882449.0, + "step": 8023 + }, + { + "epoch": 0.8811772457720185, + "grad_norm": 2.0098471641540527, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7005916833877563, + "num_tokens": 205912075.0, + "step": 8024 + }, + { + "epoch": 0.8812870634746321, + "grad_norm": 2.048879861831665, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7151968479156494, + "num_tokens": 205940796.0, + "step": 8025 + }, + { + "epoch": 0.8813968811772458, + "grad_norm": 2.2546985149383545, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7148810625076294, + "num_tokens": 205965018.0, + "step": 8026 + }, + { + "epoch": 0.8815066988798594, + "grad_norm": 1.8937971591949463, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6977487802505493, + "num_tokens": 205995038.0, + "step": 8027 + }, + { + "epoch": 0.8816165165824731, + "grad_norm": 2.089611530303955, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7200796008110046, + "num_tokens": 206021234.0, + "step": 8028 + }, + { + "epoch": 0.8817263342850867, + "grad_norm": 2.223339319229126, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7117663621902466, + "num_tokens": 206045572.0, + "step": 8029 + }, + { + "epoch": 0.8818361519877004, + "grad_norm": 1.935314416885376, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7160618305206299, + "num_tokens": 206075113.0, + "step": 8030 + }, + { + "epoch": 0.881945969690314, + "grad_norm": 2.2831828594207764, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.721515953540802, + "num_tokens": 206097073.0, + "step": 8031 + }, + { + "epoch": 0.8820557873929278, + "grad_norm": 2.6930317878723145, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7198349833488464, + "num_tokens": 206115723.0, + "step": 8032 + }, + { + "epoch": 0.8821656050955414, + "grad_norm": 2.082611560821533, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7237500548362732, + "num_tokens": 206142968.0, + "step": 8033 + }, + { + "epoch": 0.8822754227981551, + "grad_norm": 2.2968850135803223, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.721280574798584, + "num_tokens": 206166150.0, + "step": 8034 + }, + { + "epoch": 0.8823852405007687, + "grad_norm": 2.198613166809082, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7180631756782532, + "num_tokens": 206189702.0, + "step": 8035 + }, + { + "epoch": 0.8824950582033824, + "grad_norm": 2.3601698875427246, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7045073509216309, + "num_tokens": 206211122.0, + "step": 8036 + }, + { + "epoch": 0.882604875905996, + "grad_norm": 1.8862888813018799, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7067040801048279, + "num_tokens": 206242527.0, + "step": 8037 + }, + { + "epoch": 0.8827146936086097, + "grad_norm": 2.251581907272339, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.704670250415802, + "num_tokens": 206265003.0, + "step": 8038 + }, + { + "epoch": 0.8828245113112234, + "grad_norm": 2.0095787048339844, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7024259567260742, + "num_tokens": 206294691.0, + "step": 8039 + }, + { + "epoch": 0.8829343290138371, + "grad_norm": 2.085545063018799, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7121666669845581, + "num_tokens": 206322117.0, + "step": 8040 + }, + { + "epoch": 0.8830441467164507, + "grad_norm": 2.0405497550964355, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7135499715805054, + "num_tokens": 206349581.0, + "step": 8041 + }, + { + "epoch": 0.8831539644190644, + "grad_norm": 2.0894558429718018, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.691370964050293, + "num_tokens": 206378214.0, + "step": 8042 + }, + { + "epoch": 0.883263782121678, + "grad_norm": 2.3337981700897217, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7215582728385925, + "num_tokens": 206399863.0, + "step": 8043 + }, + { + "epoch": 0.8833735998242916, + "grad_norm": 2.0815882682800293, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7237052917480469, + "num_tokens": 206425352.0, + "step": 8044 + }, + { + "epoch": 0.8834834175269053, + "grad_norm": 2.2754180431365967, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6949099898338318, + "num_tokens": 206449331.0, + "step": 8045 + }, + { + "epoch": 0.883593235229519, + "grad_norm": 2.4522554874420166, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7090516090393066, + "num_tokens": 206473133.0, + "step": 8046 + }, + { + "epoch": 0.8837030529321327, + "grad_norm": 2.106889486312866, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7028380632400513, + "num_tokens": 206497702.0, + "step": 8047 + }, + { + "epoch": 0.8838128706347463, + "grad_norm": 1.9450691938400269, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7249284982681274, + "num_tokens": 206527793.0, + "step": 8048 + }, + { + "epoch": 0.88392268833736, + "grad_norm": 2.335374116897583, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7004563808441162, + "num_tokens": 206551976.0, + "step": 8049 + }, + { + "epoch": 0.8840325060399736, + "grad_norm": 2.0889461040496826, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7093929052352905, + "num_tokens": 206578294.0, + "step": 8050 + }, + { + "epoch": 0.8841423237425873, + "grad_norm": 2.438530445098877, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7427464127540588, + "num_tokens": 206597714.0, + "step": 8051 + }, + { + "epoch": 0.8842521414452009, + "grad_norm": 2.0608303546905518, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7186055183410645, + "num_tokens": 206624115.0, + "step": 8052 + }, + { + "epoch": 0.8843619591478147, + "grad_norm": 2.150308132171631, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6991763710975647, + "num_tokens": 206649977.0, + "step": 8053 + }, + { + "epoch": 0.8844717768504283, + "grad_norm": 2.2845354080200195, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7111310362815857, + "num_tokens": 206674931.0, + "step": 8054 + }, + { + "epoch": 0.884581594553042, + "grad_norm": 2.3049333095550537, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.704054594039917, + "num_tokens": 206698055.0, + "step": 8055 + }, + { + "epoch": 0.8846914122556556, + "grad_norm": 2.1976428031921387, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7173043489456177, + "num_tokens": 206722595.0, + "step": 8056 + }, + { + "epoch": 0.8848012299582693, + "grad_norm": 2.2121670246124268, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7047535181045532, + "num_tokens": 206748220.0, + "step": 8057 + }, + { + "epoch": 0.8849110476608829, + "grad_norm": 1.7763140201568604, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.698490560054779, + "num_tokens": 206784407.0, + "step": 8058 + }, + { + "epoch": 0.8850208653634966, + "grad_norm": 2.244858503341675, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7068261504173279, + "num_tokens": 206809142.0, + "step": 8059 + }, + { + "epoch": 0.8851306830661102, + "grad_norm": 1.9743516445159912, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.689573347568512, + "num_tokens": 206839851.0, + "step": 8060 + }, + { + "epoch": 0.885240500768724, + "grad_norm": 2.313706874847412, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6933872699737549, + "num_tokens": 206864757.0, + "step": 8061 + }, + { + "epoch": 0.8853503184713376, + "grad_norm": 2.08778715133667, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7043870687484741, + "num_tokens": 206892533.0, + "step": 8062 + }, + { + "epoch": 0.8854601361739513, + "grad_norm": 1.92947518825531, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7152218818664551, + "num_tokens": 206921405.0, + "step": 8063 + }, + { + "epoch": 0.8855699538765649, + "grad_norm": 2.161818742752075, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7117953300476074, + "num_tokens": 206947558.0, + "step": 8064 + }, + { + "epoch": 0.8856797715791785, + "grad_norm": 2.15616512298584, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6840884685516357, + "num_tokens": 206974321.0, + "step": 8065 + }, + { + "epoch": 0.8857895892817922, + "grad_norm": 2.0985984802246094, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7129332423210144, + "num_tokens": 207000866.0, + "step": 8066 + }, + { + "epoch": 0.8858994069844058, + "grad_norm": 2.0532073974609375, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7013360261917114, + "num_tokens": 207029175.0, + "step": 8067 + }, + { + "epoch": 0.8860092246870196, + "grad_norm": 2.235734224319458, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7354335784912109, + "num_tokens": 207052903.0, + "step": 8068 + }, + { + "epoch": 0.8861190423896332, + "grad_norm": 2.585806369781494, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7021169662475586, + "num_tokens": 207073819.0, + "step": 8069 + }, + { + "epoch": 0.8862288600922469, + "grad_norm": 2.1656036376953125, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7208548784255981, + "num_tokens": 207098602.0, + "step": 8070 + }, + { + "epoch": 0.8863386777948605, + "grad_norm": 2.1806886196136475, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7374809980392456, + "num_tokens": 207125162.0, + "step": 8071 + }, + { + "epoch": 0.8864484954974742, + "grad_norm": 2.494764804840088, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7103517055511475, + "num_tokens": 207145772.0, + "step": 8072 + }, + { + "epoch": 0.8865583132000878, + "grad_norm": 2.086733102798462, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7229043245315552, + "num_tokens": 207173246.0, + "step": 8073 + }, + { + "epoch": 0.8866681309027015, + "grad_norm": 2.136589288711548, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7334898710250854, + "num_tokens": 207200099.0, + "step": 8074 + }, + { + "epoch": 0.8867779486053152, + "grad_norm": 2.130373954772949, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7035504579544067, + "num_tokens": 207227105.0, + "step": 8075 + }, + { + "epoch": 0.8868877663079289, + "grad_norm": 2.0889813899993896, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7281883358955383, + "num_tokens": 207252276.0, + "step": 8076 + }, + { + "epoch": 0.8869975840105425, + "grad_norm": 2.3181374073028564, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.702275276184082, + "num_tokens": 207276505.0, + "step": 8077 + }, + { + "epoch": 0.8871074017131562, + "grad_norm": 1.9882885217666626, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6950638890266418, + "num_tokens": 207307280.0, + "step": 8078 + }, + { + "epoch": 0.8872172194157698, + "grad_norm": 2.4376423358917236, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7084583640098572, + "num_tokens": 207328545.0, + "step": 8079 + }, + { + "epoch": 0.8873270371183835, + "grad_norm": 2.3807432651519775, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7007606029510498, + "num_tokens": 207348994.0, + "step": 8080 + }, + { + "epoch": 0.8874368548209971, + "grad_norm": 2.2885794639587402, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6764592528343201, + "num_tokens": 207372347.0, + "step": 8081 + }, + { + "epoch": 0.8875466725236109, + "grad_norm": 2.4877214431762695, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7117917537689209, + "num_tokens": 207392622.0, + "step": 8082 + }, + { + "epoch": 0.8876564902262245, + "grad_norm": 2.3419225215911865, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7075655460357666, + "num_tokens": 207415253.0, + "step": 8083 + }, + { + "epoch": 0.8877663079288381, + "grad_norm": 2.1127796173095703, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7065281271934509, + "num_tokens": 207442012.0, + "step": 8084 + }, + { + "epoch": 0.8878761256314518, + "grad_norm": 2.268547773361206, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7077305316925049, + "num_tokens": 207465415.0, + "step": 8085 + }, + { + "epoch": 0.8879859433340654, + "grad_norm": 2.208395481109619, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7108168601989746, + "num_tokens": 207491113.0, + "step": 8086 + }, + { + "epoch": 0.8880957610366791, + "grad_norm": 2.097510576248169, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6925868988037109, + "num_tokens": 207519573.0, + "step": 8087 + }, + { + "epoch": 0.8882055787392927, + "grad_norm": 2.2706098556518555, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6997939348220825, + "num_tokens": 207546307.0, + "step": 8088 + }, + { + "epoch": 0.8883153964419065, + "grad_norm": 2.3470356464385986, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7135905623435974, + "num_tokens": 207568818.0, + "step": 8089 + }, + { + "epoch": 0.8884252141445201, + "grad_norm": 2.008190870285034, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6988404393196106, + "num_tokens": 207600375.0, + "step": 8090 + }, + { + "epoch": 0.8885350318471338, + "grad_norm": 1.7826086282730103, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.684928297996521, + "num_tokens": 207635848.0, + "step": 8091 + }, + { + "epoch": 0.8886448495497474, + "grad_norm": 2.137742519378662, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7030997276306152, + "num_tokens": 207662247.0, + "step": 8092 + }, + { + "epoch": 0.8887546672523611, + "grad_norm": 2.165987968444824, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7017512321472168, + "num_tokens": 207690173.0, + "step": 8093 + }, + { + "epoch": 0.8888644849549747, + "grad_norm": 1.9383519887924194, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7138475179672241, + "num_tokens": 207721096.0, + "step": 8094 + }, + { + "epoch": 0.8889743026575884, + "grad_norm": 2.038194179534912, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7040780782699585, + "num_tokens": 207750109.0, + "step": 8095 + }, + { + "epoch": 0.889084120360202, + "grad_norm": 2.263695478439331, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7136959433555603, + "num_tokens": 207774593.0, + "step": 8096 + }, + { + "epoch": 0.8891939380628158, + "grad_norm": 2.1195733547210693, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7185425162315369, + "num_tokens": 207800823.0, + "step": 8097 + }, + { + "epoch": 0.8893037557654294, + "grad_norm": 2.2154765129089355, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6915194392204285, + "num_tokens": 207827232.0, + "step": 8098 + }, + { + "epoch": 0.8894135734680431, + "grad_norm": 2.1596322059631348, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7457786798477173, + "num_tokens": 207851230.0, + "step": 8099 + }, + { + "epoch": 0.8895233911706567, + "grad_norm": 1.9953473806381226, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.682414710521698, + "num_tokens": 207885581.0, + "step": 8100 + }, + { + "epoch": 0.8896332088732704, + "grad_norm": 2.0357537269592285, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.7010218501091003, + "num_tokens": 207913459.0, + "step": 8101 + }, + { + "epoch": 0.889743026575884, + "grad_norm": 2.3560404777526855, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6941590309143066, + "num_tokens": 207937259.0, + "step": 8102 + }, + { + "epoch": 0.8898528442784976, + "grad_norm": 1.9187678098678589, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6930727958679199, + "num_tokens": 207970606.0, + "step": 8103 + }, + { + "epoch": 0.8899626619811114, + "grad_norm": 2.244650363922119, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.70298832654953, + "num_tokens": 207995082.0, + "step": 8104 + }, + { + "epoch": 0.890072479683725, + "grad_norm": 2.0677919387817383, + "learning_rate": 1e-06, + "loss": 1.0799, + "mean_token_accuracy": 0.6757185459136963, + "num_tokens": 208025822.0, + "step": 8105 + }, + { + "epoch": 0.8901822973863387, + "grad_norm": 1.9279814958572388, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6978937983512878, + "num_tokens": 208054059.0, + "step": 8106 + }, + { + "epoch": 0.8902921150889523, + "grad_norm": 2.2185842990875244, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7125483751296997, + "num_tokens": 208079371.0, + "step": 8107 + }, + { + "epoch": 0.890401932791566, + "grad_norm": 2.5918993949890137, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7216417789459229, + "num_tokens": 208097718.0, + "step": 8108 + }, + { + "epoch": 0.8905117504941796, + "grad_norm": 1.924234390258789, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6906692981719971, + "num_tokens": 208128275.0, + "step": 8109 + }, + { + "epoch": 0.8906215681967933, + "grad_norm": 2.1346325874328613, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.708966851234436, + "num_tokens": 208154634.0, + "step": 8110 + }, + { + "epoch": 0.890731385899407, + "grad_norm": 2.3461830615997314, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7027347087860107, + "num_tokens": 208179815.0, + "step": 8111 + }, + { + "epoch": 0.8908412036020207, + "grad_norm": 2.067674398422241, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7266103029251099, + "num_tokens": 208206206.0, + "step": 8112 + }, + { + "epoch": 0.8909510213046343, + "grad_norm": 2.5368402004241943, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7194215059280396, + "num_tokens": 208225889.0, + "step": 8113 + }, + { + "epoch": 0.891060839007248, + "grad_norm": 2.5531845092773438, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7190164923667908, + "num_tokens": 208245354.0, + "step": 8114 + }, + { + "epoch": 0.8911706567098616, + "grad_norm": 1.854310154914856, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6962159872055054, + "num_tokens": 208277472.0, + "step": 8115 + }, + { + "epoch": 0.8912804744124753, + "grad_norm": 1.9592509269714355, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7332763671875, + "num_tokens": 208305344.0, + "step": 8116 + }, + { + "epoch": 0.8913902921150889, + "grad_norm": 2.3007497787475586, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7301367521286011, + "num_tokens": 208327131.0, + "step": 8117 + }, + { + "epoch": 0.8915001098177027, + "grad_norm": 2.2423601150512695, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.696884274482727, + "num_tokens": 208350940.0, + "step": 8118 + }, + { + "epoch": 0.8916099275203163, + "grad_norm": 2.1989057064056396, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7160118818283081, + "num_tokens": 208377255.0, + "step": 8119 + }, + { + "epoch": 0.89171974522293, + "grad_norm": 2.1226584911346436, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7169994115829468, + "num_tokens": 208404723.0, + "step": 8120 + }, + { + "epoch": 0.8918295629255436, + "grad_norm": 2.001877546310425, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7021502256393433, + "num_tokens": 208434859.0, + "step": 8121 + }, + { + "epoch": 0.8919393806281573, + "grad_norm": 2.06687068939209, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.7047120928764343, + "num_tokens": 208462336.0, + "step": 8122 + }, + { + "epoch": 0.8920491983307709, + "grad_norm": 2.2102792263031006, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7006993293762207, + "num_tokens": 208486291.0, + "step": 8123 + }, + { + "epoch": 0.8921590160333845, + "grad_norm": 2.5624020099639893, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7126317024230957, + "num_tokens": 208504584.0, + "step": 8124 + }, + { + "epoch": 0.8922688337359982, + "grad_norm": 2.133267879486084, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7282863855361938, + "num_tokens": 208530581.0, + "step": 8125 + }, + { + "epoch": 0.892378651438612, + "grad_norm": 2.312623977661133, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7125049233436584, + "num_tokens": 208552623.0, + "step": 8126 + }, + { + "epoch": 0.8924884691412256, + "grad_norm": 2.1883156299591064, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7194758653640747, + "num_tokens": 208575940.0, + "step": 8127 + }, + { + "epoch": 0.8925982868438392, + "grad_norm": 2.411821126937866, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7180913686752319, + "num_tokens": 208597298.0, + "step": 8128 + }, + { + "epoch": 0.8927081045464529, + "grad_norm": 2.4700655937194824, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7404309511184692, + "num_tokens": 208616812.0, + "step": 8129 + }, + { + "epoch": 0.8928179222490665, + "grad_norm": 2.0331637859344482, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7119889259338379, + "num_tokens": 208643882.0, + "step": 8130 + }, + { + "epoch": 0.8929277399516802, + "grad_norm": 2.5637314319610596, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6998552680015564, + "num_tokens": 208664381.0, + "step": 8131 + }, + { + "epoch": 0.8930375576542938, + "grad_norm": 1.761197566986084, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7040706872940063, + "num_tokens": 208703328.0, + "step": 8132 + }, + { + "epoch": 0.8931473753569076, + "grad_norm": 2.399379253387451, + "learning_rate": 1e-06, + "loss": 1.0966, + "mean_token_accuracy": 0.6744287014007568, + "num_tokens": 208729806.0, + "step": 8133 + }, + { + "epoch": 0.8932571930595212, + "grad_norm": 2.0725722312927246, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6822391748428345, + "num_tokens": 208759036.0, + "step": 8134 + }, + { + "epoch": 0.8933670107621349, + "grad_norm": 2.0877037048339844, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.683993935585022, + "num_tokens": 208786913.0, + "step": 8135 + }, + { + "epoch": 0.8934768284647485, + "grad_norm": 2.2243077754974365, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7183749079704285, + "num_tokens": 208812219.0, + "step": 8136 + }, + { + "epoch": 0.8935866461673622, + "grad_norm": 2.0771400928497314, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7077346444129944, + "num_tokens": 208840314.0, + "step": 8137 + }, + { + "epoch": 0.8936964638699758, + "grad_norm": 2.063333511352539, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7187254428863525, + "num_tokens": 208869984.0, + "step": 8138 + }, + { + "epoch": 0.8938062815725895, + "grad_norm": 2.08547306060791, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7031029462814331, + "num_tokens": 208897953.0, + "step": 8139 + }, + { + "epoch": 0.8939160992752032, + "grad_norm": 2.3768362998962402, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.71096271276474, + "num_tokens": 208921006.0, + "step": 8140 + }, + { + "epoch": 0.8940259169778169, + "grad_norm": 2.264031171798706, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6986809968948364, + "num_tokens": 208944524.0, + "step": 8141 + }, + { + "epoch": 0.8941357346804305, + "grad_norm": 2.3631951808929443, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7071967124938965, + "num_tokens": 208966169.0, + "step": 8142 + }, + { + "epoch": 0.8942455523830442, + "grad_norm": 2.416294813156128, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7069035768508911, + "num_tokens": 208988724.0, + "step": 8143 + }, + { + "epoch": 0.8943553700856578, + "grad_norm": 1.8338149785995483, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6904369592666626, + "num_tokens": 209025001.0, + "step": 8144 + }, + { + "epoch": 0.8944651877882714, + "grad_norm": 1.8867404460906982, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6947872042655945, + "num_tokens": 209057433.0, + "step": 8145 + }, + { + "epoch": 0.8945750054908851, + "grad_norm": 1.95625901222229, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7167205810546875, + "num_tokens": 209086169.0, + "step": 8146 + }, + { + "epoch": 0.8946848231934988, + "grad_norm": 2.131828784942627, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6822386384010315, + "num_tokens": 209116586.0, + "step": 8147 + }, + { + "epoch": 0.8947946408961125, + "grad_norm": 2.434034585952759, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7157453894615173, + "num_tokens": 209137575.0, + "step": 8148 + }, + { + "epoch": 0.8949044585987261, + "grad_norm": 2.6567840576171875, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7009516358375549, + "num_tokens": 209155725.0, + "step": 8149 + }, + { + "epoch": 0.8950142763013398, + "grad_norm": 2.2768235206604004, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7043923139572144, + "num_tokens": 209179544.0, + "step": 8150 + }, + { + "epoch": 0.8951240940039534, + "grad_norm": 2.057263135910034, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7055954933166504, + "num_tokens": 209207120.0, + "step": 8151 + }, + { + "epoch": 0.8952339117065671, + "grad_norm": 2.1695728302001953, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7080497145652771, + "num_tokens": 209234360.0, + "step": 8152 + }, + { + "epoch": 0.8953437294091807, + "grad_norm": 2.0607244968414307, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7111339569091797, + "num_tokens": 209263059.0, + "step": 8153 + }, + { + "epoch": 0.8954535471117944, + "grad_norm": 2.084826707839966, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7112028002738953, + "num_tokens": 209289722.0, + "step": 8154 + }, + { + "epoch": 0.8955633648144081, + "grad_norm": 2.053248167037964, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.711506187915802, + "num_tokens": 209315138.0, + "step": 8155 + }, + { + "epoch": 0.8956731825170218, + "grad_norm": 2.1133859157562256, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7249512076377869, + "num_tokens": 209339446.0, + "step": 8156 + }, + { + "epoch": 0.8957830002196354, + "grad_norm": 2.1427974700927734, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7133527994155884, + "num_tokens": 209365437.0, + "step": 8157 + }, + { + "epoch": 0.8958928179222491, + "grad_norm": 2.0906381607055664, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7151311635971069, + "num_tokens": 209390671.0, + "step": 8158 + }, + { + "epoch": 0.8960026356248627, + "grad_norm": 2.4264941215515137, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7163504958152771, + "num_tokens": 209414501.0, + "step": 8159 + }, + { + "epoch": 0.8961124533274764, + "grad_norm": 2.1130623817443848, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6998655796051025, + "num_tokens": 209441527.0, + "step": 8160 + }, + { + "epoch": 0.89622227103009, + "grad_norm": 2.3531758785247803, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.702752947807312, + "num_tokens": 209462579.0, + "step": 8161 + }, + { + "epoch": 0.8963320887327038, + "grad_norm": 2.027250289916992, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7112939357757568, + "num_tokens": 209491187.0, + "step": 8162 + }, + { + "epoch": 0.8964419064353174, + "grad_norm": 2.580967664718628, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7215535640716553, + "num_tokens": 209509966.0, + "step": 8163 + }, + { + "epoch": 0.896551724137931, + "grad_norm": 2.2351346015930176, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7086463570594788, + "num_tokens": 209533834.0, + "step": 8164 + }, + { + "epoch": 0.8966615418405447, + "grad_norm": 2.009918212890625, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6986101269721985, + "num_tokens": 209562589.0, + "step": 8165 + }, + { + "epoch": 0.8967713595431583, + "grad_norm": 2.2017951011657715, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7258951663970947, + "num_tokens": 209588212.0, + "step": 8166 + }, + { + "epoch": 0.896881177245772, + "grad_norm": 2.002763271331787, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7113014459609985, + "num_tokens": 209616469.0, + "step": 8167 + }, + { + "epoch": 0.8969909949483856, + "grad_norm": 1.9152311086654663, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7023918628692627, + "num_tokens": 209648359.0, + "step": 8168 + }, + { + "epoch": 0.8971008126509994, + "grad_norm": 2.2338240146636963, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7115259170532227, + "num_tokens": 209671788.0, + "step": 8169 + }, + { + "epoch": 0.897210630353613, + "grad_norm": 2.0860848426818848, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7136803865432739, + "num_tokens": 209696639.0, + "step": 8170 + }, + { + "epoch": 0.8973204480562267, + "grad_norm": 2.102595806121826, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6880496144294739, + "num_tokens": 209726591.0, + "step": 8171 + }, + { + "epoch": 0.8974302657588403, + "grad_norm": 2.223341941833496, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7308275103569031, + "num_tokens": 209749610.0, + "step": 8172 + }, + { + "epoch": 0.897540083461454, + "grad_norm": 2.446981906890869, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7005773782730103, + "num_tokens": 209770456.0, + "step": 8173 + }, + { + "epoch": 0.8976499011640676, + "grad_norm": 2.2305307388305664, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7171874642372131, + "num_tokens": 209795218.0, + "step": 8174 + }, + { + "epoch": 0.8977597188666813, + "grad_norm": 2.159733533859253, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6903104782104492, + "num_tokens": 209822070.0, + "step": 8175 + }, + { + "epoch": 0.897869536569295, + "grad_norm": 2.321031093597412, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.69859778881073, + "num_tokens": 209845148.0, + "step": 8176 + }, + { + "epoch": 0.8979793542719087, + "grad_norm": 2.1343491077423096, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7076394557952881, + "num_tokens": 209871542.0, + "step": 8177 + }, + { + "epoch": 0.8980891719745223, + "grad_norm": 2.1833508014678955, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6900784969329834, + "num_tokens": 209898512.0, + "step": 8178 + }, + { + "epoch": 0.898198989677136, + "grad_norm": 2.606309652328491, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7073609828948975, + "num_tokens": 209919119.0, + "step": 8179 + }, + { + "epoch": 0.8983088073797496, + "grad_norm": 2.084805488586426, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7038582563400269, + "num_tokens": 209946373.0, + "step": 8180 + }, + { + "epoch": 0.8984186250823633, + "grad_norm": 2.319063663482666, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7278993129730225, + "num_tokens": 209969977.0, + "step": 8181 + }, + { + "epoch": 0.8985284427849769, + "grad_norm": 2.2258076667785645, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6974974870681763, + "num_tokens": 209995371.0, + "step": 8182 + }, + { + "epoch": 0.8986382604875905, + "grad_norm": 2.0217292308807373, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6960471868515015, + "num_tokens": 210025083.0, + "step": 8183 + }, + { + "epoch": 0.8987480781902043, + "grad_norm": 1.9992541074752808, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6917711496353149, + "num_tokens": 210055802.0, + "step": 8184 + }, + { + "epoch": 0.898857895892818, + "grad_norm": 2.4043028354644775, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7049441933631897, + "num_tokens": 210078841.0, + "step": 8185 + }, + { + "epoch": 0.8989677135954316, + "grad_norm": 1.929039716720581, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6879722476005554, + "num_tokens": 210113057.0, + "step": 8186 + }, + { + "epoch": 0.8990775312980452, + "grad_norm": 2.453120708465576, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7088158130645752, + "num_tokens": 210135138.0, + "step": 8187 + }, + { + "epoch": 0.8991873490006589, + "grad_norm": 2.062269687652588, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.7014065980911255, + "num_tokens": 210165254.0, + "step": 8188 + }, + { + "epoch": 0.8992971667032725, + "grad_norm": 2.116962194442749, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7013940811157227, + "num_tokens": 210193582.0, + "step": 8189 + }, + { + "epoch": 0.8994069844058862, + "grad_norm": 2.147771120071411, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6981849670410156, + "num_tokens": 210221391.0, + "step": 8190 + }, + { + "epoch": 0.8995168021084999, + "grad_norm": 2.252673625946045, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7285131216049194, + "num_tokens": 210244740.0, + "step": 8191 + }, + { + "epoch": 0.8996266198111136, + "grad_norm": 2.0929853916168213, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.692804753780365, + "num_tokens": 210272624.0, + "step": 8192 + }, + { + "epoch": 0.8997364375137272, + "grad_norm": 2.0870845317840576, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7387657165527344, + "num_tokens": 210297642.0, + "step": 8193 + }, + { + "epoch": 0.8998462552163409, + "grad_norm": 2.1014950275421143, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6971484422683716, + "num_tokens": 210325989.0, + "step": 8194 + }, + { + "epoch": 0.8999560729189545, + "grad_norm": 2.0982296466827393, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.709636926651001, + "num_tokens": 210352861.0, + "step": 8195 + }, + { + "epoch": 0.9000658906215682, + "grad_norm": 2.331928253173828, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6961321234703064, + "num_tokens": 210379874.0, + "step": 8196 + }, + { + "epoch": 0.9001757083241818, + "grad_norm": 2.081207513809204, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.715652346611023, + "num_tokens": 210406799.0, + "step": 8197 + }, + { + "epoch": 0.9002855260267956, + "grad_norm": 2.182103395462036, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.699647068977356, + "num_tokens": 210432744.0, + "step": 8198 + }, + { + "epoch": 0.9003953437294092, + "grad_norm": 1.91714608669281, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7086988687515259, + "num_tokens": 210462532.0, + "step": 8199 + }, + { + "epoch": 0.9005051614320229, + "grad_norm": 2.1244723796844482, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6965682506561279, + "num_tokens": 210489983.0, + "step": 8200 + }, + { + "epoch": 0.9006149791346365, + "grad_norm": 2.300398349761963, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7003779411315918, + "num_tokens": 210514234.0, + "step": 8201 + }, + { + "epoch": 0.9007247968372502, + "grad_norm": 2.2953453063964844, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7089269161224365, + "num_tokens": 210538377.0, + "step": 8202 + }, + { + "epoch": 0.9008346145398638, + "grad_norm": 2.0920839309692383, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6950014233589172, + "num_tokens": 210566490.0, + "step": 8203 + }, + { + "epoch": 0.9009444322424774, + "grad_norm": 2.0742878913879395, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7142471075057983, + "num_tokens": 210594501.0, + "step": 8204 + }, + { + "epoch": 0.9010542499450912, + "grad_norm": 2.4434938430786133, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7325369119644165, + "num_tokens": 210613960.0, + "step": 8205 + }, + { + "epoch": 0.9011640676477048, + "grad_norm": 2.1393487453460693, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.703946590423584, + "num_tokens": 210638731.0, + "step": 8206 + }, + { + "epoch": 0.9012738853503185, + "grad_norm": 2.081869125366211, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7264701128005981, + "num_tokens": 210666227.0, + "step": 8207 + }, + { + "epoch": 0.9013837030529321, + "grad_norm": 2.324093818664551, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7203742265701294, + "num_tokens": 210688676.0, + "step": 8208 + }, + { + "epoch": 0.9014935207555458, + "grad_norm": 1.9491299390792847, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.6863409280776978, + "num_tokens": 210721171.0, + "step": 8209 + }, + { + "epoch": 0.9016033384581594, + "grad_norm": 1.958988070487976, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6964685916900635, + "num_tokens": 210752099.0, + "step": 8210 + }, + { + "epoch": 0.9017131561607731, + "grad_norm": 1.9685304164886475, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6934566497802734, + "num_tokens": 210781802.0, + "step": 8211 + }, + { + "epoch": 0.9018229738633867, + "grad_norm": 2.415419816970825, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7214032411575317, + "num_tokens": 210802089.0, + "step": 8212 + }, + { + "epoch": 0.9019327915660005, + "grad_norm": 3.123181104660034, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7376130819320679, + "num_tokens": 210815814.0, + "step": 8213 + }, + { + "epoch": 0.9020426092686141, + "grad_norm": 2.486701250076294, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6872795820236206, + "num_tokens": 210838215.0, + "step": 8214 + }, + { + "epoch": 0.9021524269712278, + "grad_norm": 2.455165147781372, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7256815433502197, + "num_tokens": 210857830.0, + "step": 8215 + }, + { + "epoch": 0.9022622446738414, + "grad_norm": 2.16585373878479, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7118229269981384, + "num_tokens": 210883236.0, + "step": 8216 + }, + { + "epoch": 0.9023720623764551, + "grad_norm": 2.162801504135132, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6874396800994873, + "num_tokens": 210908760.0, + "step": 8217 + }, + { + "epoch": 0.9024818800790687, + "grad_norm": 2.086712121963501, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6911426186561584, + "num_tokens": 210936334.0, + "step": 8218 + }, + { + "epoch": 0.9025916977816824, + "grad_norm": 2.3932721614837646, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7254565954208374, + "num_tokens": 210956640.0, + "step": 8219 + }, + { + "epoch": 0.9027015154842961, + "grad_norm": 2.1303436756134033, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6856016516685486, + "num_tokens": 210983359.0, + "step": 8220 + }, + { + "epoch": 0.9028113331869098, + "grad_norm": 2.1821062564849854, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7322388887405396, + "num_tokens": 211008333.0, + "step": 8221 + }, + { + "epoch": 0.9029211508895234, + "grad_norm": 2.390401601791382, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7254464030265808, + "num_tokens": 211028979.0, + "step": 8222 + }, + { + "epoch": 0.903030968592137, + "grad_norm": 1.8911285400390625, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7005078196525574, + "num_tokens": 211057639.0, + "step": 8223 + }, + { + "epoch": 0.9031407862947507, + "grad_norm": 2.199907064437866, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7253884077072144, + "num_tokens": 211081542.0, + "step": 8224 + }, + { + "epoch": 0.9032506039973643, + "grad_norm": 2.3788650035858154, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7167983651161194, + "num_tokens": 211102894.0, + "step": 8225 + }, + { + "epoch": 0.903360421699978, + "grad_norm": 2.310149908065796, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7050184011459351, + "num_tokens": 211126285.0, + "step": 8226 + }, + { + "epoch": 0.9034702394025917, + "grad_norm": 2.0471913814544678, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6940338015556335, + "num_tokens": 211155903.0, + "step": 8227 + }, + { + "epoch": 0.9035800571052054, + "grad_norm": 2.1988744735717773, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7044240236282349, + "num_tokens": 211182314.0, + "step": 8228 + }, + { + "epoch": 0.903689874807819, + "grad_norm": 2.251842737197876, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7122387886047363, + "num_tokens": 211206296.0, + "step": 8229 + }, + { + "epoch": 0.9037996925104327, + "grad_norm": 2.1157517433166504, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7049872875213623, + "num_tokens": 211234588.0, + "step": 8230 + }, + { + "epoch": 0.9039095102130463, + "grad_norm": 2.382559299468994, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7034488320350647, + "num_tokens": 211256940.0, + "step": 8231 + }, + { + "epoch": 0.90401932791566, + "grad_norm": 1.9748148918151855, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6836191415786743, + "num_tokens": 211287755.0, + "step": 8232 + }, + { + "epoch": 0.9041291456182736, + "grad_norm": 2.089062213897705, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7093824744224548, + "num_tokens": 211313923.0, + "step": 8233 + }, + { + "epoch": 0.9042389633208874, + "grad_norm": 2.3027098178863525, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7334024906158447, + "num_tokens": 211335470.0, + "step": 8234 + }, + { + "epoch": 0.904348781023501, + "grad_norm": 2.1254818439483643, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6971771121025085, + "num_tokens": 211360259.0, + "step": 8235 + }, + { + "epoch": 0.9044585987261147, + "grad_norm": 2.260453224182129, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7170342206954956, + "num_tokens": 211383688.0, + "step": 8236 + }, + { + "epoch": 0.9045684164287283, + "grad_norm": 2.3752059936523438, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.710294783115387, + "num_tokens": 211406994.0, + "step": 8237 + }, + { + "epoch": 0.904678234131342, + "grad_norm": 2.0912322998046875, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6977006793022156, + "num_tokens": 211434856.0, + "step": 8238 + }, + { + "epoch": 0.9047880518339556, + "grad_norm": 2.328425884246826, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7109643220901489, + "num_tokens": 211459127.0, + "step": 8239 + }, + { + "epoch": 0.9048978695365693, + "grad_norm": 1.9340740442276, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7014060616493225, + "num_tokens": 211492766.0, + "step": 8240 + }, + { + "epoch": 0.9050076872391829, + "grad_norm": 2.157593250274658, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7033558487892151, + "num_tokens": 211521185.0, + "step": 8241 + }, + { + "epoch": 0.9051175049417967, + "grad_norm": 2.2946372032165527, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6861982345581055, + "num_tokens": 211547636.0, + "step": 8242 + }, + { + "epoch": 0.9052273226444103, + "grad_norm": 2.2803328037261963, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7189778685569763, + "num_tokens": 211572148.0, + "step": 8243 + }, + { + "epoch": 0.905337140347024, + "grad_norm": 2.5476207733154297, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.719649076461792, + "num_tokens": 211591687.0, + "step": 8244 + }, + { + "epoch": 0.9054469580496376, + "grad_norm": 2.530066967010498, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6991534233093262, + "num_tokens": 211612769.0, + "step": 8245 + }, + { + "epoch": 0.9055567757522512, + "grad_norm": 2.106447696685791, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6948779225349426, + "num_tokens": 211639728.0, + "step": 8246 + }, + { + "epoch": 0.9056665934548649, + "grad_norm": 2.17126727104187, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7423859238624573, + "num_tokens": 211665810.0, + "step": 8247 + }, + { + "epoch": 0.9057764111574785, + "grad_norm": 2.184896945953369, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7162461280822754, + "num_tokens": 211690951.0, + "step": 8248 + }, + { + "epoch": 0.9058862288600923, + "grad_norm": 2.2208058834075928, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7165272235870361, + "num_tokens": 211714052.0, + "step": 8249 + }, + { + "epoch": 0.9059960465627059, + "grad_norm": 2.347104549407959, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7027236223220825, + "num_tokens": 211736269.0, + "step": 8250 + }, + { + "epoch": 0.9061058642653196, + "grad_norm": 2.0021445751190186, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6980393528938293, + "num_tokens": 211766933.0, + "step": 8251 + }, + { + "epoch": 0.9062156819679332, + "grad_norm": 2.22261118888855, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7106189131736755, + "num_tokens": 211790470.0, + "step": 8252 + }, + { + "epoch": 0.9063254996705469, + "grad_norm": 2.0576627254486084, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7008851170539856, + "num_tokens": 211820098.0, + "step": 8253 + }, + { + "epoch": 0.9064353173731605, + "grad_norm": 2.204798936843872, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6897503733634949, + "num_tokens": 211845654.0, + "step": 8254 + }, + { + "epoch": 0.9065451350757742, + "grad_norm": 1.9748966693878174, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.6863130331039429, + "num_tokens": 211875291.0, + "step": 8255 + }, + { + "epoch": 0.9066549527783879, + "grad_norm": 2.0504724979400635, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7105711698532104, + "num_tokens": 211900216.0, + "step": 8256 + }, + { + "epoch": 0.9067647704810016, + "grad_norm": 2.4024415016174316, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7205909490585327, + "num_tokens": 211921821.0, + "step": 8257 + }, + { + "epoch": 0.9068745881836152, + "grad_norm": 1.9605460166931152, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7070823907852173, + "num_tokens": 211951319.0, + "step": 8258 + }, + { + "epoch": 0.9069844058862289, + "grad_norm": 2.277019500732422, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7005553245544434, + "num_tokens": 211976190.0, + "step": 8259 + }, + { + "epoch": 0.9070942235888425, + "grad_norm": 2.5681354999542236, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7284520864486694, + "num_tokens": 211995881.0, + "step": 8260 + }, + { + "epoch": 0.9072040412914562, + "grad_norm": 2.1639702320098877, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7198662757873535, + "num_tokens": 212018787.0, + "step": 8261 + }, + { + "epoch": 0.9073138589940698, + "grad_norm": 2.3611347675323486, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7333267331123352, + "num_tokens": 212040331.0, + "step": 8262 + }, + { + "epoch": 0.9074236766966836, + "grad_norm": 2.1001415252685547, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7090798616409302, + "num_tokens": 212066256.0, + "step": 8263 + }, + { + "epoch": 0.9075334943992972, + "grad_norm": 2.2518703937530518, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7489921450614929, + "num_tokens": 212088777.0, + "step": 8264 + }, + { + "epoch": 0.9076433121019108, + "grad_norm": 2.0045833587646484, + "learning_rate": 1e-06, + "loss": 1.1577, + "mean_token_accuracy": 0.6645627021789551, + "num_tokens": 212118470.0, + "step": 8265 + }, + { + "epoch": 0.9077531298045245, + "grad_norm": 2.2708003520965576, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7018056511878967, + "num_tokens": 212142145.0, + "step": 8266 + }, + { + "epoch": 0.9078629475071381, + "grad_norm": 1.842430830001831, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6933287978172302, + "num_tokens": 212176330.0, + "step": 8267 + }, + { + "epoch": 0.9079727652097518, + "grad_norm": 2.3348939418792725, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7049761414527893, + "num_tokens": 212198499.0, + "step": 8268 + }, + { + "epoch": 0.9080825829123654, + "grad_norm": 2.2618417739868164, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7167620062828064, + "num_tokens": 212222601.0, + "step": 8269 + }, + { + "epoch": 0.9081924006149792, + "grad_norm": 2.3252406120300293, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7112520933151245, + "num_tokens": 212245190.0, + "step": 8270 + }, + { + "epoch": 0.9083022183175928, + "grad_norm": 2.0769078731536865, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.7007604837417603, + "num_tokens": 212273430.0, + "step": 8271 + }, + { + "epoch": 0.9084120360202065, + "grad_norm": 2.1376254558563232, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6962307095527649, + "num_tokens": 212300239.0, + "step": 8272 + }, + { + "epoch": 0.9085218537228201, + "grad_norm": 1.991454005241394, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7108278274536133, + "num_tokens": 212328435.0, + "step": 8273 + }, + { + "epoch": 0.9086316714254338, + "grad_norm": 2.233098030090332, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7211804389953613, + "num_tokens": 212352029.0, + "step": 8274 + }, + { + "epoch": 0.9087414891280474, + "grad_norm": 2.038989543914795, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7115046977996826, + "num_tokens": 212380980.0, + "step": 8275 + }, + { + "epoch": 0.9088513068306611, + "grad_norm": 2.0347509384155273, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6959439516067505, + "num_tokens": 212412450.0, + "step": 8276 + }, + { + "epoch": 0.9089611245332747, + "grad_norm": 2.3542850017547607, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7206071019172668, + "num_tokens": 212435460.0, + "step": 8277 + }, + { + "epoch": 0.9090709422358885, + "grad_norm": 2.020925521850586, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.705554723739624, + "num_tokens": 212463666.0, + "step": 8278 + }, + { + "epoch": 0.9091807599385021, + "grad_norm": 2.585118293762207, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7073772549629211, + "num_tokens": 212484905.0, + "step": 8279 + }, + { + "epoch": 0.9092905776411158, + "grad_norm": 2.3217992782592773, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7056468725204468, + "num_tokens": 212508252.0, + "step": 8280 + }, + { + "epoch": 0.9094003953437294, + "grad_norm": 2.1213672161102295, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7052441835403442, + "num_tokens": 212535979.0, + "step": 8281 + }, + { + "epoch": 0.909510213046343, + "grad_norm": 1.9328542947769165, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7035541534423828, + "num_tokens": 212564680.0, + "step": 8282 + }, + { + "epoch": 0.9096200307489567, + "grad_norm": 2.4178640842437744, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.711483359336853, + "num_tokens": 212586728.0, + "step": 8283 + }, + { + "epoch": 0.9097298484515703, + "grad_norm": 2.1516427993774414, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6955179572105408, + "num_tokens": 212612927.0, + "step": 8284 + }, + { + "epoch": 0.9098396661541841, + "grad_norm": 2.473663091659546, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7228044271469116, + "num_tokens": 212633466.0, + "step": 8285 + }, + { + "epoch": 0.9099494838567977, + "grad_norm": 2.5026469230651855, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7525879144668579, + "num_tokens": 212651859.0, + "step": 8286 + }, + { + "epoch": 0.9100593015594114, + "grad_norm": 1.8131717443466187, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6869339942932129, + "num_tokens": 212688370.0, + "step": 8287 + }, + { + "epoch": 0.910169119262025, + "grad_norm": 2.0744569301605225, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6952778100967407, + "num_tokens": 212715492.0, + "step": 8288 + }, + { + "epoch": 0.9102789369646387, + "grad_norm": 2.2671754360198975, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7231850624084473, + "num_tokens": 212738520.0, + "step": 8289 + }, + { + "epoch": 0.9103887546672523, + "grad_norm": 2.072479009628296, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7194689512252808, + "num_tokens": 212765713.0, + "step": 8290 + }, + { + "epoch": 0.910498572369866, + "grad_norm": 2.3653364181518555, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7133455276489258, + "num_tokens": 212787806.0, + "step": 8291 + }, + { + "epoch": 0.9106083900724797, + "grad_norm": 2.0658226013183594, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7163597345352173, + "num_tokens": 212814580.0, + "step": 8292 + }, + { + "epoch": 0.9107182077750934, + "grad_norm": 2.2346551418304443, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.6736352443695068, + "num_tokens": 212841577.0, + "step": 8293 + }, + { + "epoch": 0.910828025477707, + "grad_norm": 2.3075788021087646, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7128393650054932, + "num_tokens": 212866011.0, + "step": 8294 + }, + { + "epoch": 0.9109378431803207, + "grad_norm": 1.9770714044570923, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6945230960845947, + "num_tokens": 212894630.0, + "step": 8295 + }, + { + "epoch": 0.9110476608829343, + "grad_norm": 2.099919319152832, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.705339789390564, + "num_tokens": 212920253.0, + "step": 8296 + }, + { + "epoch": 0.911157478585548, + "grad_norm": 2.4414608478546143, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7203940153121948, + "num_tokens": 212940373.0, + "step": 8297 + }, + { + "epoch": 0.9112672962881616, + "grad_norm": 3.1096668243408203, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7066695690155029, + "num_tokens": 212955136.0, + "step": 8298 + }, + { + "epoch": 0.9113771139907754, + "grad_norm": 2.1746580600738525, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7204923629760742, + "num_tokens": 212978546.0, + "step": 8299 + }, + { + "epoch": 0.911486931693389, + "grad_norm": 2.676085948944092, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7120564579963684, + "num_tokens": 212998564.0, + "step": 8300 + }, + { + "epoch": 0.9115967493960027, + "grad_norm": 2.14731502532959, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7093469500541687, + "num_tokens": 213026542.0, + "step": 8301 + }, + { + "epoch": 0.9117065670986163, + "grad_norm": 2.1367568969726562, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6992831230163574, + "num_tokens": 213053010.0, + "step": 8302 + }, + { + "epoch": 0.91181638480123, + "grad_norm": 2.2965688705444336, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.71567702293396, + "num_tokens": 213074801.0, + "step": 8303 + }, + { + "epoch": 0.9119262025038436, + "grad_norm": 2.0216481685638428, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6825661063194275, + "num_tokens": 213105726.0, + "step": 8304 + }, + { + "epoch": 0.9120360202064572, + "grad_norm": 2.1343111991882324, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7035760879516602, + "num_tokens": 213133890.0, + "step": 8305 + }, + { + "epoch": 0.9121458379090709, + "grad_norm": 2.1331961154937744, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7188321948051453, + "num_tokens": 213159678.0, + "step": 8306 + }, + { + "epoch": 0.9122556556116846, + "grad_norm": 2.212130546569824, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7244622707366943, + "num_tokens": 213183306.0, + "step": 8307 + }, + { + "epoch": 0.9123654733142983, + "grad_norm": 2.0825865268707275, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6944859027862549, + "num_tokens": 213211774.0, + "step": 8308 + }, + { + "epoch": 0.9124752910169119, + "grad_norm": 1.9261325597763062, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7138791084289551, + "num_tokens": 213242639.0, + "step": 8309 + }, + { + "epoch": 0.9125851087195256, + "grad_norm": 2.538547992706299, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7106339931488037, + "num_tokens": 213263505.0, + "step": 8310 + }, + { + "epoch": 0.9126949264221392, + "grad_norm": 2.0636167526245117, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.6970415115356445, + "num_tokens": 213292574.0, + "step": 8311 + }, + { + "epoch": 0.9128047441247529, + "grad_norm": 2.000580072402954, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.6883785724639893, + "num_tokens": 213324178.0, + "step": 8312 + }, + { + "epoch": 0.9129145618273665, + "grad_norm": 2.3256664276123047, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7270921468734741, + "num_tokens": 213346270.0, + "step": 8313 + }, + { + "epoch": 0.9130243795299803, + "grad_norm": 2.085519552230835, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7064386606216431, + "num_tokens": 213372650.0, + "step": 8314 + }, + { + "epoch": 0.9131341972325939, + "grad_norm": 2.170447826385498, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7086362242698669, + "num_tokens": 213399395.0, + "step": 8315 + }, + { + "epoch": 0.9132440149352076, + "grad_norm": 2.2295422554016113, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.684420108795166, + "num_tokens": 213426289.0, + "step": 8316 + }, + { + "epoch": 0.9133538326378212, + "grad_norm": 2.1243319511413574, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6898004412651062, + "num_tokens": 213454238.0, + "step": 8317 + }, + { + "epoch": 0.9134636503404349, + "grad_norm": 2.00457501411438, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7121222615242004, + "num_tokens": 213481939.0, + "step": 8318 + }, + { + "epoch": 0.9135734680430485, + "grad_norm": 2.0527005195617676, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6805199980735779, + "num_tokens": 213510399.0, + "step": 8319 + }, + { + "epoch": 0.9136832857456622, + "grad_norm": 2.32851243019104, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7103984355926514, + "num_tokens": 213533649.0, + "step": 8320 + }, + { + "epoch": 0.9137931034482759, + "grad_norm": 2.2095656394958496, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7112025618553162, + "num_tokens": 213558126.0, + "step": 8321 + }, + { + "epoch": 0.9139029211508896, + "grad_norm": 2.0167832374572754, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7186148166656494, + "num_tokens": 213586605.0, + "step": 8322 + }, + { + "epoch": 0.9140127388535032, + "grad_norm": 2.2521588802337646, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7306748628616333, + "num_tokens": 213610395.0, + "step": 8323 + }, + { + "epoch": 0.9141225565561168, + "grad_norm": 2.2624459266662598, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7353134155273438, + "num_tokens": 213632862.0, + "step": 8324 + }, + { + "epoch": 0.9142323742587305, + "grad_norm": 2.0791735649108887, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7199999094009399, + "num_tokens": 213661033.0, + "step": 8325 + }, + { + "epoch": 0.9143421919613441, + "grad_norm": 2.1226654052734375, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6897040605545044, + "num_tokens": 213689049.0, + "step": 8326 + }, + { + "epoch": 0.9144520096639578, + "grad_norm": 2.084749221801758, + "learning_rate": 1e-06, + "loss": 1.0548, + "mean_token_accuracy": 0.6819185018539429, + "num_tokens": 213716546.0, + "step": 8327 + }, + { + "epoch": 0.9145618273665715, + "grad_norm": 2.495129108428955, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7172170877456665, + "num_tokens": 213737152.0, + "step": 8328 + }, + { + "epoch": 0.9146716450691852, + "grad_norm": 2.005631446838379, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6902521848678589, + "num_tokens": 213766600.0, + "step": 8329 + }, + { + "epoch": 0.9147814627717988, + "grad_norm": 2.1077098846435547, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7051028609275818, + "num_tokens": 213793697.0, + "step": 8330 + }, + { + "epoch": 0.9148912804744125, + "grad_norm": 2.2805285453796387, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7018237113952637, + "num_tokens": 213819783.0, + "step": 8331 + }, + { + "epoch": 0.9150010981770261, + "grad_norm": 1.9247115850448608, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7028583288192749, + "num_tokens": 213850880.0, + "step": 8332 + }, + { + "epoch": 0.9151109158796398, + "grad_norm": 1.8964955806732178, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7127847671508789, + "num_tokens": 213880378.0, + "step": 8333 + }, + { + "epoch": 0.9152207335822534, + "grad_norm": 2.1579535007476807, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7189553380012512, + "num_tokens": 213904674.0, + "step": 8334 + }, + { + "epoch": 0.9153305512848671, + "grad_norm": 1.9104710817337036, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6864741444587708, + "num_tokens": 213937108.0, + "step": 8335 + }, + { + "epoch": 0.9154403689874808, + "grad_norm": 2.2538650035858154, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6970505714416504, + "num_tokens": 213961191.0, + "step": 8336 + }, + { + "epoch": 0.9155501866900945, + "grad_norm": 2.4379098415374756, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7033342123031616, + "num_tokens": 213982806.0, + "step": 8337 + }, + { + "epoch": 0.9156600043927081, + "grad_norm": 1.907213807106018, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7025809288024902, + "num_tokens": 214014452.0, + "step": 8338 + }, + { + "epoch": 0.9157698220953218, + "grad_norm": 2.401686191558838, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6982054710388184, + "num_tokens": 214037084.0, + "step": 8339 + }, + { + "epoch": 0.9158796397979354, + "grad_norm": 1.7967467308044434, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6997541189193726, + "num_tokens": 214074257.0, + "step": 8340 + }, + { + "epoch": 0.915989457500549, + "grad_norm": 2.0254414081573486, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7053182125091553, + "num_tokens": 214104074.0, + "step": 8341 + }, + { + "epoch": 0.9160992752031627, + "grad_norm": 1.9648357629776, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6937597990036011, + "num_tokens": 214133378.0, + "step": 8342 + }, + { + "epoch": 0.9162090929057765, + "grad_norm": 2.1429059505462646, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6868672370910645, + "num_tokens": 214161512.0, + "step": 8343 + }, + { + "epoch": 0.9163189106083901, + "grad_norm": 2.0539019107818604, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7057744264602661, + "num_tokens": 214190698.0, + "step": 8344 + }, + { + "epoch": 0.9164287283110037, + "grad_norm": 2.157945156097412, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7024694681167603, + "num_tokens": 214215386.0, + "step": 8345 + }, + { + "epoch": 0.9165385460136174, + "grad_norm": 2.294886827468872, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7213081121444702, + "num_tokens": 214237364.0, + "step": 8346 + }, + { + "epoch": 0.916648363716231, + "grad_norm": 2.300368070602417, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6908149719238281, + "num_tokens": 214262794.0, + "step": 8347 + }, + { + "epoch": 0.9167581814188447, + "grad_norm": 2.1382298469543457, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7143125534057617, + "num_tokens": 214288258.0, + "step": 8348 + }, + { + "epoch": 0.9168679991214583, + "grad_norm": 2.1488261222839355, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7128382921218872, + "num_tokens": 214315626.0, + "step": 8349 + }, + { + "epoch": 0.9169778168240721, + "grad_norm": 2.3848819732666016, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7061472535133362, + "num_tokens": 214337607.0, + "step": 8350 + }, + { + "epoch": 0.9170876345266857, + "grad_norm": 2.035466194152832, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7120187878608704, + "num_tokens": 214368985.0, + "step": 8351 + }, + { + "epoch": 0.9171974522292994, + "grad_norm": 2.600922107696533, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6950159072875977, + "num_tokens": 214390563.0, + "step": 8352 + }, + { + "epoch": 0.917307269931913, + "grad_norm": 2.5329949855804443, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7203202247619629, + "num_tokens": 214411020.0, + "step": 8353 + }, + { + "epoch": 0.9174170876345267, + "grad_norm": 2.235988140106201, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6933174133300781, + "num_tokens": 214436507.0, + "step": 8354 + }, + { + "epoch": 0.9175269053371403, + "grad_norm": 2.4638755321502686, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7378352880477905, + "num_tokens": 214456052.0, + "step": 8355 + }, + { + "epoch": 0.917636723039754, + "grad_norm": 2.1702966690063477, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7147870063781738, + "num_tokens": 214482555.0, + "step": 8356 + }, + { + "epoch": 0.9177465407423677, + "grad_norm": 1.908162236213684, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7074509859085083, + "num_tokens": 214514242.0, + "step": 8357 + }, + { + "epoch": 0.9178563584449814, + "grad_norm": 2.130520820617676, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.7123665809631348, + "num_tokens": 214540945.0, + "step": 8358 + }, + { + "epoch": 0.917966176147595, + "grad_norm": 1.9498807191848755, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6984595060348511, + "num_tokens": 214569908.0, + "step": 8359 + }, + { + "epoch": 0.9180759938502087, + "grad_norm": 2.424208164215088, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7222146987915039, + "num_tokens": 214591871.0, + "step": 8360 + }, + { + "epoch": 0.9181858115528223, + "grad_norm": 2.220271110534668, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7084150314331055, + "num_tokens": 214617595.0, + "step": 8361 + }, + { + "epoch": 0.918295629255436, + "grad_norm": 2.287630081176758, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6981856822967529, + "num_tokens": 214640709.0, + "step": 8362 + }, + { + "epoch": 0.9184054469580496, + "grad_norm": 2.3862648010253906, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7333474159240723, + "num_tokens": 214661084.0, + "step": 8363 + }, + { + "epoch": 0.9185152646606632, + "grad_norm": 2.4519405364990234, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7139202356338501, + "num_tokens": 214681233.0, + "step": 8364 + }, + { + "epoch": 0.918625082363277, + "grad_norm": 2.2579429149627686, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7103306651115417, + "num_tokens": 214703454.0, + "step": 8365 + }, + { + "epoch": 0.9187349000658906, + "grad_norm": 2.1808598041534424, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7114149332046509, + "num_tokens": 214728010.0, + "step": 8366 + }, + { + "epoch": 0.9188447177685043, + "grad_norm": 2.4288699626922607, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7173778414726257, + "num_tokens": 214749309.0, + "step": 8367 + }, + { + "epoch": 0.9189545354711179, + "grad_norm": 1.9766511917114258, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.700154721736908, + "num_tokens": 214778242.0, + "step": 8368 + }, + { + "epoch": 0.9190643531737316, + "grad_norm": 2.176842212677002, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7346830368041992, + "num_tokens": 214802297.0, + "step": 8369 + }, + { + "epoch": 0.9191741708763452, + "grad_norm": 1.868697166442871, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6776714324951172, + "num_tokens": 214834149.0, + "step": 8370 + }, + { + "epoch": 0.9192839885789589, + "grad_norm": 2.3128576278686523, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7477242946624756, + "num_tokens": 214854576.0, + "step": 8371 + }, + { + "epoch": 0.9193938062815726, + "grad_norm": 2.1501240730285645, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7076744437217712, + "num_tokens": 214879543.0, + "step": 8372 + }, + { + "epoch": 0.9195036239841863, + "grad_norm": 2.075328826904297, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7354009747505188, + "num_tokens": 214905788.0, + "step": 8373 + }, + { + "epoch": 0.9196134416867999, + "grad_norm": 2.4051871299743652, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7136817574501038, + "num_tokens": 214926317.0, + "step": 8374 + }, + { + "epoch": 0.9197232593894136, + "grad_norm": 2.2181692123413086, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7255061864852905, + "num_tokens": 214950577.0, + "step": 8375 + }, + { + "epoch": 0.9198330770920272, + "grad_norm": 1.8961122035980225, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6935936212539673, + "num_tokens": 214983176.0, + "step": 8376 + }, + { + "epoch": 0.9199428947946409, + "grad_norm": 2.3121602535247803, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6933659315109253, + "num_tokens": 215006925.0, + "step": 8377 + }, + { + "epoch": 0.9200527124972545, + "grad_norm": 1.9604737758636475, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.689742922782898, + "num_tokens": 215039783.0, + "step": 8378 + }, + { + "epoch": 0.9201625301998683, + "grad_norm": 2.5315918922424316, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7289159893989563, + "num_tokens": 215059263.0, + "step": 8379 + }, + { + "epoch": 0.9202723479024819, + "grad_norm": 2.1039066314697266, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6943018436431885, + "num_tokens": 215085156.0, + "step": 8380 + }, + { + "epoch": 0.9203821656050956, + "grad_norm": 2.091671943664551, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7233133316040039, + "num_tokens": 215110572.0, + "step": 8381 + }, + { + "epoch": 0.9204919833077092, + "grad_norm": 2.380856990814209, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7239325046539307, + "num_tokens": 215133881.0, + "step": 8382 + }, + { + "epoch": 0.9206018010103229, + "grad_norm": 2.35551381111145, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.7027804851531982, + "num_tokens": 215156883.0, + "step": 8383 + }, + { + "epoch": 0.9207116187129365, + "grad_norm": 2.166785717010498, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7218706011772156, + "num_tokens": 215181399.0, + "step": 8384 + }, + { + "epoch": 0.9208214364155501, + "grad_norm": 2.023928642272949, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7298009991645813, + "num_tokens": 215209644.0, + "step": 8385 + }, + { + "epoch": 0.9209312541181639, + "grad_norm": 2.239367961883545, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7073050737380981, + "num_tokens": 215234158.0, + "step": 8386 + }, + { + "epoch": 0.9210410718207775, + "grad_norm": 2.0999019145965576, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7171683311462402, + "num_tokens": 215261853.0, + "step": 8387 + }, + { + "epoch": 0.9211508895233912, + "grad_norm": 2.2516114711761475, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6862729787826538, + "num_tokens": 215286628.0, + "step": 8388 + }, + { + "epoch": 0.9212607072260048, + "grad_norm": 1.9752012491226196, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7364815473556519, + "num_tokens": 215314791.0, + "step": 8389 + }, + { + "epoch": 0.9213705249286185, + "grad_norm": 2.2304744720458984, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7272144556045532, + "num_tokens": 215338906.0, + "step": 8390 + }, + { + "epoch": 0.9214803426312321, + "grad_norm": 2.0770177841186523, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7111424207687378, + "num_tokens": 215366307.0, + "step": 8391 + }, + { + "epoch": 0.9215901603338458, + "grad_norm": 2.212028741836548, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7312479019165039, + "num_tokens": 215389869.0, + "step": 8392 + }, + { + "epoch": 0.9216999780364594, + "grad_norm": 2.175929069519043, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7296973466873169, + "num_tokens": 215414118.0, + "step": 8393 + }, + { + "epoch": 0.9218097957390732, + "grad_norm": 2.1280081272125244, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7091313600540161, + "num_tokens": 215442369.0, + "step": 8394 + }, + { + "epoch": 0.9219196134416868, + "grad_norm": 2.245558500289917, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7211698293685913, + "num_tokens": 215465623.0, + "step": 8395 + }, + { + "epoch": 0.9220294311443005, + "grad_norm": 2.172894239425659, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7116771936416626, + "num_tokens": 215493274.0, + "step": 8396 + }, + { + "epoch": 0.9221392488469141, + "grad_norm": 2.0751428604125977, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7074475288391113, + "num_tokens": 215521502.0, + "step": 8397 + }, + { + "epoch": 0.9222490665495278, + "grad_norm": 2.298356771469116, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7264618873596191, + "num_tokens": 215543073.0, + "step": 8398 + }, + { + "epoch": 0.9223588842521414, + "grad_norm": 2.236992835998535, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6960076093673706, + "num_tokens": 215568752.0, + "step": 8399 + }, + { + "epoch": 0.922468701954755, + "grad_norm": 1.902787446975708, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6930031180381775, + "num_tokens": 215601286.0, + "step": 8400 + }, + { + "epoch": 0.9225785196573688, + "grad_norm": 1.986450433731079, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6895132064819336, + "num_tokens": 215630371.0, + "step": 8401 + }, + { + "epoch": 0.9226883373599825, + "grad_norm": 2.075735330581665, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.699650228023529, + "num_tokens": 215660550.0, + "step": 8402 + }, + { + "epoch": 0.9227981550625961, + "grad_norm": 2.125715970993042, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7168607115745544, + "num_tokens": 215684779.0, + "step": 8403 + }, + { + "epoch": 0.9229079727652097, + "grad_norm": 2.085061550140381, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7033140659332275, + "num_tokens": 215712576.0, + "step": 8404 + }, + { + "epoch": 0.9230177904678234, + "grad_norm": 2.277221202850342, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.702795684337616, + "num_tokens": 215737601.0, + "step": 8405 + }, + { + "epoch": 0.923127608170437, + "grad_norm": 2.099076509475708, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7191089391708374, + "num_tokens": 215763667.0, + "step": 8406 + }, + { + "epoch": 0.9232374258730507, + "grad_norm": 2.077305793762207, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7034327983856201, + "num_tokens": 215789304.0, + "step": 8407 + }, + { + "epoch": 0.9233472435756644, + "grad_norm": 2.2280051708221436, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.715631902217865, + "num_tokens": 215814100.0, + "step": 8408 + }, + { + "epoch": 0.9234570612782781, + "grad_norm": 2.329775333404541, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7294124364852905, + "num_tokens": 215835931.0, + "step": 8409 + }, + { + "epoch": 0.9235668789808917, + "grad_norm": 1.9044137001037598, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6940126419067383, + "num_tokens": 215867284.0, + "step": 8410 + }, + { + "epoch": 0.9236766966835054, + "grad_norm": 2.0892486572265625, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6934802532196045, + "num_tokens": 215895460.0, + "step": 8411 + }, + { + "epoch": 0.923786514386119, + "grad_norm": 2.061159372329712, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.6846585273742676, + "num_tokens": 215927951.0, + "step": 8412 + }, + { + "epoch": 0.9238963320887327, + "grad_norm": 2.25126051902771, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7263497114181519, + "num_tokens": 215951003.0, + "step": 8413 + }, + { + "epoch": 0.9240061497913463, + "grad_norm": 1.977381706237793, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7139376997947693, + "num_tokens": 215980011.0, + "step": 8414 + }, + { + "epoch": 0.9241159674939601, + "grad_norm": 1.8980506658554077, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6806995868682861, + "num_tokens": 216013190.0, + "step": 8415 + }, + { + "epoch": 0.9242257851965737, + "grad_norm": 1.9025959968566895, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7062081098556519, + "num_tokens": 216044862.0, + "step": 8416 + }, + { + "epoch": 0.9243356028991874, + "grad_norm": 2.338608980178833, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7217844724655151, + "num_tokens": 216067357.0, + "step": 8417 + }, + { + "epoch": 0.924445420601801, + "grad_norm": 2.209848403930664, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6856947541236877, + "num_tokens": 216093273.0, + "step": 8418 + }, + { + "epoch": 0.9245552383044147, + "grad_norm": 2.3835809230804443, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7133737802505493, + "num_tokens": 216114985.0, + "step": 8419 + }, + { + "epoch": 0.9246650560070283, + "grad_norm": 2.247957229614258, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7126023173332214, + "num_tokens": 216138360.0, + "step": 8420 + }, + { + "epoch": 0.924774873709642, + "grad_norm": 2.7228853702545166, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7184613943099976, + "num_tokens": 216156221.0, + "step": 8421 + }, + { + "epoch": 0.9248846914122557, + "grad_norm": 2.2057852745056152, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7039176821708679, + "num_tokens": 216181358.0, + "step": 8422 + }, + { + "epoch": 0.9249945091148694, + "grad_norm": 2.0627777576446533, + "learning_rate": 1e-06, + "loss": 1.0977, + "mean_token_accuracy": 0.6818392276763916, + "num_tokens": 216212612.0, + "step": 8423 + }, + { + "epoch": 0.925104326817483, + "grad_norm": 2.079310894012451, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6912519335746765, + "num_tokens": 216243926.0, + "step": 8424 + }, + { + "epoch": 0.9252141445200966, + "grad_norm": 2.3568708896636963, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7075700759887695, + "num_tokens": 216266706.0, + "step": 8425 + }, + { + "epoch": 0.9253239622227103, + "grad_norm": 2.237481117248535, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7047730088233948, + "num_tokens": 216290177.0, + "step": 8426 + }, + { + "epoch": 0.9254337799253239, + "grad_norm": 2.0782182216644287, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7050024271011353, + "num_tokens": 216315814.0, + "step": 8427 + }, + { + "epoch": 0.9255435976279376, + "grad_norm": 1.963213562965393, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7122790217399597, + "num_tokens": 216345643.0, + "step": 8428 + }, + { + "epoch": 0.9256534153305512, + "grad_norm": 2.2334301471710205, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6991345882415771, + "num_tokens": 216370073.0, + "step": 8429 + }, + { + "epoch": 0.925763233033165, + "grad_norm": 2.013124942779541, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7072416543960571, + "num_tokens": 216400429.0, + "step": 8430 + }, + { + "epoch": 0.9258730507357786, + "grad_norm": 2.138904094696045, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6999492645263672, + "num_tokens": 216428268.0, + "step": 8431 + }, + { + "epoch": 0.9259828684383923, + "grad_norm": 2.0165698528289795, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6938912868499756, + "num_tokens": 216461252.0, + "step": 8432 + }, + { + "epoch": 0.9260926861410059, + "grad_norm": 2.2334470748901367, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7253942489624023, + "num_tokens": 216484839.0, + "step": 8433 + }, + { + "epoch": 0.9262025038436196, + "grad_norm": 2.041710138320923, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7238727807998657, + "num_tokens": 216511755.0, + "step": 8434 + }, + { + "epoch": 0.9263123215462332, + "grad_norm": 2.007798910140991, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7049752473831177, + "num_tokens": 216541040.0, + "step": 8435 + }, + { + "epoch": 0.9264221392488469, + "grad_norm": 2.1422934532165527, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6924123167991638, + "num_tokens": 216566325.0, + "step": 8436 + }, + { + "epoch": 0.9265319569514606, + "grad_norm": 2.201944351196289, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7110974192619324, + "num_tokens": 216592411.0, + "step": 8437 + }, + { + "epoch": 0.9266417746540743, + "grad_norm": 1.8219887018203735, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.693772554397583, + "num_tokens": 216628466.0, + "step": 8438 + }, + { + "epoch": 0.9267515923566879, + "grad_norm": 2.2181339263916016, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6959477066993713, + "num_tokens": 216652744.0, + "step": 8439 + }, + { + "epoch": 0.9268614100593016, + "grad_norm": 2.3435351848602295, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7403126955032349, + "num_tokens": 216675215.0, + "step": 8440 + }, + { + "epoch": 0.9269712277619152, + "grad_norm": 2.0033316612243652, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6908022165298462, + "num_tokens": 216704236.0, + "step": 8441 + }, + { + "epoch": 0.9270810454645289, + "grad_norm": 2.216709852218628, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7038900256156921, + "num_tokens": 216728012.0, + "step": 8442 + }, + { + "epoch": 0.9271908631671425, + "grad_norm": 2.0842092037200928, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7275812029838562, + "num_tokens": 216751207.0, + "step": 8443 + }, + { + "epoch": 0.9273006808697563, + "grad_norm": 2.1708714962005615, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6901626586914062, + "num_tokens": 216779960.0, + "step": 8444 + }, + { + "epoch": 0.9274104985723699, + "grad_norm": 1.8675097227096558, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6929202079772949, + "num_tokens": 216812246.0, + "step": 8445 + }, + { + "epoch": 0.9275203162749835, + "grad_norm": 2.5506041049957275, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7149993777275085, + "num_tokens": 216831403.0, + "step": 8446 + }, + { + "epoch": 0.9276301339775972, + "grad_norm": 2.188451051712036, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.702250599861145, + "num_tokens": 216859183.0, + "step": 8447 + }, + { + "epoch": 0.9277399516802108, + "grad_norm": 2.438507080078125, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7273575067520142, + "num_tokens": 216878916.0, + "step": 8448 + }, + { + "epoch": 0.9278497693828245, + "grad_norm": 2.001401662826538, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6885128021240234, + "num_tokens": 216907968.0, + "step": 8449 + }, + { + "epoch": 0.9279595870854381, + "grad_norm": 2.214311361312866, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7267906665802002, + "num_tokens": 216931904.0, + "step": 8450 + }, + { + "epoch": 0.9280694047880519, + "grad_norm": 2.244447946548462, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6867419481277466, + "num_tokens": 216956455.0, + "step": 8451 + }, + { + "epoch": 0.9281792224906655, + "grad_norm": 2.0534627437591553, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7037801742553711, + "num_tokens": 216984745.0, + "step": 8452 + }, + { + "epoch": 0.9282890401932792, + "grad_norm": 2.124238967895508, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7073127031326294, + "num_tokens": 217009732.0, + "step": 8453 + }, + { + "epoch": 0.9283988578958928, + "grad_norm": 2.306683301925659, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7120264768600464, + "num_tokens": 217032707.0, + "step": 8454 + }, + { + "epoch": 0.9285086755985065, + "grad_norm": 2.075256109237671, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.700562596321106, + "num_tokens": 217059411.0, + "step": 8455 + }, + { + "epoch": 0.9286184933011201, + "grad_norm": 1.735541582107544, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6918473243713379, + "num_tokens": 217097346.0, + "step": 8456 + }, + { + "epoch": 0.9287283110037338, + "grad_norm": 2.096849203109741, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7102967500686646, + "num_tokens": 217124821.0, + "step": 8457 + }, + { + "epoch": 0.9288381287063474, + "grad_norm": 2.4516522884368896, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7159332036972046, + "num_tokens": 217146500.0, + "step": 8458 + }, + { + "epoch": 0.9289479464089612, + "grad_norm": 1.954715609550476, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6766895055770874, + "num_tokens": 217177171.0, + "step": 8459 + }, + { + "epoch": 0.9290577641115748, + "grad_norm": 2.2552592754364014, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7190272808074951, + "num_tokens": 217201353.0, + "step": 8460 + }, + { + "epoch": 0.9291675818141885, + "grad_norm": 1.979613184928894, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7214235067367554, + "num_tokens": 217229715.0, + "step": 8461 + }, + { + "epoch": 0.9292773995168021, + "grad_norm": 2.0493876934051514, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7199119329452515, + "num_tokens": 217258487.0, + "step": 8462 + }, + { + "epoch": 0.9293872172194158, + "grad_norm": 2.4114482402801514, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7241562604904175, + "num_tokens": 217279852.0, + "step": 8463 + }, + { + "epoch": 0.9294970349220294, + "grad_norm": 2.5550177097320557, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7015101909637451, + "num_tokens": 217299604.0, + "step": 8464 + }, + { + "epoch": 0.929606852624643, + "grad_norm": 1.8372968435287476, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7261950969696045, + "num_tokens": 217334053.0, + "step": 8465 + }, + { + "epoch": 0.9297166703272568, + "grad_norm": 2.028106689453125, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7090439796447754, + "num_tokens": 217364009.0, + "step": 8466 + }, + { + "epoch": 0.9298264880298704, + "grad_norm": 2.1683123111724854, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7161800861358643, + "num_tokens": 217388345.0, + "step": 8467 + }, + { + "epoch": 0.9299363057324841, + "grad_norm": 2.2959487438201904, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7016910314559937, + "num_tokens": 217410619.0, + "step": 8468 + }, + { + "epoch": 0.9300461234350977, + "grad_norm": 2.294996500015259, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7244089841842651, + "num_tokens": 217434666.0, + "step": 8469 + }, + { + "epoch": 0.9301559411377114, + "grad_norm": 2.202058792114258, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7029606103897095, + "num_tokens": 217460852.0, + "step": 8470 + }, + { + "epoch": 0.930265758840325, + "grad_norm": 2.3215601444244385, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7161184549331665, + "num_tokens": 217488642.0, + "step": 8471 + }, + { + "epoch": 0.9303755765429387, + "grad_norm": 2.2869937419891357, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7053490877151489, + "num_tokens": 217513122.0, + "step": 8472 + }, + { + "epoch": 0.9304853942455524, + "grad_norm": 1.97992742061615, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6916331052780151, + "num_tokens": 217543547.0, + "step": 8473 + }, + { + "epoch": 0.9305952119481661, + "grad_norm": 2.5435309410095215, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7261004447937012, + "num_tokens": 217561911.0, + "step": 8474 + }, + { + "epoch": 0.9307050296507797, + "grad_norm": 2.149301290512085, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7106086611747742, + "num_tokens": 217587188.0, + "step": 8475 + }, + { + "epoch": 0.9308148473533934, + "grad_norm": 2.4280848503112793, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7212443947792053, + "num_tokens": 217607371.0, + "step": 8476 + }, + { + "epoch": 0.930924665056007, + "grad_norm": 2.6143763065338135, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7081596255302429, + "num_tokens": 217625948.0, + "step": 8477 + }, + { + "epoch": 0.9310344827586207, + "grad_norm": 2.6875646114349365, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7174069881439209, + "num_tokens": 217643803.0, + "step": 8478 + }, + { + "epoch": 0.9311443004612343, + "grad_norm": 2.1388697624206543, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7088466882705688, + "num_tokens": 217671426.0, + "step": 8479 + }, + { + "epoch": 0.9312541181638481, + "grad_norm": 2.5867393016815186, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7159122228622437, + "num_tokens": 217691618.0, + "step": 8480 + }, + { + "epoch": 0.9313639358664617, + "grad_norm": 2.2189033031463623, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6962584853172302, + "num_tokens": 217718301.0, + "step": 8481 + }, + { + "epoch": 0.9314737535690754, + "grad_norm": 2.2797796726226807, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7045305967330933, + "num_tokens": 217745784.0, + "step": 8482 + }, + { + "epoch": 0.931583571271689, + "grad_norm": 2.3177738189697266, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6909077763557434, + "num_tokens": 217769438.0, + "step": 8483 + }, + { + "epoch": 0.9316933889743026, + "grad_norm": 2.3385298252105713, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7118778228759766, + "num_tokens": 217797188.0, + "step": 8484 + }, + { + "epoch": 0.9318032066769163, + "grad_norm": 2.0790116786956787, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.712916910648346, + "num_tokens": 217824031.0, + "step": 8485 + }, + { + "epoch": 0.9319130243795299, + "grad_norm": 2.114832878112793, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6819324493408203, + "num_tokens": 217851418.0, + "step": 8486 + }, + { + "epoch": 0.9320228420821436, + "grad_norm": 2.105760097503662, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.6995487213134766, + "num_tokens": 217878540.0, + "step": 8487 + }, + { + "epoch": 0.9321326597847573, + "grad_norm": 2.1494362354278564, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6989203095436096, + "num_tokens": 217904180.0, + "step": 8488 + }, + { + "epoch": 0.932242477487371, + "grad_norm": 2.274318218231201, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7036761045455933, + "num_tokens": 217927088.0, + "step": 8489 + }, + { + "epoch": 0.9323522951899846, + "grad_norm": 2.4251720905303955, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.6962727308273315, + "num_tokens": 217948588.0, + "step": 8490 + }, + { + "epoch": 0.9324621128925983, + "grad_norm": 2.1032238006591797, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7105274200439453, + "num_tokens": 217972879.0, + "step": 8491 + }, + { + "epoch": 0.9325719305952119, + "grad_norm": 2.3140478134155273, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7016583681106567, + "num_tokens": 217995979.0, + "step": 8492 + }, + { + "epoch": 0.9326817482978256, + "grad_norm": 2.0614802837371826, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7116944193840027, + "num_tokens": 218023465.0, + "step": 8493 + }, + { + "epoch": 0.9327915660004392, + "grad_norm": 2.1161763668060303, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6980351805686951, + "num_tokens": 218050536.0, + "step": 8494 + }, + { + "epoch": 0.932901383703053, + "grad_norm": 2.0000665187835693, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7313283681869507, + "num_tokens": 218078413.0, + "step": 8495 + }, + { + "epoch": 0.9330112014056666, + "grad_norm": 2.2859976291656494, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7009747624397278, + "num_tokens": 218102083.0, + "step": 8496 + }, + { + "epoch": 0.9331210191082803, + "grad_norm": 2.163565158843994, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7206198573112488, + "num_tokens": 218127140.0, + "step": 8497 + }, + { + "epoch": 0.9332308368108939, + "grad_norm": 2.3615000247955322, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7278256416320801, + "num_tokens": 218149494.0, + "step": 8498 + }, + { + "epoch": 0.9333406545135076, + "grad_norm": 2.3212380409240723, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.723653256893158, + "num_tokens": 218172039.0, + "step": 8499 + }, + { + "epoch": 0.9334504722161212, + "grad_norm": 2.2500529289245605, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7045959234237671, + "num_tokens": 218197271.0, + "step": 8500 + }, + { + "epoch": 0.9335602899187349, + "grad_norm": 1.9509726762771606, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6967111825942993, + "num_tokens": 218227673.0, + "step": 8501 + }, + { + "epoch": 0.9336701076213486, + "grad_norm": 2.2841572761535645, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7010526657104492, + "num_tokens": 218250831.0, + "step": 8502 + }, + { + "epoch": 0.9337799253239623, + "grad_norm": 2.080345630645752, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7191202640533447, + "num_tokens": 218277364.0, + "step": 8503 + }, + { + "epoch": 0.9338897430265759, + "grad_norm": 2.194222927093506, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7033586502075195, + "num_tokens": 218303572.0, + "step": 8504 + }, + { + "epoch": 0.9339995607291895, + "grad_norm": 2.0047333240509033, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7050777673721313, + "num_tokens": 218331243.0, + "step": 8505 + }, + { + "epoch": 0.9341093784318032, + "grad_norm": 2.3484835624694824, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7061710357666016, + "num_tokens": 218353684.0, + "step": 8506 + }, + { + "epoch": 0.9342191961344168, + "grad_norm": 2.1178972721099854, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.711223304271698, + "num_tokens": 218380019.0, + "step": 8507 + }, + { + "epoch": 0.9343290138370305, + "grad_norm": 2.6051995754241943, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7082525491714478, + "num_tokens": 218399297.0, + "step": 8508 + }, + { + "epoch": 0.9344388315396442, + "grad_norm": 2.3588953018188477, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7247641086578369, + "num_tokens": 218421753.0, + "step": 8509 + }, + { + "epoch": 0.9345486492422579, + "grad_norm": 2.417360782623291, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7163123488426208, + "num_tokens": 218442458.0, + "step": 8510 + }, + { + "epoch": 0.9346584669448715, + "grad_norm": 2.2252044677734375, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7155477404594421, + "num_tokens": 218465673.0, + "step": 8511 + }, + { + "epoch": 0.9347682846474852, + "grad_norm": 2.2504172325134277, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7091491222381592, + "num_tokens": 218489332.0, + "step": 8512 + }, + { + "epoch": 0.9348781023500988, + "grad_norm": 2.1718921661376953, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6978204250335693, + "num_tokens": 218513886.0, + "step": 8513 + }, + { + "epoch": 0.9349879200527125, + "grad_norm": 2.0932443141937256, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7040724754333496, + "num_tokens": 218542474.0, + "step": 8514 + }, + { + "epoch": 0.9350977377553261, + "grad_norm": 2.228062391281128, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7252773642539978, + "num_tokens": 218562969.0, + "step": 8515 + }, + { + "epoch": 0.9352075554579398, + "grad_norm": 1.902737021446228, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.688154935836792, + "num_tokens": 218595068.0, + "step": 8516 + }, + { + "epoch": 0.9353173731605535, + "grad_norm": 2.1420857906341553, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7036433219909668, + "num_tokens": 218620833.0, + "step": 8517 + }, + { + "epoch": 0.9354271908631672, + "grad_norm": 2.2362406253814697, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7066255807876587, + "num_tokens": 218646152.0, + "step": 8518 + }, + { + "epoch": 0.9355370085657808, + "grad_norm": 2.313872814178467, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7135795950889587, + "num_tokens": 218670544.0, + "step": 8519 + }, + { + "epoch": 0.9356468262683945, + "grad_norm": 2.2071497440338135, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7052611112594604, + "num_tokens": 218695400.0, + "step": 8520 + }, + { + "epoch": 0.9357566439710081, + "grad_norm": 2.190493583679199, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7069637775421143, + "num_tokens": 218719915.0, + "step": 8521 + }, + { + "epoch": 0.9358664616736218, + "grad_norm": 2.236353874206543, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7139620780944824, + "num_tokens": 218744509.0, + "step": 8522 + }, + { + "epoch": 0.9359762793762354, + "grad_norm": 2.068899631500244, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7027930021286011, + "num_tokens": 218772447.0, + "step": 8523 + }, + { + "epoch": 0.9360860970788492, + "grad_norm": 2.589425802230835, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7242677211761475, + "num_tokens": 218789577.0, + "step": 8524 + }, + { + "epoch": 0.9361959147814628, + "grad_norm": 1.9159821271896362, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7218983173370361, + "num_tokens": 218819938.0, + "step": 8525 + }, + { + "epoch": 0.9363057324840764, + "grad_norm": 2.1178908348083496, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7085204124450684, + "num_tokens": 218846301.0, + "step": 8526 + }, + { + "epoch": 0.9364155501866901, + "grad_norm": 2.1605916023254395, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7266179323196411, + "num_tokens": 218868710.0, + "step": 8527 + }, + { + "epoch": 0.9365253678893037, + "grad_norm": 2.2346203327178955, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7364071607589722, + "num_tokens": 218891296.0, + "step": 8528 + }, + { + "epoch": 0.9366351855919174, + "grad_norm": 2.1713149547576904, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7377372980117798, + "num_tokens": 218913401.0, + "step": 8529 + }, + { + "epoch": 0.936745003294531, + "grad_norm": 2.3302252292633057, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7195544242858887, + "num_tokens": 218936341.0, + "step": 8530 + }, + { + "epoch": 0.9368548209971448, + "grad_norm": 2.1264431476593018, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6898298263549805, + "num_tokens": 218964928.0, + "step": 8531 + }, + { + "epoch": 0.9369646386997584, + "grad_norm": 2.093951463699341, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7020179629325867, + "num_tokens": 218992977.0, + "step": 8532 + }, + { + "epoch": 0.9370744564023721, + "grad_norm": 2.6123714447021484, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7104611396789551, + "num_tokens": 219011642.0, + "step": 8533 + }, + { + "epoch": 0.9371842741049857, + "grad_norm": 2.328496217727661, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7077978849411011, + "num_tokens": 219034529.0, + "step": 8534 + }, + { + "epoch": 0.9372940918075994, + "grad_norm": 2.1923046112060547, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7069569230079651, + "num_tokens": 219060505.0, + "step": 8535 + }, + { + "epoch": 0.937403909510213, + "grad_norm": 1.941594123840332, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7198789119720459, + "num_tokens": 219090708.0, + "step": 8536 + }, + { + "epoch": 0.9375137272128267, + "grad_norm": 2.003938674926758, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6955894231796265, + "num_tokens": 219119460.0, + "step": 8537 + }, + { + "epoch": 0.9376235449154404, + "grad_norm": 2.0446949005126953, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7036750912666321, + "num_tokens": 219148096.0, + "step": 8538 + }, + { + "epoch": 0.9377333626180541, + "grad_norm": 2.1330509185791016, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.700518012046814, + "num_tokens": 219175007.0, + "step": 8539 + }, + { + "epoch": 0.9378431803206677, + "grad_norm": 2.000061273574829, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7096758484840393, + "num_tokens": 219203749.0, + "step": 8540 + }, + { + "epoch": 0.9379529980232814, + "grad_norm": 2.2196531295776367, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7141994833946228, + "num_tokens": 219227679.0, + "step": 8541 + }, + { + "epoch": 0.938062815725895, + "grad_norm": 2.384782552719116, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7159590721130371, + "num_tokens": 219250618.0, + "step": 8542 + }, + { + "epoch": 0.9381726334285087, + "grad_norm": 2.3082306385040283, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7132240533828735, + "num_tokens": 219274609.0, + "step": 8543 + }, + { + "epoch": 0.9382824511311223, + "grad_norm": 2.0451011657714844, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.7003694176673889, + "num_tokens": 219303884.0, + "step": 8544 + }, + { + "epoch": 0.9383922688337359, + "grad_norm": 2.0906929969787598, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7117663621902466, + "num_tokens": 219329576.0, + "step": 8545 + }, + { + "epoch": 0.9385020865363497, + "grad_norm": 2.080449104309082, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7088171243667603, + "num_tokens": 219356133.0, + "step": 8546 + }, + { + "epoch": 0.9386119042389633, + "grad_norm": 1.9075181484222412, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6936584711074829, + "num_tokens": 219387386.0, + "step": 8547 + }, + { + "epoch": 0.938721721941577, + "grad_norm": 2.2897143363952637, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7034632563591003, + "num_tokens": 219412293.0, + "step": 8548 + }, + { + "epoch": 0.9388315396441906, + "grad_norm": 2.2910754680633545, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7119377851486206, + "num_tokens": 219435250.0, + "step": 8549 + }, + { + "epoch": 0.9389413573468043, + "grad_norm": 2.1424756050109863, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6970515847206116, + "num_tokens": 219462067.0, + "step": 8550 + }, + { + "epoch": 0.9390511750494179, + "grad_norm": 2.0720536708831787, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7345343232154846, + "num_tokens": 219488120.0, + "step": 8551 + }, + { + "epoch": 0.9391609927520316, + "grad_norm": 2.3320446014404297, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7259231805801392, + "num_tokens": 219510803.0, + "step": 8552 + }, + { + "epoch": 0.9392708104546453, + "grad_norm": 2.6254260540008545, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7009483575820923, + "num_tokens": 219531260.0, + "step": 8553 + }, + { + "epoch": 0.939380628157259, + "grad_norm": 2.7214088439941406, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.697719156742096, + "num_tokens": 219549253.0, + "step": 8554 + }, + { + "epoch": 0.9394904458598726, + "grad_norm": 2.237668991088867, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7074226140975952, + "num_tokens": 219575037.0, + "step": 8555 + }, + { + "epoch": 0.9396002635624863, + "grad_norm": 2.13496994972229, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6914543509483337, + "num_tokens": 219601741.0, + "step": 8556 + }, + { + "epoch": 0.9397100812650999, + "grad_norm": 2.086625814437866, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7094430923461914, + "num_tokens": 219630330.0, + "step": 8557 + }, + { + "epoch": 0.9398198989677136, + "grad_norm": 2.140352725982666, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6905255913734436, + "num_tokens": 219655723.0, + "step": 8558 + }, + { + "epoch": 0.9399297166703272, + "grad_norm": 2.5318596363067627, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7252598404884338, + "num_tokens": 219674698.0, + "step": 8559 + }, + { + "epoch": 0.940039534372941, + "grad_norm": 2.241730213165283, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.7016938328742981, + "num_tokens": 219698684.0, + "step": 8560 + }, + { + "epoch": 0.9401493520755546, + "grad_norm": 2.141326904296875, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7180287837982178, + "num_tokens": 219724317.0, + "step": 8561 + }, + { + "epoch": 0.9402591697781683, + "grad_norm": 2.010591983795166, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7008787393569946, + "num_tokens": 219755104.0, + "step": 8562 + }, + { + "epoch": 0.9403689874807819, + "grad_norm": 2.314866781234741, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7198334336280823, + "num_tokens": 219776767.0, + "step": 8563 + }, + { + "epoch": 0.9404788051833955, + "grad_norm": 2.0005996227264404, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7104386687278748, + "num_tokens": 219804951.0, + "step": 8564 + }, + { + "epoch": 0.9405886228860092, + "grad_norm": 2.0265920162200928, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7115663290023804, + "num_tokens": 219834988.0, + "step": 8565 + }, + { + "epoch": 0.9406984405886228, + "grad_norm": 2.317840099334717, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7244589328765869, + "num_tokens": 219856938.0, + "step": 8566 + }, + { + "epoch": 0.9408082582912366, + "grad_norm": 2.232905387878418, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7230932712554932, + "num_tokens": 219880176.0, + "step": 8567 + }, + { + "epoch": 0.9409180759938502, + "grad_norm": 1.995107889175415, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7085105180740356, + "num_tokens": 219910062.0, + "step": 8568 + }, + { + "epoch": 0.9410278936964639, + "grad_norm": 2.0282328128814697, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7061272859573364, + "num_tokens": 219939311.0, + "step": 8569 + }, + { + "epoch": 0.9411377113990775, + "grad_norm": 2.520411252975464, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7228391170501709, + "num_tokens": 219960464.0, + "step": 8570 + }, + { + "epoch": 0.9412475291016912, + "grad_norm": 2.113457202911377, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7253313064575195, + "num_tokens": 219987248.0, + "step": 8571 + }, + { + "epoch": 0.9413573468043048, + "grad_norm": 2.2443294525146484, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6935501098632812, + "num_tokens": 220012650.0, + "step": 8572 + }, + { + "epoch": 0.9414671645069185, + "grad_norm": 2.2355129718780518, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.733791172504425, + "num_tokens": 220034328.0, + "step": 8573 + }, + { + "epoch": 0.9415769822095322, + "grad_norm": 2.068183183670044, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7147688865661621, + "num_tokens": 220063879.0, + "step": 8574 + }, + { + "epoch": 0.9416867999121459, + "grad_norm": 2.076427936553955, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6987333297729492, + "num_tokens": 220091777.0, + "step": 8575 + }, + { + "epoch": 0.9417966176147595, + "grad_norm": 2.0704119205474854, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7087762951850891, + "num_tokens": 220120802.0, + "step": 8576 + }, + { + "epoch": 0.9419064353173732, + "grad_norm": 2.085463523864746, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7050222754478455, + "num_tokens": 220147743.0, + "step": 8577 + }, + { + "epoch": 0.9420162530199868, + "grad_norm": 2.5135271549224854, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7184669971466064, + "num_tokens": 220167504.0, + "step": 8578 + }, + { + "epoch": 0.9421260707226005, + "grad_norm": 2.1750383377075195, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7236837148666382, + "num_tokens": 220194134.0, + "step": 8579 + }, + { + "epoch": 0.9422358884252141, + "grad_norm": 2.373929738998413, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7135094404220581, + "num_tokens": 220216541.0, + "step": 8580 + }, + { + "epoch": 0.9423457061278278, + "grad_norm": 2.1320126056671143, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.71238112449646, + "num_tokens": 220241907.0, + "step": 8581 + }, + { + "epoch": 0.9424555238304415, + "grad_norm": 2.1508872509002686, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6832457780838013, + "num_tokens": 220268436.0, + "step": 8582 + }, + { + "epoch": 0.9425653415330552, + "grad_norm": 2.6325557231903076, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7163144946098328, + "num_tokens": 220286413.0, + "step": 8583 + }, + { + "epoch": 0.9426751592356688, + "grad_norm": 2.046865701675415, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6915656328201294, + "num_tokens": 220315179.0, + "step": 8584 + }, + { + "epoch": 0.9427849769382824, + "grad_norm": 2.3727867603302, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7500926852226257, + "num_tokens": 220335878.0, + "step": 8585 + }, + { + "epoch": 0.9428947946408961, + "grad_norm": 1.9627858400344849, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6992884874343872, + "num_tokens": 220368389.0, + "step": 8586 + }, + { + "epoch": 0.9430046123435097, + "grad_norm": 2.2061409950256348, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7096103429794312, + "num_tokens": 220393444.0, + "step": 8587 + }, + { + "epoch": 0.9431144300461234, + "grad_norm": 2.343334436416626, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7207406759262085, + "num_tokens": 220417139.0, + "step": 8588 + }, + { + "epoch": 0.9432242477487371, + "grad_norm": 2.4931490421295166, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7387973070144653, + "num_tokens": 220436054.0, + "step": 8589 + }, + { + "epoch": 0.9433340654513508, + "grad_norm": 2.41143798828125, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7183444499969482, + "num_tokens": 220458064.0, + "step": 8590 + }, + { + "epoch": 0.9434438831539644, + "grad_norm": 2.218013286590576, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7029614448547363, + "num_tokens": 220481749.0, + "step": 8591 + }, + { + "epoch": 0.9435537008565781, + "grad_norm": 2.212249517440796, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7073609828948975, + "num_tokens": 220505695.0, + "step": 8592 + }, + { + "epoch": 0.9436635185591917, + "grad_norm": 2.1957924365997314, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7324725389480591, + "num_tokens": 220529157.0, + "step": 8593 + }, + { + "epoch": 0.9437733362618054, + "grad_norm": 1.839604139328003, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6971728801727295, + "num_tokens": 220562890.0, + "step": 8594 + }, + { + "epoch": 0.943883153964419, + "grad_norm": 2.047335386276245, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6904994249343872, + "num_tokens": 220590044.0, + "step": 8595 + }, + { + "epoch": 0.9439929716670328, + "grad_norm": 2.2150652408599854, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7155337333679199, + "num_tokens": 220614579.0, + "step": 8596 + }, + { + "epoch": 0.9441027893696464, + "grad_norm": 1.8060697317123413, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.72327721118927, + "num_tokens": 220645902.0, + "step": 8597 + }, + { + "epoch": 0.9442126070722601, + "grad_norm": 2.226036548614502, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7293696403503418, + "num_tokens": 220671743.0, + "step": 8598 + }, + { + "epoch": 0.9443224247748737, + "grad_norm": 2.262155294418335, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7024077773094177, + "num_tokens": 220694789.0, + "step": 8599 + }, + { + "epoch": 0.9444322424774874, + "grad_norm": 2.2315921783447266, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6967039704322815, + "num_tokens": 220721731.0, + "step": 8600 + }, + { + "epoch": 0.944542060180101, + "grad_norm": 2.311237335205078, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.699498176574707, + "num_tokens": 220744759.0, + "step": 8601 + }, + { + "epoch": 0.9446518778827147, + "grad_norm": 2.073837995529175, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7084769010543823, + "num_tokens": 220772749.0, + "step": 8602 + }, + { + "epoch": 0.9447616955853284, + "grad_norm": 2.265158176422119, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7178685665130615, + "num_tokens": 220796837.0, + "step": 8603 + }, + { + "epoch": 0.944871513287942, + "grad_norm": 2.1920835971832275, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7209848165512085, + "num_tokens": 220820827.0, + "step": 8604 + }, + { + "epoch": 0.9449813309905557, + "grad_norm": 2.055591106414795, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6989172697067261, + "num_tokens": 220847430.0, + "step": 8605 + }, + { + "epoch": 0.9450911486931693, + "grad_norm": 2.141122579574585, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6923831701278687, + "num_tokens": 220875653.0, + "step": 8606 + }, + { + "epoch": 0.945200966395783, + "grad_norm": 2.0608019828796387, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6786167025566101, + "num_tokens": 220906381.0, + "step": 8607 + }, + { + "epoch": 0.9453107840983966, + "grad_norm": 1.777870535850525, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7004172205924988, + "num_tokens": 220939708.0, + "step": 8608 + }, + { + "epoch": 0.9454206018010103, + "grad_norm": 2.137291431427002, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6839393377304077, + "num_tokens": 220967679.0, + "step": 8609 + }, + { + "epoch": 0.9455304195036239, + "grad_norm": 2.107285976409912, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.685737133026123, + "num_tokens": 220995254.0, + "step": 8610 + }, + { + "epoch": 0.9456402372062377, + "grad_norm": 1.8777508735656738, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7111629247665405, + "num_tokens": 221025810.0, + "step": 8611 + }, + { + "epoch": 0.9457500549088513, + "grad_norm": 2.036921977996826, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7306516170501709, + "num_tokens": 221050235.0, + "step": 8612 + }, + { + "epoch": 0.945859872611465, + "grad_norm": 2.235283613204956, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.703175961971283, + "num_tokens": 221073753.0, + "step": 8613 + }, + { + "epoch": 0.9459696903140786, + "grad_norm": 2.8122329711914062, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7134392261505127, + "num_tokens": 221090233.0, + "step": 8614 + }, + { + "epoch": 0.9460795080166923, + "grad_norm": 2.0638248920440674, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6885030269622803, + "num_tokens": 221116453.0, + "step": 8615 + }, + { + "epoch": 0.9461893257193059, + "grad_norm": 2.2287986278533936, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6944838762283325, + "num_tokens": 221141415.0, + "step": 8616 + }, + { + "epoch": 0.9462991434219196, + "grad_norm": 2.3798749446868896, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.720237135887146, + "num_tokens": 221161476.0, + "step": 8617 + }, + { + "epoch": 0.9464089611245333, + "grad_norm": 2.041034460067749, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7008779048919678, + "num_tokens": 221188841.0, + "step": 8618 + }, + { + "epoch": 0.946518778827147, + "grad_norm": 2.196479320526123, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7222660779953003, + "num_tokens": 221212937.0, + "step": 8619 + }, + { + "epoch": 0.9466285965297606, + "grad_norm": 2.0579833984375, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6828041076660156, + "num_tokens": 221241390.0, + "step": 8620 + }, + { + "epoch": 0.9467384142323743, + "grad_norm": 1.8033498525619507, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6850773692131042, + "num_tokens": 221278880.0, + "step": 8621 + }, + { + "epoch": 0.9468482319349879, + "grad_norm": 2.5679595470428467, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7020245790481567, + "num_tokens": 221299896.0, + "step": 8622 + }, + { + "epoch": 0.9469580496376016, + "grad_norm": 2.1631884574890137, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6875411868095398, + "num_tokens": 221327930.0, + "step": 8623 + }, + { + "epoch": 0.9470678673402152, + "grad_norm": 2.148587465286255, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.714759111404419, + "num_tokens": 221354219.0, + "step": 8624 + }, + { + "epoch": 0.947177685042829, + "grad_norm": 2.1827428340911865, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7189103960990906, + "num_tokens": 221379928.0, + "step": 8625 + }, + { + "epoch": 0.9472875027454426, + "grad_norm": 2.0983083248138428, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7035239934921265, + "num_tokens": 221407891.0, + "step": 8626 + }, + { + "epoch": 0.9473973204480562, + "grad_norm": 2.3218162059783936, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7280275821685791, + "num_tokens": 221429642.0, + "step": 8627 + }, + { + "epoch": 0.9475071381506699, + "grad_norm": 2.2627103328704834, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7197725772857666, + "num_tokens": 221453723.0, + "step": 8628 + }, + { + "epoch": 0.9476169558532835, + "grad_norm": 2.187687635421753, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7146279811859131, + "num_tokens": 221480048.0, + "step": 8629 + }, + { + "epoch": 0.9477267735558972, + "grad_norm": 2.172466278076172, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7064343690872192, + "num_tokens": 221506309.0, + "step": 8630 + }, + { + "epoch": 0.9478365912585108, + "grad_norm": 2.1590938568115234, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6981542110443115, + "num_tokens": 221532476.0, + "step": 8631 + }, + { + "epoch": 0.9479464089611246, + "grad_norm": 2.6043760776519775, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7120099663734436, + "num_tokens": 221551951.0, + "step": 8632 + }, + { + "epoch": 0.9480562266637382, + "grad_norm": 2.59134840965271, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7269900441169739, + "num_tokens": 221571745.0, + "step": 8633 + }, + { + "epoch": 0.9481660443663519, + "grad_norm": 2.067880392074585, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7167121171951294, + "num_tokens": 221597560.0, + "step": 8634 + }, + { + "epoch": 0.9482758620689655, + "grad_norm": 2.2265262603759766, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7134647965431213, + "num_tokens": 221621625.0, + "step": 8635 + }, + { + "epoch": 0.9483856797715792, + "grad_norm": 2.331484317779541, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.714937686920166, + "num_tokens": 221644598.0, + "step": 8636 + }, + { + "epoch": 0.9484954974741928, + "grad_norm": 2.313309907913208, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7346140742301941, + "num_tokens": 221666036.0, + "step": 8637 + }, + { + "epoch": 0.9486053151768065, + "grad_norm": 2.138089179992676, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.700675368309021, + "num_tokens": 221693233.0, + "step": 8638 + }, + { + "epoch": 0.9487151328794201, + "grad_norm": 2.142827272415161, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7290037870407104, + "num_tokens": 221717018.0, + "step": 8639 + }, + { + "epoch": 0.9488249505820339, + "grad_norm": 2.6534383296966553, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7430800199508667, + "num_tokens": 221735071.0, + "step": 8640 + }, + { + "epoch": 0.9489347682846475, + "grad_norm": 2.3689565658569336, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.701058030128479, + "num_tokens": 221758457.0, + "step": 8641 + }, + { + "epoch": 0.9490445859872612, + "grad_norm": 2.377865791320801, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7092271447181702, + "num_tokens": 221779340.0, + "step": 8642 + }, + { + "epoch": 0.9491544036898748, + "grad_norm": 2.0803825855255127, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7210963368415833, + "num_tokens": 221804845.0, + "step": 8643 + }, + { + "epoch": 0.9492642213924884, + "grad_norm": 2.086838722229004, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7069116830825806, + "num_tokens": 221834545.0, + "step": 8644 + }, + { + "epoch": 0.9493740390951021, + "grad_norm": 2.1218197345733643, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6985039710998535, + "num_tokens": 221861527.0, + "step": 8645 + }, + { + "epoch": 0.9494838567977157, + "grad_norm": 2.155125379562378, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7185574769973755, + "num_tokens": 221885613.0, + "step": 8646 + }, + { + "epoch": 0.9495936745003295, + "grad_norm": 2.1840293407440186, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7044461965560913, + "num_tokens": 221910939.0, + "step": 8647 + }, + { + "epoch": 0.9497034922029431, + "grad_norm": 2.374565601348877, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7062706351280212, + "num_tokens": 221932686.0, + "step": 8648 + }, + { + "epoch": 0.9498133099055568, + "grad_norm": 2.003434658050537, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6868332624435425, + "num_tokens": 221960699.0, + "step": 8649 + }, + { + "epoch": 0.9499231276081704, + "grad_norm": 1.9775241613388062, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.7027242183685303, + "num_tokens": 221990277.0, + "step": 8650 + }, + { + "epoch": 0.9500329453107841, + "grad_norm": 2.2961478233337402, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6973422765731812, + "num_tokens": 222015025.0, + "step": 8651 + }, + { + "epoch": 0.9501427630133977, + "grad_norm": 2.3399271965026855, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7172375917434692, + "num_tokens": 222035401.0, + "step": 8652 + }, + { + "epoch": 0.9502525807160114, + "grad_norm": 2.221177816390991, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7099775075912476, + "num_tokens": 222058819.0, + "step": 8653 + }, + { + "epoch": 0.9503623984186251, + "grad_norm": 1.8466953039169312, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6978613138198853, + "num_tokens": 222091380.0, + "step": 8654 + }, + { + "epoch": 0.9504722161212388, + "grad_norm": 2.0765748023986816, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7071689963340759, + "num_tokens": 222119032.0, + "step": 8655 + }, + { + "epoch": 0.9505820338238524, + "grad_norm": 2.227804183959961, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6986774206161499, + "num_tokens": 222144393.0, + "step": 8656 + }, + { + "epoch": 0.9506918515264661, + "grad_norm": 1.9224181175231934, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6983200907707214, + "num_tokens": 222176005.0, + "step": 8657 + }, + { + "epoch": 0.9508016692290797, + "grad_norm": 2.099151134490967, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6995173096656799, + "num_tokens": 222201885.0, + "step": 8658 + }, + { + "epoch": 0.9509114869316934, + "grad_norm": 2.27361798286438, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7309393882751465, + "num_tokens": 222223949.0, + "step": 8659 + }, + { + "epoch": 0.951021304634307, + "grad_norm": 2.2698171138763428, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7194097638130188, + "num_tokens": 222246341.0, + "step": 8660 + }, + { + "epoch": 0.9511311223369208, + "grad_norm": 2.0989203453063965, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6961773633956909, + "num_tokens": 222272167.0, + "step": 8661 + }, + { + "epoch": 0.9512409400395344, + "grad_norm": 2.1816155910491943, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7113742828369141, + "num_tokens": 222295069.0, + "step": 8662 + }, + { + "epoch": 0.9513507577421481, + "grad_norm": 2.1652939319610596, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7063978910446167, + "num_tokens": 222319960.0, + "step": 8663 + }, + { + "epoch": 0.9514605754447617, + "grad_norm": 3.0370311737060547, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7299272418022156, + "num_tokens": 222337990.0, + "step": 8664 + }, + { + "epoch": 0.9515703931473753, + "grad_norm": 2.17946720123291, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6884976625442505, + "num_tokens": 222364185.0, + "step": 8665 + }, + { + "epoch": 0.951680210849989, + "grad_norm": 2.1222245693206787, + "learning_rate": 1e-06, + "loss": 1.0792, + "mean_token_accuracy": 0.6758447885513306, + "num_tokens": 222391362.0, + "step": 8666 + }, + { + "epoch": 0.9517900285526026, + "grad_norm": 2.3758816719055176, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7065579891204834, + "num_tokens": 222415888.0, + "step": 8667 + }, + { + "epoch": 0.9518998462552163, + "grad_norm": 2.194974422454834, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7117627859115601, + "num_tokens": 222440921.0, + "step": 8668 + }, + { + "epoch": 0.95200966395783, + "grad_norm": 2.3163464069366455, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7100048065185547, + "num_tokens": 222463974.0, + "step": 8669 + }, + { + "epoch": 0.9521194816604437, + "grad_norm": 2.125079870223999, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7076170444488525, + "num_tokens": 222491830.0, + "step": 8670 + }, + { + "epoch": 0.9522292993630573, + "grad_norm": 2.025517463684082, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7150976657867432, + "num_tokens": 222520840.0, + "step": 8671 + }, + { + "epoch": 0.952339117065671, + "grad_norm": 2.2090694904327393, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7079412341117859, + "num_tokens": 222546280.0, + "step": 8672 + }, + { + "epoch": 0.9524489347682846, + "grad_norm": 2.1820404529571533, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6999896764755249, + "num_tokens": 222573721.0, + "step": 8673 + }, + { + "epoch": 0.9525587524708983, + "grad_norm": 2.24884295463562, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7069244980812073, + "num_tokens": 222595332.0, + "step": 8674 + }, + { + "epoch": 0.9526685701735119, + "grad_norm": 2.0281407833099365, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7330918312072754, + "num_tokens": 222620253.0, + "step": 8675 + }, + { + "epoch": 0.9527783878761257, + "grad_norm": 1.9066046476364136, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.696715235710144, + "num_tokens": 222652902.0, + "step": 8676 + }, + { + "epoch": 0.9528882055787393, + "grad_norm": 2.4232637882232666, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7274860739707947, + "num_tokens": 222674131.0, + "step": 8677 + }, + { + "epoch": 0.952998023281353, + "grad_norm": 2.4350123405456543, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7264537811279297, + "num_tokens": 222694929.0, + "step": 8678 + }, + { + "epoch": 0.9531078409839666, + "grad_norm": 1.8100334405899048, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6895163655281067, + "num_tokens": 222729367.0, + "step": 8679 + }, + { + "epoch": 0.9532176586865803, + "grad_norm": 2.1763947010040283, + "learning_rate": 1e-06, + "loss": 1.0829, + "mean_token_accuracy": 0.66934734582901, + "num_tokens": 222758433.0, + "step": 8680 + }, + { + "epoch": 0.9533274763891939, + "grad_norm": 1.9073779582977295, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7055648565292358, + "num_tokens": 222790125.0, + "step": 8681 + }, + { + "epoch": 0.9534372940918076, + "grad_norm": 2.237072467803955, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.739909291267395, + "num_tokens": 222815424.0, + "step": 8682 + }, + { + "epoch": 0.9535471117944213, + "grad_norm": 2.033827304840088, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.690950870513916, + "num_tokens": 222843397.0, + "step": 8683 + }, + { + "epoch": 0.953656929497035, + "grad_norm": 2.1585044860839844, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.703002393245697, + "num_tokens": 222870009.0, + "step": 8684 + }, + { + "epoch": 0.9537667471996486, + "grad_norm": 1.9240068197250366, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7015300393104553, + "num_tokens": 222900844.0, + "step": 8685 + }, + { + "epoch": 0.9538765649022622, + "grad_norm": 2.0992538928985596, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6964377164840698, + "num_tokens": 222929367.0, + "step": 8686 + }, + { + "epoch": 0.9539863826048759, + "grad_norm": 2.1415059566497803, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7129022479057312, + "num_tokens": 222955627.0, + "step": 8687 + }, + { + "epoch": 0.9540962003074895, + "grad_norm": 2.166234016418457, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7010546326637268, + "num_tokens": 222980867.0, + "step": 8688 + }, + { + "epoch": 0.9542060180101032, + "grad_norm": 2.0570626258850098, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6939171552658081, + "num_tokens": 223008870.0, + "step": 8689 + }, + { + "epoch": 0.9543158357127169, + "grad_norm": 2.609182119369507, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7099119424819946, + "num_tokens": 223028383.0, + "step": 8690 + }, + { + "epoch": 0.9544256534153306, + "grad_norm": 2.227742910385132, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6968441009521484, + "num_tokens": 223052112.0, + "step": 8691 + }, + { + "epoch": 0.9545354711179442, + "grad_norm": 2.675248861312866, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7216920852661133, + "num_tokens": 223070091.0, + "step": 8692 + }, + { + "epoch": 0.9546452888205579, + "grad_norm": 2.307539701461792, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7315945029258728, + "num_tokens": 223091542.0, + "step": 8693 + }, + { + "epoch": 0.9547551065231715, + "grad_norm": 2.538917303085327, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7082528471946716, + "num_tokens": 223112706.0, + "step": 8694 + }, + { + "epoch": 0.9548649242257852, + "grad_norm": 2.2610127925872803, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7116076946258545, + "num_tokens": 223137419.0, + "step": 8695 + }, + { + "epoch": 0.9549747419283988, + "grad_norm": 2.293248414993286, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7185717821121216, + "num_tokens": 223159269.0, + "step": 8696 + }, + { + "epoch": 0.9550845596310125, + "grad_norm": 2.3369853496551514, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7199478149414062, + "num_tokens": 223182553.0, + "step": 8697 + }, + { + "epoch": 0.9551943773336262, + "grad_norm": 2.3774781227111816, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.726973831653595, + "num_tokens": 223204983.0, + "step": 8698 + }, + { + "epoch": 0.9553041950362399, + "grad_norm": 2.155756711959839, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6902636289596558, + "num_tokens": 223232351.0, + "step": 8699 + }, + { + "epoch": 0.9554140127388535, + "grad_norm": 2.344025135040283, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7304878234863281, + "num_tokens": 223255474.0, + "step": 8700 + }, + { + "epoch": 0.9555238304414672, + "grad_norm": 2.09798526763916, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7076230645179749, + "num_tokens": 223280024.0, + "step": 8701 + }, + { + "epoch": 0.9556336481440808, + "grad_norm": 2.3518903255462646, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6836416721343994, + "num_tokens": 223304610.0, + "step": 8702 + }, + { + "epoch": 0.9557434658466945, + "grad_norm": 2.0752100944519043, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.685798168182373, + "num_tokens": 223331789.0, + "step": 8703 + }, + { + "epoch": 0.9558532835493081, + "grad_norm": 2.0691463947296143, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6909986138343811, + "num_tokens": 223361711.0, + "step": 8704 + }, + { + "epoch": 0.9559631012519219, + "grad_norm": 2.1101369857788086, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7198022603988647, + "num_tokens": 223389359.0, + "step": 8705 + }, + { + "epoch": 0.9560729189545355, + "grad_norm": 2.0471463203430176, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7179794311523438, + "num_tokens": 223415910.0, + "step": 8706 + }, + { + "epoch": 0.9561827366571491, + "grad_norm": 1.9860284328460693, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.723770022392273, + "num_tokens": 223444269.0, + "step": 8707 + }, + { + "epoch": 0.9562925543597628, + "grad_norm": 2.2086760997772217, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7194116115570068, + "num_tokens": 223467838.0, + "step": 8708 + }, + { + "epoch": 0.9564023720623764, + "grad_norm": 1.9736045598983765, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7145211696624756, + "num_tokens": 223497845.0, + "step": 8709 + }, + { + "epoch": 0.9565121897649901, + "grad_norm": 2.1676299571990967, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7169576287269592, + "num_tokens": 223522692.0, + "step": 8710 + }, + { + "epoch": 0.9566220074676037, + "grad_norm": 1.9011189937591553, + "learning_rate": 1e-06, + "loss": 1.093, + "mean_token_accuracy": 0.6718721389770508, + "num_tokens": 223557234.0, + "step": 8711 + }, + { + "epoch": 0.9567318251702175, + "grad_norm": 2.12737774848938, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7427984476089478, + "num_tokens": 223580698.0, + "step": 8712 + }, + { + "epoch": 0.9568416428728311, + "grad_norm": 2.3735852241516113, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.711608350276947, + "num_tokens": 223603284.0, + "step": 8713 + }, + { + "epoch": 0.9569514605754448, + "grad_norm": 2.1425764560699463, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6930927038192749, + "num_tokens": 223631457.0, + "step": 8714 + }, + { + "epoch": 0.9570612782780584, + "grad_norm": 2.2550134658813477, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7260458469390869, + "num_tokens": 223654516.0, + "step": 8715 + }, + { + "epoch": 0.9571710959806721, + "grad_norm": 2.0097367763519287, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7414571046829224, + "num_tokens": 223681840.0, + "step": 8716 + }, + { + "epoch": 0.9572809136832857, + "grad_norm": 2.1813623905181885, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7209533452987671, + "num_tokens": 223706896.0, + "step": 8717 + }, + { + "epoch": 0.9573907313858994, + "grad_norm": 2.4771647453308105, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7100164890289307, + "num_tokens": 223729445.0, + "step": 8718 + }, + { + "epoch": 0.9575005490885131, + "grad_norm": 2.262760639190674, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6925239562988281, + "num_tokens": 223753866.0, + "step": 8719 + }, + { + "epoch": 0.9576103667911268, + "grad_norm": 1.9783326387405396, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7084460258483887, + "num_tokens": 223781299.0, + "step": 8720 + }, + { + "epoch": 0.9577201844937404, + "grad_norm": 1.9785164594650269, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7106921672821045, + "num_tokens": 223808583.0, + "step": 8721 + }, + { + "epoch": 0.9578300021963541, + "grad_norm": 2.0476346015930176, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7145284414291382, + "num_tokens": 223833827.0, + "step": 8722 + }, + { + "epoch": 0.9579398198989677, + "grad_norm": 2.152592182159424, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.7009922862052917, + "num_tokens": 223861425.0, + "step": 8723 + }, + { + "epoch": 0.9580496376015813, + "grad_norm": 1.9009026288986206, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6824577450752258, + "num_tokens": 223895129.0, + "step": 8724 + }, + { + "epoch": 0.958159455304195, + "grad_norm": 2.2699408531188965, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7243834733963013, + "num_tokens": 223917462.0, + "step": 8725 + }, + { + "epoch": 0.9582692730068088, + "grad_norm": 2.225224256515503, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6910985708236694, + "num_tokens": 223945037.0, + "step": 8726 + }, + { + "epoch": 0.9583790907094224, + "grad_norm": 2.0145885944366455, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6812260746955872, + "num_tokens": 223974411.0, + "step": 8727 + }, + { + "epoch": 0.958488908412036, + "grad_norm": 2.0811426639556885, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7227586507797241, + "num_tokens": 223999506.0, + "step": 8728 + }, + { + "epoch": 0.9585987261146497, + "grad_norm": 2.076429605484009, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7214990258216858, + "num_tokens": 224026442.0, + "step": 8729 + }, + { + "epoch": 0.9587085438172633, + "grad_norm": 2.154230833053589, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7068363428115845, + "num_tokens": 224054451.0, + "step": 8730 + }, + { + "epoch": 0.958818361519877, + "grad_norm": 2.058793306350708, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7047528028488159, + "num_tokens": 224080935.0, + "step": 8731 + }, + { + "epoch": 0.9589281792224906, + "grad_norm": 2.284053087234497, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7069222927093506, + "num_tokens": 224105731.0, + "step": 8732 + }, + { + "epoch": 0.9590379969251043, + "grad_norm": 2.060027599334717, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7150534391403198, + "num_tokens": 224134352.0, + "step": 8733 + }, + { + "epoch": 0.959147814627718, + "grad_norm": 2.085162878036499, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7107535004615784, + "num_tokens": 224160172.0, + "step": 8734 + }, + { + "epoch": 0.9592576323303317, + "grad_norm": 2.1042048931121826, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6957898736000061, + "num_tokens": 224187163.0, + "step": 8735 + }, + { + "epoch": 0.9593674500329453, + "grad_norm": 1.8572032451629639, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6953393220901489, + "num_tokens": 224222604.0, + "step": 8736 + }, + { + "epoch": 0.959477267735559, + "grad_norm": 1.982459545135498, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.6993736624717712, + "num_tokens": 224254491.0, + "step": 8737 + }, + { + "epoch": 0.9595870854381726, + "grad_norm": 1.9482805728912354, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7287968397140503, + "num_tokens": 224284260.0, + "step": 8738 + }, + { + "epoch": 0.9596969031407863, + "grad_norm": 1.8891726732254028, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6850887537002563, + "num_tokens": 224315907.0, + "step": 8739 + }, + { + "epoch": 0.9598067208433999, + "grad_norm": 2.3222806453704834, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7194746732711792, + "num_tokens": 224339642.0, + "step": 8740 + }, + { + "epoch": 0.9599165385460137, + "grad_norm": 2.466649055480957, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7210210561752319, + "num_tokens": 224361229.0, + "step": 8741 + }, + { + "epoch": 0.9600263562486273, + "grad_norm": 2.242293119430542, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7069076299667358, + "num_tokens": 224385344.0, + "step": 8742 + }, + { + "epoch": 0.960136173951241, + "grad_norm": 2.27645206451416, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6988193988800049, + "num_tokens": 224411858.0, + "step": 8743 + }, + { + "epoch": 0.9602459916538546, + "grad_norm": 2.272289276123047, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7272494435310364, + "num_tokens": 224433727.0, + "step": 8744 + }, + { + "epoch": 0.9603558093564682, + "grad_norm": 2.1772654056549072, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.732012152671814, + "num_tokens": 224458150.0, + "step": 8745 + }, + { + "epoch": 0.9604656270590819, + "grad_norm": 2.423130512237549, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7267060875892639, + "num_tokens": 224476916.0, + "step": 8746 + }, + { + "epoch": 0.9605754447616955, + "grad_norm": 2.21370267868042, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7172510623931885, + "num_tokens": 224501709.0, + "step": 8747 + }, + { + "epoch": 0.9606852624643093, + "grad_norm": 2.2128124237060547, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7025320529937744, + "num_tokens": 224527793.0, + "step": 8748 + }, + { + "epoch": 0.9607950801669229, + "grad_norm": 2.1730728149414062, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.703978419303894, + "num_tokens": 224554011.0, + "step": 8749 + }, + { + "epoch": 0.9609048978695366, + "grad_norm": 2.2925870418548584, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.707154393196106, + "num_tokens": 224576660.0, + "step": 8750 + }, + { + "epoch": 0.9610147155721502, + "grad_norm": 2.2630889415740967, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7166372537612915, + "num_tokens": 224599427.0, + "step": 8751 + }, + { + "epoch": 0.9611245332747639, + "grad_norm": 2.319460391998291, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6963125467300415, + "num_tokens": 224622704.0, + "step": 8752 + }, + { + "epoch": 0.9612343509773775, + "grad_norm": 2.3016607761383057, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7247656583786011, + "num_tokens": 224645011.0, + "step": 8753 + }, + { + "epoch": 0.9613441686799912, + "grad_norm": 2.4065821170806885, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7151901721954346, + "num_tokens": 224666533.0, + "step": 8754 + }, + { + "epoch": 0.9614539863826049, + "grad_norm": 2.228529691696167, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.712726891040802, + "num_tokens": 224690000.0, + "step": 8755 + }, + { + "epoch": 0.9615638040852186, + "grad_norm": 2.0242762565612793, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.748979926109314, + "num_tokens": 224715904.0, + "step": 8756 + }, + { + "epoch": 0.9616736217878322, + "grad_norm": 1.9851218461990356, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7098167538642883, + "num_tokens": 224746127.0, + "step": 8757 + }, + { + "epoch": 0.9617834394904459, + "grad_norm": 2.119313955307007, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6970899701118469, + "num_tokens": 224770564.0, + "step": 8758 + }, + { + "epoch": 0.9618932571930595, + "grad_norm": 2.1731059551239014, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6921752691268921, + "num_tokens": 224797050.0, + "step": 8759 + }, + { + "epoch": 0.9620030748956732, + "grad_norm": 1.925026774406433, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7236250042915344, + "num_tokens": 224825039.0, + "step": 8760 + }, + { + "epoch": 0.9621128925982868, + "grad_norm": 2.2651848793029785, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7114081382751465, + "num_tokens": 224848612.0, + "step": 8761 + }, + { + "epoch": 0.9622227103009005, + "grad_norm": 1.911662220954895, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7042366862297058, + "num_tokens": 224881311.0, + "step": 8762 + }, + { + "epoch": 0.9623325280035142, + "grad_norm": 2.135301351547241, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6908628940582275, + "num_tokens": 224907067.0, + "step": 8763 + }, + { + "epoch": 0.9624423457061279, + "grad_norm": 2.2254793643951416, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7060869932174683, + "num_tokens": 224931287.0, + "step": 8764 + }, + { + "epoch": 0.9625521634087415, + "grad_norm": 2.150142192840576, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7282090187072754, + "num_tokens": 224956785.0, + "step": 8765 + }, + { + "epoch": 0.9626619811113551, + "grad_norm": 2.1693789958953857, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6829609274864197, + "num_tokens": 224983682.0, + "step": 8766 + }, + { + "epoch": 0.9627717988139688, + "grad_norm": 2.3128974437713623, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7195260524749756, + "num_tokens": 225005972.0, + "step": 8767 + }, + { + "epoch": 0.9628816165165824, + "grad_norm": 2.1150825023651123, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7027463316917419, + "num_tokens": 225032562.0, + "step": 8768 + }, + { + "epoch": 0.9629914342191961, + "grad_norm": 1.9063488245010376, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.70976722240448, + "num_tokens": 225064827.0, + "step": 8769 + }, + { + "epoch": 0.9631012519218098, + "grad_norm": 2.0816218852996826, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7165756821632385, + "num_tokens": 225091776.0, + "step": 8770 + }, + { + "epoch": 0.9632110696244235, + "grad_norm": 2.261976480484009, + "learning_rate": 1e-06, + "loss": 1.0604, + "mean_token_accuracy": 0.6846009492874146, + "num_tokens": 225117763.0, + "step": 8771 + }, + { + "epoch": 0.9633208873270371, + "grad_norm": 2.0716824531555176, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.708337664604187, + "num_tokens": 225147019.0, + "step": 8772 + }, + { + "epoch": 0.9634307050296508, + "grad_norm": 2.2831766605377197, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7113556265830994, + "num_tokens": 225169657.0, + "step": 8773 + }, + { + "epoch": 0.9635405227322644, + "grad_norm": 2.058631420135498, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7060824632644653, + "num_tokens": 225196846.0, + "step": 8774 + }, + { + "epoch": 0.9636503404348781, + "grad_norm": 1.9591138362884521, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7013176083564758, + "num_tokens": 225226558.0, + "step": 8775 + }, + { + "epoch": 0.9637601581374917, + "grad_norm": 2.194368839263916, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7162032127380371, + "num_tokens": 225249361.0, + "step": 8776 + }, + { + "epoch": 0.9638699758401055, + "grad_norm": 2.3660402297973633, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7139060497283936, + "num_tokens": 225271642.0, + "step": 8777 + }, + { + "epoch": 0.9639797935427191, + "grad_norm": 2.383622646331787, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7054908871650696, + "num_tokens": 225295459.0, + "step": 8778 + }, + { + "epoch": 0.9640896112453328, + "grad_norm": 2.3640167713165283, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7190563678741455, + "num_tokens": 225316756.0, + "step": 8779 + }, + { + "epoch": 0.9641994289479464, + "grad_norm": 1.914626955986023, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6870337724685669, + "num_tokens": 225349031.0, + "step": 8780 + }, + { + "epoch": 0.9643092466505601, + "grad_norm": 1.8794641494750977, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7119802832603455, + "num_tokens": 225380023.0, + "step": 8781 + }, + { + "epoch": 0.9644190643531737, + "grad_norm": 2.407614231109619, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7053952217102051, + "num_tokens": 225402223.0, + "step": 8782 + }, + { + "epoch": 0.9645288820557874, + "grad_norm": 2.5538463592529297, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7008249759674072, + "num_tokens": 225423664.0, + "step": 8783 + }, + { + "epoch": 0.9646386997584011, + "grad_norm": 2.3854873180389404, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7093146443367004, + "num_tokens": 225446789.0, + "step": 8784 + }, + { + "epoch": 0.9647485174610148, + "grad_norm": 2.2162134647369385, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7012789845466614, + "num_tokens": 225471337.0, + "step": 8785 + }, + { + "epoch": 0.9648583351636284, + "grad_norm": 2.1214826107025146, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7028873562812805, + "num_tokens": 225499081.0, + "step": 8786 + }, + { + "epoch": 0.964968152866242, + "grad_norm": 2.3160736560821533, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7045679092407227, + "num_tokens": 225522728.0, + "step": 8787 + }, + { + "epoch": 0.9650779705688557, + "grad_norm": 2.266162157058716, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7155709266662598, + "num_tokens": 225544642.0, + "step": 8788 + }, + { + "epoch": 0.9651877882714693, + "grad_norm": 2.1593523025512695, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7153336405754089, + "num_tokens": 225569351.0, + "step": 8789 + }, + { + "epoch": 0.965297605974083, + "grad_norm": 2.264760732650757, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7115221619606018, + "num_tokens": 225593300.0, + "step": 8790 + }, + { + "epoch": 0.9654074236766966, + "grad_norm": 2.414665937423706, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7075311541557312, + "num_tokens": 225614946.0, + "step": 8791 + }, + { + "epoch": 0.9655172413793104, + "grad_norm": 2.5114591121673584, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7476599216461182, + "num_tokens": 225634091.0, + "step": 8792 + }, + { + "epoch": 0.965627059081924, + "grad_norm": 2.354916572570801, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.709089457988739, + "num_tokens": 225656906.0, + "step": 8793 + }, + { + "epoch": 0.9657368767845377, + "grad_norm": 2.0784389972686768, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7133917808532715, + "num_tokens": 225685781.0, + "step": 8794 + }, + { + "epoch": 0.9658466944871513, + "grad_norm": 2.054560422897339, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7112305164337158, + "num_tokens": 225713583.0, + "step": 8795 + }, + { + "epoch": 0.965956512189765, + "grad_norm": 2.190699577331543, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7033980488777161, + "num_tokens": 225740419.0, + "step": 8796 + }, + { + "epoch": 0.9660663298923786, + "grad_norm": 2.324549913406372, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7338408827781677, + "num_tokens": 225761557.0, + "step": 8797 + }, + { + "epoch": 0.9661761475949923, + "grad_norm": 2.264575481414795, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7192829847335815, + "num_tokens": 225783921.0, + "step": 8798 + }, + { + "epoch": 0.966285965297606, + "grad_norm": 2.1014649868011475, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7052216529846191, + "num_tokens": 225811117.0, + "step": 8799 + }, + { + "epoch": 0.9663957830002197, + "grad_norm": 2.0785579681396484, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7318457365036011, + "num_tokens": 225835477.0, + "step": 8800 + }, + { + "epoch": 0.9665056007028333, + "grad_norm": 2.3593623638153076, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7059872150421143, + "num_tokens": 225858827.0, + "step": 8801 + }, + { + "epoch": 0.966615418405447, + "grad_norm": 2.1624972820281982, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7106104493141174, + "num_tokens": 225883138.0, + "step": 8802 + }, + { + "epoch": 0.9667252361080606, + "grad_norm": 2.3498623371124268, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7242430448532104, + "num_tokens": 225904617.0, + "step": 8803 + }, + { + "epoch": 0.9668350538106742, + "grad_norm": 2.513618230819702, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7220503687858582, + "num_tokens": 225923856.0, + "step": 8804 + }, + { + "epoch": 0.9669448715132879, + "grad_norm": 2.276409387588501, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.729755163192749, + "num_tokens": 225945749.0, + "step": 8805 + }, + { + "epoch": 0.9670546892159017, + "grad_norm": 2.2159433364868164, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7334141731262207, + "num_tokens": 225968836.0, + "step": 8806 + }, + { + "epoch": 0.9671645069185153, + "grad_norm": 2.104954957962036, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7075289487838745, + "num_tokens": 225995775.0, + "step": 8807 + }, + { + "epoch": 0.967274324621129, + "grad_norm": 2.0793814659118652, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6894123554229736, + "num_tokens": 226026071.0, + "step": 8808 + }, + { + "epoch": 0.9673841423237426, + "grad_norm": 2.290405511856079, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7118620872497559, + "num_tokens": 226049534.0, + "step": 8809 + }, + { + "epoch": 0.9674939600263562, + "grad_norm": 2.235161066055298, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7146975994110107, + "num_tokens": 226072983.0, + "step": 8810 + }, + { + "epoch": 0.9676037777289699, + "grad_norm": 1.991909146308899, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7292742729187012, + "num_tokens": 226100261.0, + "step": 8811 + }, + { + "epoch": 0.9677135954315835, + "grad_norm": 2.557028293609619, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.715693473815918, + "num_tokens": 226120130.0, + "step": 8812 + }, + { + "epoch": 0.9678234131341973, + "grad_norm": 2.139517068862915, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7158117890357971, + "num_tokens": 226148311.0, + "step": 8813 + }, + { + "epoch": 0.9679332308368109, + "grad_norm": 2.165158987045288, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7290261387825012, + "num_tokens": 226174837.0, + "step": 8814 + }, + { + "epoch": 0.9680430485394246, + "grad_norm": 2.2675602436065674, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7091407775878906, + "num_tokens": 226199737.0, + "step": 8815 + }, + { + "epoch": 0.9681528662420382, + "grad_norm": 2.243255376815796, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7221099138259888, + "num_tokens": 226222566.0, + "step": 8816 + }, + { + "epoch": 0.9682626839446519, + "grad_norm": 2.2472054958343506, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7049870491027832, + "num_tokens": 226245959.0, + "step": 8817 + }, + { + "epoch": 0.9683725016472655, + "grad_norm": 2.1089928150177, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7227553725242615, + "num_tokens": 226273355.0, + "step": 8818 + }, + { + "epoch": 0.9684823193498792, + "grad_norm": 2.311249256134033, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6905211210250854, + "num_tokens": 226297936.0, + "step": 8819 + }, + { + "epoch": 0.9685921370524928, + "grad_norm": 1.9931824207305908, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7182611227035522, + "num_tokens": 226325363.0, + "step": 8820 + }, + { + "epoch": 0.9687019547551066, + "grad_norm": 2.218435764312744, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6880494356155396, + "num_tokens": 226349429.0, + "step": 8821 + }, + { + "epoch": 0.9688117724577202, + "grad_norm": 2.146907329559326, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.716941773891449, + "num_tokens": 226373322.0, + "step": 8822 + }, + { + "epoch": 0.9689215901603339, + "grad_norm": 2.269432306289673, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7302640080451965, + "num_tokens": 226396088.0, + "step": 8823 + }, + { + "epoch": 0.9690314078629475, + "grad_norm": 2.454792022705078, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7199861407279968, + "num_tokens": 226416931.0, + "step": 8824 + }, + { + "epoch": 0.9691412255655611, + "grad_norm": 2.191302537918091, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6941500902175903, + "num_tokens": 226442293.0, + "step": 8825 + }, + { + "epoch": 0.9692510432681748, + "grad_norm": 2.109018325805664, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6847501397132874, + "num_tokens": 226469972.0, + "step": 8826 + }, + { + "epoch": 0.9693608609707884, + "grad_norm": 2.226022720336914, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7146172523498535, + "num_tokens": 226494426.0, + "step": 8827 + }, + { + "epoch": 0.9694706786734022, + "grad_norm": 2.11787748336792, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7441941499710083, + "num_tokens": 226519133.0, + "step": 8828 + }, + { + "epoch": 0.9695804963760158, + "grad_norm": 2.0511155128479004, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7255045175552368, + "num_tokens": 226547743.0, + "step": 8829 + }, + { + "epoch": 0.9696903140786295, + "grad_norm": 2.106870174407959, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.710520327091217, + "num_tokens": 226574525.0, + "step": 8830 + }, + { + "epoch": 0.9698001317812431, + "grad_norm": 2.1342926025390625, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6964980363845825, + "num_tokens": 226602754.0, + "step": 8831 + }, + { + "epoch": 0.9699099494838568, + "grad_norm": 2.0535645484924316, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6931501626968384, + "num_tokens": 226634030.0, + "step": 8832 + }, + { + "epoch": 0.9700197671864704, + "grad_norm": 2.4870054721832275, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7152702212333679, + "num_tokens": 226655945.0, + "step": 8833 + }, + { + "epoch": 0.9701295848890841, + "grad_norm": 2.1434528827667236, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7172223925590515, + "num_tokens": 226680610.0, + "step": 8834 + }, + { + "epoch": 0.9702394025916978, + "grad_norm": 2.3212647438049316, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7330373525619507, + "num_tokens": 226702988.0, + "step": 8835 + }, + { + "epoch": 0.9703492202943115, + "grad_norm": 2.061925172805786, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.7046323418617249, + "num_tokens": 226731717.0, + "step": 8836 + }, + { + "epoch": 0.9704590379969251, + "grad_norm": 1.919150710105896, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.7044988870620728, + "num_tokens": 226762788.0, + "step": 8837 + }, + { + "epoch": 0.9705688556995388, + "grad_norm": 2.075925350189209, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7248773574829102, + "num_tokens": 226788971.0, + "step": 8838 + }, + { + "epoch": 0.9706786734021524, + "grad_norm": 2.3237836360931396, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6908082365989685, + "num_tokens": 226812599.0, + "step": 8839 + }, + { + "epoch": 0.9707884911047661, + "grad_norm": 2.05558443069458, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7093257904052734, + "num_tokens": 226839907.0, + "step": 8840 + }, + { + "epoch": 0.9708983088073797, + "grad_norm": 2.1657440662384033, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.711322546005249, + "num_tokens": 226865491.0, + "step": 8841 + }, + { + "epoch": 0.9710081265099935, + "grad_norm": 2.1643283367156982, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7026317119598389, + "num_tokens": 226890464.0, + "step": 8842 + }, + { + "epoch": 0.9711179442126071, + "grad_norm": 1.888482928276062, + "learning_rate": 1e-06, + "loss": 1.0785, + "mean_token_accuracy": 0.687143087387085, + "num_tokens": 226925285.0, + "step": 8843 + }, + { + "epoch": 0.9712277619152208, + "grad_norm": 1.855303168296814, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6921699643135071, + "num_tokens": 226959218.0, + "step": 8844 + }, + { + "epoch": 0.9713375796178344, + "grad_norm": 2.232902765274048, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7109197378158569, + "num_tokens": 226982724.0, + "step": 8845 + }, + { + "epoch": 0.971447397320448, + "grad_norm": 2.263129472732544, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.703904926776886, + "num_tokens": 227006938.0, + "step": 8846 + }, + { + "epoch": 0.9715572150230617, + "grad_norm": 2.0955729484558105, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6945658922195435, + "num_tokens": 227034109.0, + "step": 8847 + }, + { + "epoch": 0.9716670327256753, + "grad_norm": 2.3998098373413086, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7143930196762085, + "num_tokens": 227055530.0, + "step": 8848 + }, + { + "epoch": 0.971776850428289, + "grad_norm": 1.9062455892562866, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7222670316696167, + "num_tokens": 227084725.0, + "step": 8849 + }, + { + "epoch": 0.9718866681309027, + "grad_norm": 2.400597095489502, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7226145267486572, + "num_tokens": 227104523.0, + "step": 8850 + }, + { + "epoch": 0.9719964858335164, + "grad_norm": 1.9786016941070557, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7042146325111389, + "num_tokens": 227136863.0, + "step": 8851 + }, + { + "epoch": 0.97210630353613, + "grad_norm": 2.0245001316070557, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.6955551505088806, + "num_tokens": 227162419.0, + "step": 8852 + }, + { + "epoch": 0.9722161212387437, + "grad_norm": 1.7559640407562256, + "learning_rate": 1e-06, + "loss": 1.0844, + "mean_token_accuracy": 0.6796069741249084, + "num_tokens": 227198695.0, + "step": 8853 + }, + { + "epoch": 0.9723259389413573, + "grad_norm": 2.387162208557129, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7167292833328247, + "num_tokens": 227219542.0, + "step": 8854 + }, + { + "epoch": 0.972435756643971, + "grad_norm": 2.173154830932617, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7416780591011047, + "num_tokens": 227244478.0, + "step": 8855 + }, + { + "epoch": 0.9725455743465846, + "grad_norm": 2.1628572940826416, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7372074127197266, + "num_tokens": 227268417.0, + "step": 8856 + }, + { + "epoch": 0.9726553920491984, + "grad_norm": 1.8325077295303345, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6901910305023193, + "num_tokens": 227302912.0, + "step": 8857 + }, + { + "epoch": 0.972765209751812, + "grad_norm": 2.188839912414551, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7072116136550903, + "num_tokens": 227329080.0, + "step": 8858 + }, + { + "epoch": 0.9728750274544257, + "grad_norm": 2.256249189376831, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7016547322273254, + "num_tokens": 227353328.0, + "step": 8859 + }, + { + "epoch": 0.9729848451570393, + "grad_norm": 2.1416573524475098, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7069844007492065, + "num_tokens": 227379853.0, + "step": 8860 + }, + { + "epoch": 0.973094662859653, + "grad_norm": 2.1238863468170166, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7224945425987244, + "num_tokens": 227405361.0, + "step": 8861 + }, + { + "epoch": 0.9732044805622666, + "grad_norm": 2.126046895980835, + "learning_rate": 1e-06, + "loss": 1.0631, + "mean_token_accuracy": 0.6774361729621887, + "num_tokens": 227433704.0, + "step": 8862 + }, + { + "epoch": 0.9733142982648803, + "grad_norm": 2.4517135620117188, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7291024923324585, + "num_tokens": 227453705.0, + "step": 8863 + }, + { + "epoch": 0.973424115967494, + "grad_norm": 2.409414052963257, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7022147178649902, + "num_tokens": 227476802.0, + "step": 8864 + }, + { + "epoch": 0.9735339336701077, + "grad_norm": 2.125288486480713, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.6998767852783203, + "num_tokens": 227503667.0, + "step": 8865 + }, + { + "epoch": 0.9736437513727213, + "grad_norm": 2.0974388122558594, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7043498754501343, + "num_tokens": 227529482.0, + "step": 8866 + }, + { + "epoch": 0.973753569075335, + "grad_norm": 2.1608455181121826, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6961170434951782, + "num_tokens": 227557425.0, + "step": 8867 + }, + { + "epoch": 0.9738633867779486, + "grad_norm": 2.2137198448181152, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6843782663345337, + "num_tokens": 227583564.0, + "step": 8868 + }, + { + "epoch": 0.9739732044805622, + "grad_norm": 2.2709813117980957, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7145414352416992, + "num_tokens": 227607903.0, + "step": 8869 + }, + { + "epoch": 0.9740830221831759, + "grad_norm": 2.369844913482666, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.706118643283844, + "num_tokens": 227628495.0, + "step": 8870 + }, + { + "epoch": 0.9741928398857896, + "grad_norm": 2.300429344177246, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.707132875919342, + "num_tokens": 227652356.0, + "step": 8871 + }, + { + "epoch": 0.9743026575884033, + "grad_norm": 2.307175874710083, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.735637903213501, + "num_tokens": 227675460.0, + "step": 8872 + }, + { + "epoch": 0.9744124752910169, + "grad_norm": 2.1084699630737305, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6914609670639038, + "num_tokens": 227702291.0, + "step": 8873 + }, + { + "epoch": 0.9745222929936306, + "grad_norm": 2.2477028369903564, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7181196808815002, + "num_tokens": 227725307.0, + "step": 8874 + }, + { + "epoch": 0.9746321106962442, + "grad_norm": 2.3913748264312744, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.734245777130127, + "num_tokens": 227746528.0, + "step": 8875 + }, + { + "epoch": 0.9747419283988579, + "grad_norm": 2.20414400100708, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6989418864250183, + "num_tokens": 227772162.0, + "step": 8876 + }, + { + "epoch": 0.9748517461014715, + "grad_norm": 2.040294885635376, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6890470385551453, + "num_tokens": 227800110.0, + "step": 8877 + }, + { + "epoch": 0.9749615638040853, + "grad_norm": 2.3661322593688965, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6987959146499634, + "num_tokens": 227822280.0, + "step": 8878 + }, + { + "epoch": 0.9750713815066989, + "grad_norm": 2.481985092163086, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7128087878227234, + "num_tokens": 227842471.0, + "step": 8879 + }, + { + "epoch": 0.9751811992093126, + "grad_norm": 2.2917232513427734, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7112387418746948, + "num_tokens": 227864061.0, + "step": 8880 + }, + { + "epoch": 0.9752910169119262, + "grad_norm": 2.0550899505615234, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.71429443359375, + "num_tokens": 227891978.0, + "step": 8881 + }, + { + "epoch": 0.9754008346145399, + "grad_norm": 2.3557240962982178, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7026964426040649, + "num_tokens": 227915741.0, + "step": 8882 + }, + { + "epoch": 0.9755106523171535, + "grad_norm": 2.1008739471435547, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7104280591011047, + "num_tokens": 227942727.0, + "step": 8883 + }, + { + "epoch": 0.9756204700197671, + "grad_norm": 2.4414446353912354, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7070464491844177, + "num_tokens": 227963903.0, + "step": 8884 + }, + { + "epoch": 0.9757302877223808, + "grad_norm": 2.3109583854675293, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.71783447265625, + "num_tokens": 227988082.0, + "step": 8885 + }, + { + "epoch": 0.9758401054249946, + "grad_norm": 2.2794253826141357, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6928077936172485, + "num_tokens": 228012632.0, + "step": 8886 + }, + { + "epoch": 0.9759499231276082, + "grad_norm": 1.9229050874710083, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.685602068901062, + "num_tokens": 228046559.0, + "step": 8887 + }, + { + "epoch": 0.9760597408302218, + "grad_norm": 1.9831947088241577, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7159357070922852, + "num_tokens": 228077381.0, + "step": 8888 + }, + { + "epoch": 0.9761695585328355, + "grad_norm": 1.9466661214828491, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6983464956283569, + "num_tokens": 228108250.0, + "step": 8889 + }, + { + "epoch": 0.9762793762354491, + "grad_norm": 2.0672292709350586, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6954753398895264, + "num_tokens": 228138253.0, + "step": 8890 + }, + { + "epoch": 0.9763891939380628, + "grad_norm": 2.279271364212036, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7279538512229919, + "num_tokens": 228161614.0, + "step": 8891 + }, + { + "epoch": 0.9764990116406764, + "grad_norm": 2.0309770107269287, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7168353199958801, + "num_tokens": 228189436.0, + "step": 8892 + }, + { + "epoch": 0.9766088293432902, + "grad_norm": 2.4631454944610596, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7276585102081299, + "num_tokens": 228212048.0, + "step": 8893 + }, + { + "epoch": 0.9767186470459038, + "grad_norm": 2.08980655670166, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.715704083442688, + "num_tokens": 228238393.0, + "step": 8894 + }, + { + "epoch": 0.9768284647485175, + "grad_norm": 2.2865424156188965, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7310127019882202, + "num_tokens": 228260756.0, + "step": 8895 + }, + { + "epoch": 0.9769382824511311, + "grad_norm": 2.117608070373535, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.716352105140686, + "num_tokens": 228286860.0, + "step": 8896 + }, + { + "epoch": 0.9770481001537448, + "grad_norm": 2.0430350303649902, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7254552841186523, + "num_tokens": 228315294.0, + "step": 8897 + }, + { + "epoch": 0.9771579178563584, + "grad_norm": 1.9953560829162598, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7009535431861877, + "num_tokens": 228343379.0, + "step": 8898 + }, + { + "epoch": 0.9772677355589721, + "grad_norm": 2.129178047180176, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7005855441093445, + "num_tokens": 228369984.0, + "step": 8899 + }, + { + "epoch": 0.9773775532615858, + "grad_norm": 2.260133981704712, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7113571166992188, + "num_tokens": 228393997.0, + "step": 8900 + }, + { + "epoch": 0.9774873709641995, + "grad_norm": 2.2914161682128906, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7106842994689941, + "num_tokens": 228417182.0, + "step": 8901 + }, + { + "epoch": 0.9775971886668131, + "grad_norm": 2.1740946769714355, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7136380076408386, + "num_tokens": 228443453.0, + "step": 8902 + }, + { + "epoch": 0.9777070063694268, + "grad_norm": 2.2531068325042725, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7036240100860596, + "num_tokens": 228467742.0, + "step": 8903 + }, + { + "epoch": 0.9778168240720404, + "grad_norm": 1.9297667741775513, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7179778218269348, + "num_tokens": 228498143.0, + "step": 8904 + }, + { + "epoch": 0.977926641774654, + "grad_norm": 2.423208236694336, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7143537998199463, + "num_tokens": 228519931.0, + "step": 8905 + }, + { + "epoch": 0.9780364594772677, + "grad_norm": 2.2975516319274902, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.707263171672821, + "num_tokens": 228543467.0, + "step": 8906 + }, + { + "epoch": 0.9781462771798815, + "grad_norm": 1.925870418548584, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6818133592605591, + "num_tokens": 228575895.0, + "step": 8907 + }, + { + "epoch": 0.9782560948824951, + "grad_norm": 2.8582441806793213, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.71417236328125, + "num_tokens": 228591722.0, + "step": 8908 + }, + { + "epoch": 0.9783659125851087, + "grad_norm": 2.0068910121917725, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.724655270576477, + "num_tokens": 228620712.0, + "step": 8909 + }, + { + "epoch": 0.9784757302877224, + "grad_norm": 2.2140579223632812, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7227161526679993, + "num_tokens": 228644317.0, + "step": 8910 + }, + { + "epoch": 0.978585547990336, + "grad_norm": 1.8580455780029297, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6940441727638245, + "num_tokens": 228678219.0, + "step": 8911 + }, + { + "epoch": 0.9786953656929497, + "grad_norm": 2.133219003677368, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6842260956764221, + "num_tokens": 228706155.0, + "step": 8912 + }, + { + "epoch": 0.9788051833955633, + "grad_norm": 2.1558279991149902, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7138145565986633, + "num_tokens": 228731695.0, + "step": 8913 + }, + { + "epoch": 0.978915001098177, + "grad_norm": 2.188065528869629, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7336463332176208, + "num_tokens": 228755113.0, + "step": 8914 + }, + { + "epoch": 0.9790248188007907, + "grad_norm": 2.016958236694336, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7045700550079346, + "num_tokens": 228784219.0, + "step": 8915 + }, + { + "epoch": 0.9791346365034044, + "grad_norm": 2.2157676219940186, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6881841421127319, + "num_tokens": 228808778.0, + "step": 8916 + }, + { + "epoch": 0.979244454206018, + "grad_norm": 2.2189157009124756, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6997223496437073, + "num_tokens": 228834351.0, + "step": 8917 + }, + { + "epoch": 0.9793542719086317, + "grad_norm": 1.8570225238800049, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7317193746566772, + "num_tokens": 228866863.0, + "step": 8918 + }, + { + "epoch": 0.9794640896112453, + "grad_norm": 1.9446296691894531, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7009363174438477, + "num_tokens": 228899351.0, + "step": 8919 + }, + { + "epoch": 0.979573907313859, + "grad_norm": 2.065161943435669, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.718836784362793, + "num_tokens": 228925564.0, + "step": 8920 + }, + { + "epoch": 0.9796837250164726, + "grad_norm": 2.2137627601623535, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7023501992225647, + "num_tokens": 228951409.0, + "step": 8921 + }, + { + "epoch": 0.9797935427190864, + "grad_norm": 2.1583175659179688, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7016160488128662, + "num_tokens": 228977243.0, + "step": 8922 + }, + { + "epoch": 0.9799033604217, + "grad_norm": 2.0862526893615723, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6998945474624634, + "num_tokens": 229006041.0, + "step": 8923 + }, + { + "epoch": 0.9800131781243137, + "grad_norm": 2.021895170211792, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7221074104309082, + "num_tokens": 229032751.0, + "step": 8924 + }, + { + "epoch": 0.9801229958269273, + "grad_norm": 2.292571783065796, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7272164821624756, + "num_tokens": 229054882.0, + "step": 8925 + }, + { + "epoch": 0.980232813529541, + "grad_norm": 1.9631521701812744, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7122727036476135, + "num_tokens": 229081334.0, + "step": 8926 + }, + { + "epoch": 0.9803426312321546, + "grad_norm": 2.7852542400360107, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7120994329452515, + "num_tokens": 229106192.0, + "step": 8927 + }, + { + "epoch": 0.9804524489347682, + "grad_norm": 1.9750392436981201, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7211425304412842, + "num_tokens": 229135610.0, + "step": 8928 + }, + { + "epoch": 0.980562266637382, + "grad_norm": 2.093334436416626, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.690390944480896, + "num_tokens": 229162755.0, + "step": 8929 + }, + { + "epoch": 0.9806720843399956, + "grad_norm": 2.3107893466949463, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6993042826652527, + "num_tokens": 229185013.0, + "step": 8930 + }, + { + "epoch": 0.9807819020426093, + "grad_norm": 2.365696668624878, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6851320266723633, + "num_tokens": 229208436.0, + "step": 8931 + }, + { + "epoch": 0.9808917197452229, + "grad_norm": 2.1138534545898438, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6963479518890381, + "num_tokens": 229233825.0, + "step": 8932 + }, + { + "epoch": 0.9810015374478366, + "grad_norm": 2.2455382347106934, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6992762684822083, + "num_tokens": 229258652.0, + "step": 8933 + }, + { + "epoch": 0.9811113551504502, + "grad_norm": 2.216735363006592, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6956905722618103, + "num_tokens": 229285003.0, + "step": 8934 + }, + { + "epoch": 0.9812211728530639, + "grad_norm": 2.2019054889678955, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7085607647895813, + "num_tokens": 229310508.0, + "step": 8935 + }, + { + "epoch": 0.9813309905556776, + "grad_norm": 2.2671167850494385, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6947689056396484, + "num_tokens": 229335159.0, + "step": 8936 + }, + { + "epoch": 0.9814408082582913, + "grad_norm": 2.4645538330078125, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7157058119773865, + "num_tokens": 229355505.0, + "step": 8937 + }, + { + "epoch": 0.9815506259609049, + "grad_norm": 2.504544973373413, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7294492125511169, + "num_tokens": 229375591.0, + "step": 8938 + }, + { + "epoch": 0.9816604436635186, + "grad_norm": 2.229586362838745, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6936007142066956, + "num_tokens": 229400557.0, + "step": 8939 + }, + { + "epoch": 0.9817702613661322, + "grad_norm": 2.5676043033599854, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7264058589935303, + "num_tokens": 229419095.0, + "step": 8940 + }, + { + "epoch": 0.9818800790687459, + "grad_norm": 2.130868673324585, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7292525768280029, + "num_tokens": 229443384.0, + "step": 8941 + }, + { + "epoch": 0.9819898967713595, + "grad_norm": 2.2683446407318115, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7326176762580872, + "num_tokens": 229466423.0, + "step": 8942 + }, + { + "epoch": 0.9820997144739732, + "grad_norm": 2.040553331375122, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.689511775970459, + "num_tokens": 229495133.0, + "step": 8943 + }, + { + "epoch": 0.9822095321765869, + "grad_norm": 2.6221446990966797, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.710090160369873, + "num_tokens": 229514982.0, + "step": 8944 + }, + { + "epoch": 0.9823193498792006, + "grad_norm": 2.0470457077026367, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6995468139648438, + "num_tokens": 229544608.0, + "step": 8945 + }, + { + "epoch": 0.9824291675818142, + "grad_norm": 2.352389097213745, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6934307217597961, + "num_tokens": 229568395.0, + "step": 8946 + }, + { + "epoch": 0.9825389852844278, + "grad_norm": 2.453521490097046, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.741197943687439, + "num_tokens": 229586987.0, + "step": 8947 + }, + { + "epoch": 0.9826488029870415, + "grad_norm": 2.0732617378234863, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7249598503112793, + "num_tokens": 229613636.0, + "step": 8948 + }, + { + "epoch": 0.9827586206896551, + "grad_norm": 2.16796612739563, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6864126324653625, + "num_tokens": 229640253.0, + "step": 8949 + }, + { + "epoch": 0.9828684383922688, + "grad_norm": 2.3150131702423096, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6884963512420654, + "num_tokens": 229664410.0, + "step": 8950 + }, + { + "epoch": 0.9829782560948825, + "grad_norm": 1.9241751432418823, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6950833797454834, + "num_tokens": 229695725.0, + "step": 8951 + }, + { + "epoch": 0.9830880737974962, + "grad_norm": 2.2059340476989746, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7203247547149658, + "num_tokens": 229718388.0, + "step": 8952 + }, + { + "epoch": 0.9831978915001098, + "grad_norm": 2.413992404937744, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7111306190490723, + "num_tokens": 229738122.0, + "step": 8953 + }, + { + "epoch": 0.9833077092027235, + "grad_norm": 2.0266361236572266, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7168943881988525, + "num_tokens": 229765400.0, + "step": 8954 + }, + { + "epoch": 0.9834175269053371, + "grad_norm": 2.5918564796447754, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7016187906265259, + "num_tokens": 229785376.0, + "step": 8955 + }, + { + "epoch": 0.9835273446079508, + "grad_norm": 2.021766424179077, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7059668302536011, + "num_tokens": 229812759.0, + "step": 8956 + }, + { + "epoch": 0.9836371623105644, + "grad_norm": 2.1737074851989746, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.694420576095581, + "num_tokens": 229838690.0, + "step": 8957 + }, + { + "epoch": 0.9837469800131782, + "grad_norm": 2.2053956985473633, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.715695858001709, + "num_tokens": 229864338.0, + "step": 8958 + }, + { + "epoch": 0.9838567977157918, + "grad_norm": 2.377333641052246, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6988776922225952, + "num_tokens": 229888992.0, + "step": 8959 + }, + { + "epoch": 0.9839666154184055, + "grad_norm": 2.098245859146118, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7330915927886963, + "num_tokens": 229913627.0, + "step": 8960 + }, + { + "epoch": 0.9840764331210191, + "grad_norm": 2.114109516143799, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7136498093605042, + "num_tokens": 229939840.0, + "step": 8961 + }, + { + "epoch": 0.9841862508236328, + "grad_norm": 2.1276679039001465, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7348890900611877, + "num_tokens": 229964888.0, + "step": 8962 + }, + { + "epoch": 0.9842960685262464, + "grad_norm": 2.4808883666992188, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7087866067886353, + "num_tokens": 229987242.0, + "step": 8963 + }, + { + "epoch": 0.98440588622886, + "grad_norm": 1.89372980594635, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7058959007263184, + "num_tokens": 230020035.0, + "step": 8964 + }, + { + "epoch": 0.9845157039314738, + "grad_norm": 2.4418694972991943, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7051702737808228, + "num_tokens": 230041379.0, + "step": 8965 + }, + { + "epoch": 0.9846255216340875, + "grad_norm": 2.241224765777588, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6964988708496094, + "num_tokens": 230065813.0, + "step": 8966 + }, + { + "epoch": 0.9847353393367011, + "grad_norm": 1.867809534072876, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7089129090309143, + "num_tokens": 230097604.0, + "step": 8967 + }, + { + "epoch": 0.9848451570393147, + "grad_norm": 1.9752395153045654, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7011301517486572, + "num_tokens": 230127345.0, + "step": 8968 + }, + { + "epoch": 0.9849549747419284, + "grad_norm": 1.980652093887329, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7106616497039795, + "num_tokens": 230159655.0, + "step": 8969 + }, + { + "epoch": 0.985064792444542, + "grad_norm": 2.2745466232299805, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7133762240409851, + "num_tokens": 230183427.0, + "step": 8970 + }, + { + "epoch": 0.9851746101471557, + "grad_norm": 2.1199963092803955, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7298718690872192, + "num_tokens": 230210572.0, + "step": 8971 + }, + { + "epoch": 0.9852844278497693, + "grad_norm": 2.317986249923706, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6941168904304504, + "num_tokens": 230233939.0, + "step": 8972 + }, + { + "epoch": 0.9853942455523831, + "grad_norm": 2.05830979347229, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7168411612510681, + "num_tokens": 230263259.0, + "step": 8973 + }, + { + "epoch": 0.9855040632549967, + "grad_norm": 2.0141663551330566, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6815345287322998, + "num_tokens": 230292891.0, + "step": 8974 + }, + { + "epoch": 0.9856138809576104, + "grad_norm": 2.146491289138794, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6953558921813965, + "num_tokens": 230320342.0, + "step": 8975 + }, + { + "epoch": 0.985723698660224, + "grad_norm": 2.1820898056030273, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7290265560150146, + "num_tokens": 230345197.0, + "step": 8976 + }, + { + "epoch": 0.9858335163628377, + "grad_norm": 2.4485361576080322, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7149380445480347, + "num_tokens": 230364566.0, + "step": 8977 + }, + { + "epoch": 0.9859433340654513, + "grad_norm": 2.4338736534118652, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.698298454284668, + "num_tokens": 230385614.0, + "step": 8978 + }, + { + "epoch": 0.986053151768065, + "grad_norm": 1.978731632232666, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6899475455284119, + "num_tokens": 230418129.0, + "step": 8979 + }, + { + "epoch": 0.9861629694706787, + "grad_norm": 2.6067280769348145, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7159966826438904, + "num_tokens": 230436517.0, + "step": 8980 + }, + { + "epoch": 0.9862727871732924, + "grad_norm": 2.2491564750671387, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7201211452484131, + "num_tokens": 230458477.0, + "step": 8981 + }, + { + "epoch": 0.986382604875906, + "grad_norm": 2.5416085720062256, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7169145941734314, + "num_tokens": 230478347.0, + "step": 8982 + }, + { + "epoch": 0.9864924225785197, + "grad_norm": 1.9696800708770752, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7233543395996094, + "num_tokens": 230508630.0, + "step": 8983 + }, + { + "epoch": 0.9866022402811333, + "grad_norm": 2.185342788696289, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7084984183311462, + "num_tokens": 230533478.0, + "step": 8984 + }, + { + "epoch": 0.986712057983747, + "grad_norm": 2.3803887367248535, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7154961824417114, + "num_tokens": 230555739.0, + "step": 8985 + }, + { + "epoch": 0.9868218756863606, + "grad_norm": 2.360734224319458, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7083027362823486, + "num_tokens": 230578246.0, + "step": 8986 + }, + { + "epoch": 0.9869316933889744, + "grad_norm": 2.4264817237854004, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7088079452514648, + "num_tokens": 230600091.0, + "step": 8987 + }, + { + "epoch": 0.987041511091588, + "grad_norm": 2.2267158031463623, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7278651595115662, + "num_tokens": 230625039.0, + "step": 8988 + }, + { + "epoch": 0.9871513287942016, + "grad_norm": 2.3987553119659424, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6866414546966553, + "num_tokens": 230648603.0, + "step": 8989 + }, + { + "epoch": 0.9872611464968153, + "grad_norm": 2.748919725418091, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.714397132396698, + "num_tokens": 230667060.0, + "step": 8990 + }, + { + "epoch": 0.9873709641994289, + "grad_norm": 2.7009479999542236, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7232645750045776, + "num_tokens": 230686324.0, + "step": 8991 + }, + { + "epoch": 0.9874807819020426, + "grad_norm": 2.3410632610321045, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7306243777275085, + "num_tokens": 230707881.0, + "step": 8992 + }, + { + "epoch": 0.9875905996046562, + "grad_norm": 2.113694906234741, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6825850009918213, + "num_tokens": 230736417.0, + "step": 8993 + }, + { + "epoch": 0.98770041730727, + "grad_norm": 1.93276047706604, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7193648815155029, + "num_tokens": 230765924.0, + "step": 8994 + }, + { + "epoch": 0.9878102350098836, + "grad_norm": 2.383888006210327, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7479998469352722, + "num_tokens": 230785777.0, + "step": 8995 + }, + { + "epoch": 0.9879200527124973, + "grad_norm": 2.2469167709350586, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7112116813659668, + "num_tokens": 230811019.0, + "step": 8996 + }, + { + "epoch": 0.9880298704151109, + "grad_norm": 2.2224526405334473, + "learning_rate": 1e-06, + "loss": 1.0631, + "mean_token_accuracy": 0.6809343099594116, + "num_tokens": 230839557.0, + "step": 8997 + }, + { + "epoch": 0.9881396881177246, + "grad_norm": 2.163616418838501, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7066253423690796, + "num_tokens": 230864844.0, + "step": 8998 + }, + { + "epoch": 0.9882495058203382, + "grad_norm": 2.5181655883789062, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7427394986152649, + "num_tokens": 230885171.0, + "step": 8999 + }, + { + "epoch": 0.9883593235229519, + "grad_norm": 2.1670193672180176, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7016069293022156, + "num_tokens": 230912188.0, + "step": 9000 + }, + { + "epoch": 0.9884691412255655, + "grad_norm": 2.0739829540252686, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7121889591217041, + "num_tokens": 230938125.0, + "step": 9001 + }, + { + "epoch": 0.9885789589281793, + "grad_norm": 2.299356698989868, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7031799554824829, + "num_tokens": 230962159.0, + "step": 9002 + }, + { + "epoch": 0.9886887766307929, + "grad_norm": 1.9760915040969849, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7028458714485168, + "num_tokens": 230993101.0, + "step": 9003 + }, + { + "epoch": 0.9887985943334066, + "grad_norm": 2.2884132862091064, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7363059520721436, + "num_tokens": 231014816.0, + "step": 9004 + }, + { + "epoch": 0.9889084120360202, + "grad_norm": 2.0010621547698975, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7386327981948853, + "num_tokens": 231042683.0, + "step": 9005 + }, + { + "epoch": 0.9890182297386338, + "grad_norm": 2.2164347171783447, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6917803287506104, + "num_tokens": 231065952.0, + "step": 9006 + }, + { + "epoch": 0.9891280474412475, + "grad_norm": 2.373541831970215, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6996393203735352, + "num_tokens": 231089727.0, + "step": 9007 + }, + { + "epoch": 0.9892378651438611, + "grad_norm": 2.407180070877075, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6931682825088501, + "num_tokens": 231114010.0, + "step": 9008 + }, + { + "epoch": 0.9893476828464749, + "grad_norm": 2.1574931144714355, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6991292834281921, + "num_tokens": 231139217.0, + "step": 9009 + }, + { + "epoch": 0.9894575005490885, + "grad_norm": 2.577241897583008, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7317138910293579, + "num_tokens": 231157269.0, + "step": 9010 + }, + { + "epoch": 0.9895673182517022, + "grad_norm": 2.1772124767303467, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7014709711074829, + "num_tokens": 231183630.0, + "step": 9011 + }, + { + "epoch": 0.9896771359543158, + "grad_norm": 2.245523452758789, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7145965099334717, + "num_tokens": 231208478.0, + "step": 9012 + }, + { + "epoch": 0.9897869536569295, + "grad_norm": 2.1582884788513184, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7253760099411011, + "num_tokens": 231235004.0, + "step": 9013 + }, + { + "epoch": 0.9898967713595431, + "grad_norm": 2.2267847061157227, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7043271064758301, + "num_tokens": 231258429.0, + "step": 9014 + }, + { + "epoch": 0.9900065890621568, + "grad_norm": 2.144174575805664, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7410112619400024, + "num_tokens": 231283259.0, + "step": 9015 + }, + { + "epoch": 0.9901164067647705, + "grad_norm": 2.235509157180786, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7042560577392578, + "num_tokens": 231307575.0, + "step": 9016 + }, + { + "epoch": 0.9902262244673842, + "grad_norm": 1.9145511388778687, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6732874512672424, + "num_tokens": 231341802.0, + "step": 9017 + }, + { + "epoch": 0.9903360421699978, + "grad_norm": 2.105409860610962, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7206807136535645, + "num_tokens": 231368603.0, + "step": 9018 + }, + { + "epoch": 0.9904458598726115, + "grad_norm": 2.5442845821380615, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.714501142501831, + "num_tokens": 231387547.0, + "step": 9019 + }, + { + "epoch": 0.9905556775752251, + "grad_norm": 2.1945712566375732, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6823514699935913, + "num_tokens": 231414639.0, + "step": 9020 + }, + { + "epoch": 0.9906654952778388, + "grad_norm": 2.2386085987091064, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7127710580825806, + "num_tokens": 231439510.0, + "step": 9021 + }, + { + "epoch": 0.9907753129804524, + "grad_norm": 2.197219133377075, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7240647077560425, + "num_tokens": 231465238.0, + "step": 9022 + }, + { + "epoch": 0.9908851306830662, + "grad_norm": 2.0521292686462402, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7188825011253357, + "num_tokens": 231491641.0, + "step": 9023 + }, + { + "epoch": 0.9909949483856798, + "grad_norm": 2.377669334411621, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7079598307609558, + "num_tokens": 231514737.0, + "step": 9024 + }, + { + "epoch": 0.9911047660882935, + "grad_norm": 1.8689543008804321, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.715043306350708, + "num_tokens": 231547800.0, + "step": 9025 + }, + { + "epoch": 0.9912145837909071, + "grad_norm": 2.3678717613220215, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7119314074516296, + "num_tokens": 231569076.0, + "step": 9026 + }, + { + "epoch": 0.9913244014935207, + "grad_norm": 2.5986502170562744, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7186782360076904, + "num_tokens": 231587457.0, + "step": 9027 + }, + { + "epoch": 0.9914342191961344, + "grad_norm": 2.0580742359161377, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6999425888061523, + "num_tokens": 231617182.0, + "step": 9028 + }, + { + "epoch": 0.991544036898748, + "grad_norm": 2.046219825744629, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7053390741348267, + "num_tokens": 231645446.0, + "step": 9029 + }, + { + "epoch": 0.9916538546013618, + "grad_norm": 1.9985440969467163, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6904855966567993, + "num_tokens": 231677155.0, + "step": 9030 + }, + { + "epoch": 0.9917636723039754, + "grad_norm": 2.5578715801239014, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7194068431854248, + "num_tokens": 231696878.0, + "step": 9031 + }, + { + "epoch": 0.9918734900065891, + "grad_norm": 2.259506940841675, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7322208881378174, + "num_tokens": 231718347.0, + "step": 9032 + }, + { + "epoch": 0.9919833077092027, + "grad_norm": 2.0204763412475586, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.733025074005127, + "num_tokens": 231743924.0, + "step": 9033 + }, + { + "epoch": 0.9920931254118164, + "grad_norm": 2.062971830368042, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7302300333976746, + "num_tokens": 231773201.0, + "step": 9034 + }, + { + "epoch": 0.99220294311443, + "grad_norm": 2.2517287731170654, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7144725322723389, + "num_tokens": 231797249.0, + "step": 9035 + }, + { + "epoch": 0.9923127608170437, + "grad_norm": 2.307770013809204, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7089833617210388, + "num_tokens": 231819876.0, + "step": 9036 + }, + { + "epoch": 0.9924225785196573, + "grad_norm": 2.299546718597412, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7232911586761475, + "num_tokens": 231843600.0, + "step": 9037 + }, + { + "epoch": 0.9925323962222711, + "grad_norm": 2.1859099864959717, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7266528606414795, + "num_tokens": 231867571.0, + "step": 9038 + }, + { + "epoch": 0.9926422139248847, + "grad_norm": 2.2609498500823975, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7254186868667603, + "num_tokens": 231890518.0, + "step": 9039 + }, + { + "epoch": 0.9927520316274984, + "grad_norm": 2.0514471530914307, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7166789770126343, + "num_tokens": 231918052.0, + "step": 9040 + }, + { + "epoch": 0.992861849330112, + "grad_norm": 2.1179378032684326, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.7065428495407104, + "num_tokens": 231945763.0, + "step": 9041 + }, + { + "epoch": 0.9929716670327257, + "grad_norm": 2.328850746154785, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7019389867782593, + "num_tokens": 231968741.0, + "step": 9042 + }, + { + "epoch": 0.9930814847353393, + "grad_norm": 1.7944645881652832, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6922074556350708, + "num_tokens": 232005188.0, + "step": 9043 + }, + { + "epoch": 0.993191302437953, + "grad_norm": 2.281954288482666, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7371416091918945, + "num_tokens": 232028319.0, + "step": 9044 + }, + { + "epoch": 0.9933011201405667, + "grad_norm": 2.02837872505188, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6879141330718994, + "num_tokens": 232055959.0, + "step": 9045 + }, + { + "epoch": 0.9934109378431804, + "grad_norm": 2.0131125450134277, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7268542051315308, + "num_tokens": 232082593.0, + "step": 9046 + }, + { + "epoch": 0.993520755545794, + "grad_norm": 2.6138956546783447, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7354065775871277, + "num_tokens": 232102564.0, + "step": 9047 + }, + { + "epoch": 0.9936305732484076, + "grad_norm": 2.2103137969970703, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7134469151496887, + "num_tokens": 232128401.0, + "step": 9048 + }, + { + "epoch": 0.9937403909510213, + "grad_norm": 2.2793962955474854, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.72261643409729, + "num_tokens": 232152826.0, + "step": 9049 + }, + { + "epoch": 0.9938502086536349, + "grad_norm": 2.0416672229766846, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7303551435470581, + "num_tokens": 232181676.0, + "step": 9050 + }, + { + "epoch": 0.9939600263562486, + "grad_norm": 2.148923635482788, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7046811580657959, + "num_tokens": 232206074.0, + "step": 9051 + }, + { + "epoch": 0.9940698440588623, + "grad_norm": 2.242971897125244, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7284588813781738, + "num_tokens": 232229915.0, + "step": 9052 + }, + { + "epoch": 0.994179661761476, + "grad_norm": 2.199741840362549, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7052782773971558, + "num_tokens": 232258714.0, + "step": 9053 + }, + { + "epoch": 0.9942894794640896, + "grad_norm": 2.314746141433716, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7048117518424988, + "num_tokens": 232281163.0, + "step": 9054 + }, + { + "epoch": 0.9943992971667033, + "grad_norm": 2.367136240005493, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7125958204269409, + "num_tokens": 232303436.0, + "step": 9055 + }, + { + "epoch": 0.9945091148693169, + "grad_norm": 2.2719476222991943, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7077499628067017, + "num_tokens": 232327281.0, + "step": 9056 + }, + { + "epoch": 0.9946189325719306, + "grad_norm": 2.473224639892578, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6972517371177673, + "num_tokens": 232347213.0, + "step": 9057 + }, + { + "epoch": 0.9947287502745442, + "grad_norm": 2.269531011581421, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7021647095680237, + "num_tokens": 232373879.0, + "step": 9058 + }, + { + "epoch": 0.994838567977158, + "grad_norm": 2.2813966274261475, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7081940174102783, + "num_tokens": 232397003.0, + "step": 9059 + }, + { + "epoch": 0.9949483856797716, + "grad_norm": 2.257124423980713, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7097755670547485, + "num_tokens": 232421978.0, + "step": 9060 + }, + { + "epoch": 0.9950582033823853, + "grad_norm": 2.0605392456054688, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7348070740699768, + "num_tokens": 232446804.0, + "step": 9061 + }, + { + "epoch": 0.9951680210849989, + "grad_norm": 1.8505223989486694, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7217825651168823, + "num_tokens": 232478842.0, + "step": 9062 + }, + { + "epoch": 0.9952778387876126, + "grad_norm": 2.5533244609832764, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7026337385177612, + "num_tokens": 232498319.0, + "step": 9063 + }, + { + "epoch": 0.9953876564902262, + "grad_norm": 2.1003851890563965, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7132778763771057, + "num_tokens": 232524973.0, + "step": 9064 + }, + { + "epoch": 0.9954974741928398, + "grad_norm": 2.2241299152374268, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7156674861907959, + "num_tokens": 232549035.0, + "step": 9065 + }, + { + "epoch": 0.9956072918954535, + "grad_norm": 2.0642826557159424, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6947290301322937, + "num_tokens": 232576386.0, + "step": 9066 + }, + { + "epoch": 0.9957171095980673, + "grad_norm": 2.4518117904663086, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7325179576873779, + "num_tokens": 232595471.0, + "step": 9067 + }, + { + "epoch": 0.9958269273006809, + "grad_norm": 1.9372950792312622, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6931885480880737, + "num_tokens": 232627004.0, + "step": 9068 + }, + { + "epoch": 0.9959367450032945, + "grad_norm": 2.192619562149048, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7265418767929077, + "num_tokens": 232650791.0, + "step": 9069 + }, + { + "epoch": 0.9960465627059082, + "grad_norm": 1.9937459230422974, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7158092856407166, + "num_tokens": 232679119.0, + "step": 9070 + }, + { + "epoch": 0.9961563804085218, + "grad_norm": 2.1230340003967285, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7062057256698608, + "num_tokens": 232705396.0, + "step": 9071 + }, + { + "epoch": 0.9962661981111355, + "grad_norm": 2.255882740020752, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7167590856552124, + "num_tokens": 232728101.0, + "step": 9072 + }, + { + "epoch": 0.9963760158137491, + "grad_norm": 2.3626132011413574, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7208777070045471, + "num_tokens": 232749867.0, + "step": 9073 + }, + { + "epoch": 0.9964858335163629, + "grad_norm": 2.3373539447784424, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6997750401496887, + "num_tokens": 232774114.0, + "step": 9074 + }, + { + "epoch": 0.9965956512189765, + "grad_norm": 2.2119691371917725, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7150013446807861, + "num_tokens": 232801126.0, + "step": 9075 + }, + { + "epoch": 0.9967054689215902, + "grad_norm": 2.660794973373413, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7134355306625366, + "num_tokens": 232820227.0, + "step": 9076 + }, + { + "epoch": 0.9968152866242038, + "grad_norm": 2.369262933731079, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7164890766143799, + "num_tokens": 232841796.0, + "step": 9077 + }, + { + "epoch": 0.9969251043268175, + "grad_norm": 2.184556484222412, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7111164927482605, + "num_tokens": 232866078.0, + "step": 9078 + }, + { + "epoch": 0.9970349220294311, + "grad_norm": 2.154371976852417, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6992504596710205, + "num_tokens": 232893801.0, + "step": 9079 + }, + { + "epoch": 0.9971447397320448, + "grad_norm": 2.317467212677002, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7013919353485107, + "num_tokens": 232917640.0, + "step": 9080 + }, + { + "epoch": 0.9972545574346585, + "grad_norm": 2.2400877475738525, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7190318703651428, + "num_tokens": 232944416.0, + "step": 9081 + }, + { + "epoch": 0.9973643751372722, + "grad_norm": 2.113787889480591, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7237558960914612, + "num_tokens": 232970732.0, + "step": 9082 + }, + { + "epoch": 0.9974741928398858, + "grad_norm": 2.0810089111328125, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7120181322097778, + "num_tokens": 232997047.0, + "step": 9083 + }, + { + "epoch": 0.9975840105424995, + "grad_norm": 2.210843324661255, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7281874418258667, + "num_tokens": 233020633.0, + "step": 9084 + }, + { + "epoch": 0.9976938282451131, + "grad_norm": 2.0510969161987305, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7008168697357178, + "num_tokens": 233048028.0, + "step": 9085 + }, + { + "epoch": 0.9978036459477267, + "grad_norm": 2.1024134159088135, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6993892788887024, + "num_tokens": 233074051.0, + "step": 9086 + }, + { + "epoch": 0.9979134636503404, + "grad_norm": 2.374952554702759, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7502920031547546, + "num_tokens": 233093375.0, + "step": 9087 + }, + { + "epoch": 0.9980232813529542, + "grad_norm": 2.5272226333618164, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7277975082397461, + "num_tokens": 233112982.0, + "step": 9088 + }, + { + "epoch": 0.9981330990555678, + "grad_norm": 2.195302724838257, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7048359513282776, + "num_tokens": 233136813.0, + "step": 9089 + }, + { + "epoch": 0.9982429167581814, + "grad_norm": 2.274324655532837, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.722936749458313, + "num_tokens": 233158638.0, + "step": 9090 + }, + { + "epoch": 0.9983527344607951, + "grad_norm": 2.0938172340393066, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7058071494102478, + "num_tokens": 233185540.0, + "step": 9091 + }, + { + "epoch": 0.9984625521634087, + "grad_norm": 2.464400053024292, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7087175250053406, + "num_tokens": 233206436.0, + "step": 9092 + }, + { + "epoch": 0.9985723698660224, + "grad_norm": 2.430659532546997, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7331044673919678, + "num_tokens": 233228370.0, + "step": 9093 + }, + { + "epoch": 0.998682187568636, + "grad_norm": 2.1177990436553955, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.6839672327041626, + "num_tokens": 233256732.0, + "step": 9094 + }, + { + "epoch": 0.9987920052712497, + "grad_norm": 2.3694019317626953, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7061677575111389, + "num_tokens": 233279001.0, + "step": 9095 + }, + { + "epoch": 0.9989018229738634, + "grad_norm": 2.17616868019104, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.707445502281189, + "num_tokens": 233301389.0, + "step": 9096 + }, + { + "epoch": 0.9990116406764771, + "grad_norm": 1.7721152305603027, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.6983023881912231, + "num_tokens": 233336607.0, + "step": 9097 + }, + { + "epoch": 0.9991214583790907, + "grad_norm": 2.18489933013916, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6962319016456604, + "num_tokens": 233362926.0, + "step": 9098 + }, + { + "epoch": 0.9992312760817044, + "grad_norm": 2.1313693523406982, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6895408630371094, + "num_tokens": 233389366.0, + "step": 9099 + }, + { + "epoch": 0.999341093784318, + "grad_norm": 2.180025577545166, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7195461392402649, + "num_tokens": 233416385.0, + "step": 9100 + }, + { + "epoch": 0.9994509114869317, + "grad_norm": 2.061859130859375, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.683163046836853, + "num_tokens": 233445959.0, + "step": 9101 + }, + { + "epoch": 0.9995607291895453, + "grad_norm": 2.377943515777588, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7023448944091797, + "num_tokens": 233467684.0, + "step": 9102 + }, + { + "epoch": 0.9996705468921591, + "grad_norm": 2.128931760787964, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7116519212722778, + "num_tokens": 233494418.0, + "step": 9103 + }, + { + "epoch": 0.9997803645947727, + "grad_norm": 1.9429908990859985, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7078179717063904, + "num_tokens": 233522705.0, + "step": 9104 + }, + { + "epoch": 0.9998901822973864, + "grad_norm": 2.227027654647827, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.728553295135498, + "num_tokens": 233545541.0, + "step": 9105 + }, + { + "epoch": 1.0, + "grad_norm": 2.103618621826172, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7074710726737976, + "num_tokens": 233571819.0, + "step": 9106 + }, + { + "epoch": 1.0001098177026138, + "grad_norm": 2.182295799255371, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7059391736984253, + "num_tokens": 233595612.0, + "step": 9107 + }, + { + "epoch": 1.0002196354052273, + "grad_norm": 2.256544828414917, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7017661333084106, + "num_tokens": 233618469.0, + "step": 9108 + }, + { + "epoch": 1.000329453107841, + "grad_norm": 2.152226686477661, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7277024984359741, + "num_tokens": 233644431.0, + "step": 9109 + }, + { + "epoch": 1.0004392708104546, + "grad_norm": 2.1912763118743896, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7235567569732666, + "num_tokens": 233668024.0, + "step": 9110 + }, + { + "epoch": 1.0005490885130683, + "grad_norm": 2.3591530323028564, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7182525992393494, + "num_tokens": 233688939.0, + "step": 9111 + }, + { + "epoch": 1.0006589062156819, + "grad_norm": 2.1258788108825684, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7132267355918884, + "num_tokens": 233714588.0, + "step": 9112 + }, + { + "epoch": 1.0007687239182956, + "grad_norm": 2.158660650253296, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7207622528076172, + "num_tokens": 233740748.0, + "step": 9113 + }, + { + "epoch": 1.0008785416209094, + "grad_norm": 2.01318097114563, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7148303985595703, + "num_tokens": 233769191.0, + "step": 9114 + }, + { + "epoch": 1.000988359323523, + "grad_norm": 2.106173515319824, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7248620986938477, + "num_tokens": 233795483.0, + "step": 9115 + }, + { + "epoch": 1.0010981770261367, + "grad_norm": 1.71840238571167, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7083317637443542, + "num_tokens": 233836304.0, + "step": 9116 + }, + { + "epoch": 1.0012079947287502, + "grad_norm": 2.1701505184173584, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6997659206390381, + "num_tokens": 233863402.0, + "step": 9117 + }, + { + "epoch": 1.001317812431364, + "grad_norm": 2.137864828109741, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.732035219669342, + "num_tokens": 233888867.0, + "step": 9118 + }, + { + "epoch": 1.0014276301339775, + "grad_norm": 2.045668601989746, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7216175198554993, + "num_tokens": 233919015.0, + "step": 9119 + }, + { + "epoch": 1.0015374478365913, + "grad_norm": 2.2732982635498047, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7291761636734009, + "num_tokens": 233942853.0, + "step": 9120 + }, + { + "epoch": 1.001647265539205, + "grad_norm": 2.036137342453003, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7347480058670044, + "num_tokens": 233971662.0, + "step": 9121 + }, + { + "epoch": 1.0017570832418186, + "grad_norm": 2.135066509246826, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7234048843383789, + "num_tokens": 233997835.0, + "step": 9122 + }, + { + "epoch": 1.0018669009444323, + "grad_norm": 2.414609432220459, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7021497488021851, + "num_tokens": 234021510.0, + "step": 9123 + }, + { + "epoch": 1.0019767186470458, + "grad_norm": 2.10105037689209, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7190418243408203, + "num_tokens": 234047895.0, + "step": 9124 + }, + { + "epoch": 1.0020865363496596, + "grad_norm": 1.9551643133163452, + "learning_rate": 1e-06, + "loss": 1.0833, + "mean_token_accuracy": 0.6711664199829102, + "num_tokens": 234079401.0, + "step": 9125 + }, + { + "epoch": 1.0021963540522731, + "grad_norm": 2.4941229820251465, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.729825496673584, + "num_tokens": 234100145.0, + "step": 9126 + }, + { + "epoch": 1.002306171754887, + "grad_norm": 2.2737221717834473, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7179819941520691, + "num_tokens": 234123932.0, + "step": 9127 + }, + { + "epoch": 1.0024159894575007, + "grad_norm": 2.306626558303833, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6924893856048584, + "num_tokens": 234147775.0, + "step": 9128 + }, + { + "epoch": 1.0025258071601142, + "grad_norm": 2.4758810997009277, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7299821972846985, + "num_tokens": 234168982.0, + "step": 9129 + }, + { + "epoch": 1.002635624862728, + "grad_norm": 2.209418773651123, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7350193858146667, + "num_tokens": 234192180.0, + "step": 9130 + }, + { + "epoch": 1.0027454425653415, + "grad_norm": 2.0982882976531982, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7134314179420471, + "num_tokens": 234218831.0, + "step": 9131 + }, + { + "epoch": 1.0028552602679552, + "grad_norm": 2.09273099899292, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.6779003143310547, + "num_tokens": 234249293.0, + "step": 9132 + }, + { + "epoch": 1.0029650779705688, + "grad_norm": 2.194918632507324, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.732270359992981, + "num_tokens": 234274252.0, + "step": 9133 + }, + { + "epoch": 1.0030748956731825, + "grad_norm": 2.1178975105285645, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7288441061973572, + "num_tokens": 234299103.0, + "step": 9134 + }, + { + "epoch": 1.0031847133757963, + "grad_norm": 2.2738540172576904, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7170950174331665, + "num_tokens": 234325403.0, + "step": 9135 + }, + { + "epoch": 1.0032945310784098, + "grad_norm": 1.9447773694992065, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7089102268218994, + "num_tokens": 234356831.0, + "step": 9136 + }, + { + "epoch": 1.0034043487810236, + "grad_norm": 2.26165509223938, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7377654910087585, + "num_tokens": 234380018.0, + "step": 9137 + }, + { + "epoch": 1.0035141664836371, + "grad_norm": 2.233499526977539, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7227833271026611, + "num_tokens": 234404438.0, + "step": 9138 + }, + { + "epoch": 1.0036239841862509, + "grad_norm": 2.1973321437835693, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7111096978187561, + "num_tokens": 234430519.0, + "step": 9139 + }, + { + "epoch": 1.0037338018888644, + "grad_norm": 2.4352731704711914, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7198554873466492, + "num_tokens": 234453117.0, + "step": 9140 + }, + { + "epoch": 1.0038436195914782, + "grad_norm": 2.2203667163848877, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7082603573799133, + "num_tokens": 234478340.0, + "step": 9141 + }, + { + "epoch": 1.0039534372940917, + "grad_norm": 2.0372138023376465, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6906266808509827, + "num_tokens": 234510827.0, + "step": 9142 + }, + { + "epoch": 1.0040632549967055, + "grad_norm": 1.9921329021453857, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7097797393798828, + "num_tokens": 234540810.0, + "step": 9143 + }, + { + "epoch": 1.0041730726993192, + "grad_norm": 2.8155276775360107, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7303994297981262, + "num_tokens": 234559921.0, + "step": 9144 + }, + { + "epoch": 1.0042828904019327, + "grad_norm": 1.9730181694030762, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7170660495758057, + "num_tokens": 234591527.0, + "step": 9145 + }, + { + "epoch": 1.0043927081045465, + "grad_norm": 2.1674928665161133, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7133442163467407, + "num_tokens": 234618783.0, + "step": 9146 + }, + { + "epoch": 1.00450252580716, + "grad_norm": 1.868943214416504, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6897870898246765, + "num_tokens": 234653821.0, + "step": 9147 + }, + { + "epoch": 1.0046123435097738, + "grad_norm": 2.2715766429901123, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7277952432632446, + "num_tokens": 234679337.0, + "step": 9148 + }, + { + "epoch": 1.0047221612123873, + "grad_norm": 2.169734239578247, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7169337272644043, + "num_tokens": 234704005.0, + "step": 9149 + }, + { + "epoch": 1.004831978915001, + "grad_norm": 2.6547632217407227, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.722861647605896, + "num_tokens": 234722507.0, + "step": 9150 + }, + { + "epoch": 1.0049417966176148, + "grad_norm": 2.254849672317505, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7069843411445618, + "num_tokens": 234748944.0, + "step": 9151 + }, + { + "epoch": 1.0050516143202284, + "grad_norm": 2.068342447280884, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7001787424087524, + "num_tokens": 234779719.0, + "step": 9152 + }, + { + "epoch": 1.0051614320228421, + "grad_norm": 2.494454860687256, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.729544460773468, + "num_tokens": 234801075.0, + "step": 9153 + }, + { + "epoch": 1.0052712497254557, + "grad_norm": 2.0441153049468994, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7093332409858704, + "num_tokens": 234828137.0, + "step": 9154 + }, + { + "epoch": 1.0053810674280694, + "grad_norm": 2.246133327484131, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.722687840461731, + "num_tokens": 234854923.0, + "step": 9155 + }, + { + "epoch": 1.005490885130683, + "grad_norm": 2.022045373916626, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.725540041923523, + "num_tokens": 234884601.0, + "step": 9156 + }, + { + "epoch": 1.0056007028332967, + "grad_norm": 2.1184210777282715, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7237470746040344, + "num_tokens": 234913055.0, + "step": 9157 + }, + { + "epoch": 1.0057105205359105, + "grad_norm": 2.4966254234313965, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.702060878276825, + "num_tokens": 234934097.0, + "step": 9158 + }, + { + "epoch": 1.005820338238524, + "grad_norm": 2.04781436920166, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6991560459136963, + "num_tokens": 234965181.0, + "step": 9159 + }, + { + "epoch": 1.0059301559411378, + "grad_norm": 2.3784260749816895, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.728309154510498, + "num_tokens": 234986613.0, + "step": 9160 + }, + { + "epoch": 1.0060399736437513, + "grad_norm": 2.107992649078369, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7215354442596436, + "num_tokens": 235015301.0, + "step": 9161 + }, + { + "epoch": 1.006149791346365, + "grad_norm": 2.6734466552734375, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7153013944625854, + "num_tokens": 235033481.0, + "step": 9162 + }, + { + "epoch": 1.0062596090489786, + "grad_norm": 2.7045633792877197, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7153812646865845, + "num_tokens": 235052984.0, + "step": 9163 + }, + { + "epoch": 1.0063694267515924, + "grad_norm": 2.3197569847106934, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7103564739227295, + "num_tokens": 235075538.0, + "step": 9164 + }, + { + "epoch": 1.0064792444542061, + "grad_norm": 2.383936643600464, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.709324300289154, + "num_tokens": 235099743.0, + "step": 9165 + }, + { + "epoch": 1.0065890621568196, + "grad_norm": 2.2329301834106445, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7183295488357544, + "num_tokens": 235125694.0, + "step": 9166 + }, + { + "epoch": 1.0066988798594334, + "grad_norm": 2.4777419567108154, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7293837070465088, + "num_tokens": 235146102.0, + "step": 9167 + }, + { + "epoch": 1.006808697562047, + "grad_norm": 2.3298168182373047, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7314426898956299, + "num_tokens": 235168501.0, + "step": 9168 + }, + { + "epoch": 1.0069185152646607, + "grad_norm": 2.063459873199463, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7177916765213013, + "num_tokens": 235196504.0, + "step": 9169 + }, + { + "epoch": 1.0070283329672742, + "grad_norm": 2.354652166366577, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7323119640350342, + "num_tokens": 235217116.0, + "step": 9170 + }, + { + "epoch": 1.007138150669888, + "grad_norm": 2.2362852096557617, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7240786552429199, + "num_tokens": 235242735.0, + "step": 9171 + }, + { + "epoch": 1.0072479683725017, + "grad_norm": 2.301774263381958, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7133604288101196, + "num_tokens": 235265784.0, + "step": 9172 + }, + { + "epoch": 1.0073577860751153, + "grad_norm": 1.9576952457427979, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7060149908065796, + "num_tokens": 235299353.0, + "step": 9173 + }, + { + "epoch": 1.007467603777729, + "grad_norm": 2.0239787101745605, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7122557163238525, + "num_tokens": 235328990.0, + "step": 9174 + }, + { + "epoch": 1.0075774214803426, + "grad_norm": 2.332801580429077, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7161869406700134, + "num_tokens": 235351843.0, + "step": 9175 + }, + { + "epoch": 1.0076872391829563, + "grad_norm": 2.2136125564575195, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7224074602127075, + "num_tokens": 235374953.0, + "step": 9176 + }, + { + "epoch": 1.0077970568855699, + "grad_norm": 2.4798474311828613, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7412654757499695, + "num_tokens": 235396470.0, + "step": 9177 + }, + { + "epoch": 1.0079068745881836, + "grad_norm": 2.209995985031128, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7255011200904846, + "num_tokens": 235420027.0, + "step": 9178 + }, + { + "epoch": 1.0080166922907974, + "grad_norm": 1.9327583312988281, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7038658261299133, + "num_tokens": 235451328.0, + "step": 9179 + }, + { + "epoch": 1.008126509993411, + "grad_norm": 2.158931255340576, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7339460849761963, + "num_tokens": 235478400.0, + "step": 9180 + }, + { + "epoch": 1.0082363276960247, + "grad_norm": 2.175327777862549, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7374324202537537, + "num_tokens": 235503708.0, + "step": 9181 + }, + { + "epoch": 1.0083461453986382, + "grad_norm": 2.1082990169525146, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7358983755111694, + "num_tokens": 235528496.0, + "step": 9182 + }, + { + "epoch": 1.008455963101252, + "grad_norm": 2.3631808757781982, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7330071926116943, + "num_tokens": 235551101.0, + "step": 9183 + }, + { + "epoch": 1.0085657808038655, + "grad_norm": 2.4866769313812256, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7117758989334106, + "num_tokens": 235572130.0, + "step": 9184 + }, + { + "epoch": 1.0086755985064793, + "grad_norm": 2.257484197616577, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7064735293388367, + "num_tokens": 235598415.0, + "step": 9185 + }, + { + "epoch": 1.008785416209093, + "grad_norm": 2.0931503772735596, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7125543355941772, + "num_tokens": 235628084.0, + "step": 9186 + }, + { + "epoch": 1.0088952339117065, + "grad_norm": 2.256870746612549, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6984647512435913, + "num_tokens": 235655612.0, + "step": 9187 + }, + { + "epoch": 1.0090050516143203, + "grad_norm": 2.268167734146118, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6941487789154053, + "num_tokens": 235682510.0, + "step": 9188 + }, + { + "epoch": 1.0091148693169338, + "grad_norm": 2.1849451065063477, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7216118574142456, + "num_tokens": 235707436.0, + "step": 9189 + }, + { + "epoch": 1.0092246870195476, + "grad_norm": 2.644507884979248, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7441437840461731, + "num_tokens": 235725020.0, + "step": 9190 + }, + { + "epoch": 1.0093345047221611, + "grad_norm": 2.196345567703247, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7223312854766846, + "num_tokens": 235751235.0, + "step": 9191 + }, + { + "epoch": 1.0094443224247749, + "grad_norm": 2.2700438499450684, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7110410332679749, + "num_tokens": 235775798.0, + "step": 9192 + }, + { + "epoch": 1.0095541401273886, + "grad_norm": 2.5602428913116455, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7413734197616577, + "num_tokens": 235793665.0, + "step": 9193 + }, + { + "epoch": 1.0096639578300022, + "grad_norm": 2.4201364517211914, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7166335582733154, + "num_tokens": 235815127.0, + "step": 9194 + }, + { + "epoch": 1.009773775532616, + "grad_norm": 1.9299248456954956, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7142993211746216, + "num_tokens": 235847672.0, + "step": 9195 + }, + { + "epoch": 1.0098835932352295, + "grad_norm": 2.493365526199341, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7296127080917358, + "num_tokens": 235869177.0, + "step": 9196 + }, + { + "epoch": 1.0099934109378432, + "grad_norm": 2.338064193725586, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7124331593513489, + "num_tokens": 235891787.0, + "step": 9197 + }, + { + "epoch": 1.0101032286404568, + "grad_norm": 2.2138001918792725, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7196444272994995, + "num_tokens": 235918830.0, + "step": 9198 + }, + { + "epoch": 1.0102130463430705, + "grad_norm": 2.047846555709839, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6975924968719482, + "num_tokens": 235948065.0, + "step": 9199 + }, + { + "epoch": 1.010322864045684, + "grad_norm": 2.1970274448394775, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7147207260131836, + "num_tokens": 235974644.0, + "step": 9200 + }, + { + "epoch": 1.0104326817482978, + "grad_norm": 2.1062474250793457, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6899863481521606, + "num_tokens": 236002029.0, + "step": 9201 + }, + { + "epoch": 1.0105424994509116, + "grad_norm": 2.138979911804199, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7224857807159424, + "num_tokens": 236028138.0, + "step": 9202 + }, + { + "epoch": 1.010652317153525, + "grad_norm": 1.9521117210388184, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7163974046707153, + "num_tokens": 236061013.0, + "step": 9203 + }, + { + "epoch": 1.0107621348561389, + "grad_norm": 2.202303647994995, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.711100697517395, + "num_tokens": 236087969.0, + "step": 9204 + }, + { + "epoch": 1.0108719525587524, + "grad_norm": 2.0620081424713135, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.742128849029541, + "num_tokens": 236114955.0, + "step": 9205 + }, + { + "epoch": 1.0109817702613662, + "grad_norm": 2.1891286373138428, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7219822406768799, + "num_tokens": 236141782.0, + "step": 9206 + }, + { + "epoch": 1.0110915879639797, + "grad_norm": 2.166057825088501, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7120466232299805, + "num_tokens": 236170036.0, + "step": 9207 + }, + { + "epoch": 1.0112014056665934, + "grad_norm": 2.3369572162628174, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6972286701202393, + "num_tokens": 236193960.0, + "step": 9208 + }, + { + "epoch": 1.0113112233692072, + "grad_norm": 1.915899634361267, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7215820550918579, + "num_tokens": 236227620.0, + "step": 9209 + }, + { + "epoch": 1.0114210410718207, + "grad_norm": 2.0348472595214844, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6906686425209045, + "num_tokens": 236263856.0, + "step": 9210 + }, + { + "epoch": 1.0115308587744345, + "grad_norm": 2.2855494022369385, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7313849925994873, + "num_tokens": 236289100.0, + "step": 9211 + }, + { + "epoch": 1.011640676477048, + "grad_norm": 2.2514445781707764, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7294164896011353, + "num_tokens": 236312635.0, + "step": 9212 + }, + { + "epoch": 1.0117504941796618, + "grad_norm": 2.4110283851623535, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7346993684768677, + "num_tokens": 236334884.0, + "step": 9213 + }, + { + "epoch": 1.0118603118822753, + "grad_norm": 1.9602478742599487, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7196560502052307, + "num_tokens": 236364345.0, + "step": 9214 + }, + { + "epoch": 1.011970129584889, + "grad_norm": 2.2424235343933105, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7150648832321167, + "num_tokens": 236388976.0, + "step": 9215 + }, + { + "epoch": 1.0120799472875028, + "grad_norm": 2.2567138671875, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7206578254699707, + "num_tokens": 236414298.0, + "step": 9216 + }, + { + "epoch": 1.0121897649901164, + "grad_norm": 2.3436367511749268, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7150994539260864, + "num_tokens": 236437744.0, + "step": 9217 + }, + { + "epoch": 1.0122995826927301, + "grad_norm": 2.5216197967529297, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7269158363342285, + "num_tokens": 236456898.0, + "step": 9218 + }, + { + "epoch": 1.0124094003953437, + "grad_norm": 2.087751626968384, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.717816948890686, + "num_tokens": 236487145.0, + "step": 9219 + }, + { + "epoch": 1.0125192180979574, + "grad_norm": 2.267929792404175, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7707586884498596, + "num_tokens": 236509530.0, + "step": 9220 + }, + { + "epoch": 1.012629035800571, + "grad_norm": 2.6933484077453613, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7145664095878601, + "num_tokens": 236529234.0, + "step": 9221 + }, + { + "epoch": 1.0127388535031847, + "grad_norm": 2.674020290374756, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7451502084732056, + "num_tokens": 236547854.0, + "step": 9222 + }, + { + "epoch": 1.0128486712057985, + "grad_norm": 2.532313346862793, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7043567299842834, + "num_tokens": 236570364.0, + "step": 9223 + }, + { + "epoch": 1.012958488908412, + "grad_norm": 2.0371766090393066, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7017526626586914, + "num_tokens": 236602479.0, + "step": 9224 + }, + { + "epoch": 1.0130683066110258, + "grad_norm": 2.453153610229492, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7374407649040222, + "num_tokens": 236624553.0, + "step": 9225 + }, + { + "epoch": 1.0131781243136393, + "grad_norm": 2.1589114665985107, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7204360365867615, + "num_tokens": 236653384.0, + "step": 9226 + }, + { + "epoch": 1.013287942016253, + "grad_norm": 2.4837491512298584, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7301561236381531, + "num_tokens": 236674183.0, + "step": 9227 + }, + { + "epoch": 1.0133977597188666, + "grad_norm": 2.556238889694214, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7225225567817688, + "num_tokens": 236695092.0, + "step": 9228 + }, + { + "epoch": 1.0135075774214803, + "grad_norm": 2.2430777549743652, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7435122728347778, + "num_tokens": 236718190.0, + "step": 9229 + }, + { + "epoch": 1.013617395124094, + "grad_norm": 2.102597951889038, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7215385437011719, + "num_tokens": 236746051.0, + "step": 9230 + }, + { + "epoch": 1.0137272128267076, + "grad_norm": 2.2178330421447754, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7238890528678894, + "num_tokens": 236771309.0, + "step": 9231 + }, + { + "epoch": 1.0138370305293214, + "grad_norm": 2.6093733310699463, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7283299565315247, + "num_tokens": 236791525.0, + "step": 9232 + }, + { + "epoch": 1.013946848231935, + "grad_norm": 2.0655181407928467, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7184527516365051, + "num_tokens": 236820620.0, + "step": 9233 + }, + { + "epoch": 1.0140566659345487, + "grad_norm": 2.170010805130005, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7031779289245605, + "num_tokens": 236849185.0, + "step": 9234 + }, + { + "epoch": 1.0141664836371622, + "grad_norm": 2.3507630825042725, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7216935157775879, + "num_tokens": 236873082.0, + "step": 9235 + }, + { + "epoch": 1.014276301339776, + "grad_norm": 2.3963117599487305, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7048570513725281, + "num_tokens": 236894234.0, + "step": 9236 + }, + { + "epoch": 1.0143861190423897, + "grad_norm": 2.251908302307129, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7010684609413147, + "num_tokens": 236923315.0, + "step": 9237 + }, + { + "epoch": 1.0144959367450033, + "grad_norm": 2.54536771774292, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.715887725353241, + "num_tokens": 236944139.0, + "step": 9238 + }, + { + "epoch": 1.014605754447617, + "grad_norm": 2.2656643390655518, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7341344952583313, + "num_tokens": 236969184.0, + "step": 9239 + }, + { + "epoch": 1.0147155721502306, + "grad_norm": 2.046875238418579, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7196804285049438, + "num_tokens": 236999101.0, + "step": 9240 + }, + { + "epoch": 1.0148253898528443, + "grad_norm": 2.1685128211975098, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7152448892593384, + "num_tokens": 237025287.0, + "step": 9241 + }, + { + "epoch": 1.0149352075554579, + "grad_norm": 2.3880724906921387, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7311338186264038, + "num_tokens": 237047809.0, + "step": 9242 + }, + { + "epoch": 1.0150450252580716, + "grad_norm": 2.265939474105835, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7162885069847107, + "num_tokens": 237071847.0, + "step": 9243 + }, + { + "epoch": 1.0151548429606854, + "grad_norm": 2.4171717166900635, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7346251010894775, + "num_tokens": 237093303.0, + "step": 9244 + }, + { + "epoch": 1.015264660663299, + "grad_norm": 1.9496355056762695, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6965177059173584, + "num_tokens": 237123482.0, + "step": 9245 + }, + { + "epoch": 1.0153744783659127, + "grad_norm": 2.037419557571411, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7182790637016296, + "num_tokens": 237154870.0, + "step": 9246 + }, + { + "epoch": 1.0154842960685262, + "grad_norm": 2.247795820236206, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7235256433486938, + "num_tokens": 237181179.0, + "step": 9247 + }, + { + "epoch": 1.01559411377114, + "grad_norm": 2.016968011856079, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7134762406349182, + "num_tokens": 237209561.0, + "step": 9248 + }, + { + "epoch": 1.0157039314737535, + "grad_norm": 2.1358628273010254, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7374669313430786, + "num_tokens": 237235296.0, + "step": 9249 + }, + { + "epoch": 1.0158137491763672, + "grad_norm": 2.2821969985961914, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7364524006843567, + "num_tokens": 237257153.0, + "step": 9250 + }, + { + "epoch": 1.015923566878981, + "grad_norm": 2.5183231830596924, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7095195055007935, + "num_tokens": 237278652.0, + "step": 9251 + }, + { + "epoch": 1.0160333845815945, + "grad_norm": 2.3266124725341797, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.70980304479599, + "num_tokens": 237302483.0, + "step": 9252 + }, + { + "epoch": 1.0161432022842083, + "grad_norm": 2.2780981063842773, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7468705177307129, + "num_tokens": 237325216.0, + "step": 9253 + }, + { + "epoch": 1.0162530199868218, + "grad_norm": 2.2118589878082275, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7339595556259155, + "num_tokens": 237350199.0, + "step": 9254 + }, + { + "epoch": 1.0163628376894356, + "grad_norm": 2.417802095413208, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7121559381484985, + "num_tokens": 237371719.0, + "step": 9255 + }, + { + "epoch": 1.0164726553920491, + "grad_norm": 2.276202917098999, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7126917243003845, + "num_tokens": 237395816.0, + "step": 9256 + }, + { + "epoch": 1.0165824730946629, + "grad_norm": 2.421743631362915, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7371079325675964, + "num_tokens": 237416049.0, + "step": 9257 + }, + { + "epoch": 1.0166922907972764, + "grad_norm": 2.2993829250335693, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7036367058753967, + "num_tokens": 237439065.0, + "step": 9258 + }, + { + "epoch": 1.0168021084998902, + "grad_norm": 2.4156839847564697, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7302286624908447, + "num_tokens": 237462024.0, + "step": 9259 + }, + { + "epoch": 1.016911926202504, + "grad_norm": 1.9306055307388306, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6944164633750916, + "num_tokens": 237492017.0, + "step": 9260 + }, + { + "epoch": 1.0170217439051175, + "grad_norm": 2.352160930633545, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7080215215682983, + "num_tokens": 237516137.0, + "step": 9261 + }, + { + "epoch": 1.0171315616077312, + "grad_norm": 2.1681509017944336, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7038933038711548, + "num_tokens": 237544146.0, + "step": 9262 + }, + { + "epoch": 1.0172413793103448, + "grad_norm": 2.3967788219451904, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7342230081558228, + "num_tokens": 237564598.0, + "step": 9263 + }, + { + "epoch": 1.0173511970129585, + "grad_norm": 2.4208738803863525, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7502135038375854, + "num_tokens": 237585086.0, + "step": 9264 + }, + { + "epoch": 1.017461014715572, + "grad_norm": 2.3789634704589844, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7155113220214844, + "num_tokens": 237607419.0, + "step": 9265 + }, + { + "epoch": 1.0175708324181858, + "grad_norm": 2.2560391426086426, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7169176340103149, + "num_tokens": 237633068.0, + "step": 9266 + }, + { + "epoch": 1.0176806501207996, + "grad_norm": 2.1433238983154297, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7296686172485352, + "num_tokens": 237661258.0, + "step": 9267 + }, + { + "epoch": 1.017790467823413, + "grad_norm": 2.3386600017547607, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6974825859069824, + "num_tokens": 237686593.0, + "step": 9268 + }, + { + "epoch": 1.0179002855260268, + "grad_norm": 2.611050844192505, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7218437790870667, + "num_tokens": 237705197.0, + "step": 9269 + }, + { + "epoch": 1.0180101032286404, + "grad_norm": 2.233553886413574, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7115598917007446, + "num_tokens": 237730453.0, + "step": 9270 + }, + { + "epoch": 1.0181199209312541, + "grad_norm": 2.1628384590148926, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7122894525527954, + "num_tokens": 237757294.0, + "step": 9271 + }, + { + "epoch": 1.0182297386338677, + "grad_norm": 1.933377742767334, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7176287770271301, + "num_tokens": 237788831.0, + "step": 9272 + }, + { + "epoch": 1.0183395563364814, + "grad_norm": 1.9704725742340088, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7447760105133057, + "num_tokens": 237816046.0, + "step": 9273 + }, + { + "epoch": 1.0184493740390952, + "grad_norm": 2.475660800933838, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7238174080848694, + "num_tokens": 237836913.0, + "step": 9274 + }, + { + "epoch": 1.0185591917417087, + "grad_norm": 2.145569324493408, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6925895810127258, + "num_tokens": 237865448.0, + "step": 9275 + }, + { + "epoch": 1.0186690094443225, + "grad_norm": 2.3490712642669678, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7098947167396545, + "num_tokens": 237890158.0, + "step": 9276 + }, + { + "epoch": 1.018778827146936, + "grad_norm": 2.737311601638794, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7304138541221619, + "num_tokens": 237907408.0, + "step": 9277 + }, + { + "epoch": 1.0188886448495498, + "grad_norm": 2.1979124546051025, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7055030465126038, + "num_tokens": 237933205.0, + "step": 9278 + }, + { + "epoch": 1.0189984625521633, + "grad_norm": 2.0641865730285645, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7116110324859619, + "num_tokens": 237961595.0, + "step": 9279 + }, + { + "epoch": 1.019108280254777, + "grad_norm": 2.3035213947296143, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.719233512878418, + "num_tokens": 237986481.0, + "step": 9280 + }, + { + "epoch": 1.0192180979573908, + "grad_norm": 2.5088467597961426, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7331676483154297, + "num_tokens": 238006103.0, + "step": 9281 + }, + { + "epoch": 1.0193279156600044, + "grad_norm": 2.4773213863372803, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7234537601470947, + "num_tokens": 238027556.0, + "step": 9282 + }, + { + "epoch": 1.0194377333626181, + "grad_norm": 2.56870174407959, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7223665714263916, + "num_tokens": 238048260.0, + "step": 9283 + }, + { + "epoch": 1.0195475510652316, + "grad_norm": 2.2258615493774414, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7306805849075317, + "num_tokens": 238073780.0, + "step": 9284 + }, + { + "epoch": 1.0196573687678454, + "grad_norm": 2.322380304336548, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7117785215377808, + "num_tokens": 238098513.0, + "step": 9285 + }, + { + "epoch": 1.019767186470459, + "grad_norm": 2.0238566398620605, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7174744606018066, + "num_tokens": 238127838.0, + "step": 9286 + }, + { + "epoch": 1.0198770041730727, + "grad_norm": 2.2167422771453857, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7200294733047485, + "num_tokens": 238152434.0, + "step": 9287 + }, + { + "epoch": 1.0199868218756865, + "grad_norm": 1.8564599752426147, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7010109424591064, + "num_tokens": 238189416.0, + "step": 9288 + }, + { + "epoch": 1.0200966395783, + "grad_norm": 2.3750507831573486, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7222821116447449, + "num_tokens": 238212115.0, + "step": 9289 + }, + { + "epoch": 1.0202064572809137, + "grad_norm": 2.207886219024658, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7340919971466064, + "num_tokens": 238237502.0, + "step": 9290 + }, + { + "epoch": 1.0203162749835273, + "grad_norm": 2.2882542610168457, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7131016850471497, + "num_tokens": 238261006.0, + "step": 9291 + }, + { + "epoch": 1.020426092686141, + "grad_norm": 2.3237178325653076, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7440847158432007, + "num_tokens": 238282568.0, + "step": 9292 + }, + { + "epoch": 1.0205359103887546, + "grad_norm": 1.9610110521316528, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.744610607624054, + "num_tokens": 238313487.0, + "step": 9293 + }, + { + "epoch": 1.0206457280913683, + "grad_norm": 2.3224270343780518, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7067760825157166, + "num_tokens": 238339966.0, + "step": 9294 + }, + { + "epoch": 1.020755545793982, + "grad_norm": 2.512145757675171, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7183156609535217, + "num_tokens": 238360851.0, + "step": 9295 + }, + { + "epoch": 1.0208653634965956, + "grad_norm": 2.301647663116455, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7484244108200073, + "num_tokens": 238383015.0, + "step": 9296 + }, + { + "epoch": 1.0209751811992094, + "grad_norm": 2.399657964706421, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7113623023033142, + "num_tokens": 238408539.0, + "step": 9297 + }, + { + "epoch": 1.021084998901823, + "grad_norm": 2.1304616928100586, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.6993256211280823, + "num_tokens": 238437452.0, + "step": 9298 + }, + { + "epoch": 1.0211948166044367, + "grad_norm": 2.1186304092407227, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7224768400192261, + "num_tokens": 238464450.0, + "step": 9299 + }, + { + "epoch": 1.0213046343070502, + "grad_norm": 2.4701497554779053, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7328643202781677, + "num_tokens": 238486888.0, + "step": 9300 + }, + { + "epoch": 1.021414452009664, + "grad_norm": 2.587761402130127, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7014829516410828, + "num_tokens": 238507519.0, + "step": 9301 + }, + { + "epoch": 1.0215242697122777, + "grad_norm": 2.094369649887085, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7243692278862, + "num_tokens": 238533886.0, + "step": 9302 + }, + { + "epoch": 1.0216340874148913, + "grad_norm": 2.288977861404419, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7305309772491455, + "num_tokens": 238556977.0, + "step": 9303 + }, + { + "epoch": 1.021743905117505, + "grad_norm": 2.520404100418091, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7111116647720337, + "num_tokens": 238578295.0, + "step": 9304 + }, + { + "epoch": 1.0218537228201185, + "grad_norm": 1.9767591953277588, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7108184695243835, + "num_tokens": 238611520.0, + "step": 9305 + }, + { + "epoch": 1.0219635405227323, + "grad_norm": 2.29990291595459, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7279462814331055, + "num_tokens": 238638499.0, + "step": 9306 + }, + { + "epoch": 1.0220733582253458, + "grad_norm": 2.0825164318084717, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7069051265716553, + "num_tokens": 238667512.0, + "step": 9307 + }, + { + "epoch": 1.0221831759279596, + "grad_norm": 2.1494557857513428, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7193353176116943, + "num_tokens": 238694908.0, + "step": 9308 + }, + { + "epoch": 1.0222929936305734, + "grad_norm": 1.9883257150650024, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7238917350769043, + "num_tokens": 238723098.0, + "step": 9309 + }, + { + "epoch": 1.022402811333187, + "grad_norm": 2.507960557937622, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7064225673675537, + "num_tokens": 238746479.0, + "step": 9310 + }, + { + "epoch": 1.0225126290358006, + "grad_norm": 2.338218927383423, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7176733016967773, + "num_tokens": 238773102.0, + "step": 9311 + }, + { + "epoch": 1.0226224467384142, + "grad_norm": 2.040544271469116, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7103347778320312, + "num_tokens": 238805652.0, + "step": 9312 + }, + { + "epoch": 1.022732264441028, + "grad_norm": 2.111788034439087, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7183338403701782, + "num_tokens": 238831431.0, + "step": 9313 + }, + { + "epoch": 1.0228420821436415, + "grad_norm": 2.1540846824645996, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7302349209785461, + "num_tokens": 238857271.0, + "step": 9314 + }, + { + "epoch": 1.0229518998462552, + "grad_norm": 2.404214859008789, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7233456969261169, + "num_tokens": 238881212.0, + "step": 9315 + }, + { + "epoch": 1.023061717548869, + "grad_norm": 2.3007266521453857, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7426535487174988, + "num_tokens": 238904065.0, + "step": 9316 + }, + { + "epoch": 1.0231715352514825, + "grad_norm": 2.1764533519744873, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7141035795211792, + "num_tokens": 238931734.0, + "step": 9317 + }, + { + "epoch": 1.0232813529540963, + "grad_norm": 2.184582471847534, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7195440530776978, + "num_tokens": 238957576.0, + "step": 9318 + }, + { + "epoch": 1.0233911706567098, + "grad_norm": 2.26204514503479, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7173658609390259, + "num_tokens": 238984439.0, + "step": 9319 + }, + { + "epoch": 1.0235009883593236, + "grad_norm": 2.0546956062316895, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7055909633636475, + "num_tokens": 239018184.0, + "step": 9320 + }, + { + "epoch": 1.023610806061937, + "grad_norm": 2.1642990112304688, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7129076719284058, + "num_tokens": 239048211.0, + "step": 9321 + }, + { + "epoch": 1.0237206237645509, + "grad_norm": 2.6067543029785156, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7316194772720337, + "num_tokens": 239068130.0, + "step": 9322 + }, + { + "epoch": 1.0238304414671644, + "grad_norm": 2.184438943862915, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7150211334228516, + "num_tokens": 239094717.0, + "step": 9323 + }, + { + "epoch": 1.0239402591697782, + "grad_norm": 2.040696620941162, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7009135484695435, + "num_tokens": 239126642.0, + "step": 9324 + }, + { + "epoch": 1.024050076872392, + "grad_norm": 1.9880207777023315, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7177383303642273, + "num_tokens": 239156314.0, + "step": 9325 + }, + { + "epoch": 1.0241598945750054, + "grad_norm": 2.2467353343963623, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7259601354598999, + "num_tokens": 239181237.0, + "step": 9326 + }, + { + "epoch": 1.0242697122776192, + "grad_norm": 2.215696096420288, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7024823427200317, + "num_tokens": 239206024.0, + "step": 9327 + }, + { + "epoch": 1.0243795299802327, + "grad_norm": 2.389803171157837, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7481464147567749, + "num_tokens": 239226432.0, + "step": 9328 + }, + { + "epoch": 1.0244893476828465, + "grad_norm": 2.0545578002929688, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7127351760864258, + "num_tokens": 239254456.0, + "step": 9329 + }, + { + "epoch": 1.02459916538546, + "grad_norm": 2.426805257797241, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7263879179954529, + "num_tokens": 239275276.0, + "step": 9330 + }, + { + "epoch": 1.0247089830880738, + "grad_norm": 2.3371028900146484, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7348452806472778, + "num_tokens": 239297251.0, + "step": 9331 + }, + { + "epoch": 1.0248188007906875, + "grad_norm": 1.9865028858184814, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7109798789024353, + "num_tokens": 239327426.0, + "step": 9332 + }, + { + "epoch": 1.024928618493301, + "grad_norm": 2.228764772415161, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7456488609313965, + "num_tokens": 239350611.0, + "step": 9333 + }, + { + "epoch": 1.0250384361959148, + "grad_norm": 2.390448808670044, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7078997492790222, + "num_tokens": 239373741.0, + "step": 9334 + }, + { + "epoch": 1.0251482538985284, + "grad_norm": 1.96427321434021, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7348055243492126, + "num_tokens": 239404733.0, + "step": 9335 + }, + { + "epoch": 1.0252580716011421, + "grad_norm": 2.4538233280181885, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7207385897636414, + "num_tokens": 239425162.0, + "step": 9336 + }, + { + "epoch": 1.0253678893037557, + "grad_norm": 1.9729030132293701, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.712763249874115, + "num_tokens": 239459381.0, + "step": 9337 + }, + { + "epoch": 1.0254777070063694, + "grad_norm": 2.3206305503845215, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7344039678573608, + "num_tokens": 239482468.0, + "step": 9338 + }, + { + "epoch": 1.0255875247089832, + "grad_norm": 2.3965682983398438, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7375260591506958, + "num_tokens": 239504313.0, + "step": 9339 + }, + { + "epoch": 1.0256973424115967, + "grad_norm": 1.9699535369873047, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6983858346939087, + "num_tokens": 239537573.0, + "step": 9340 + }, + { + "epoch": 1.0258071601142105, + "grad_norm": 2.216691255569458, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7243130207061768, + "num_tokens": 239563580.0, + "step": 9341 + }, + { + "epoch": 1.025916977816824, + "grad_norm": 2.1547675132751465, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7097612619400024, + "num_tokens": 239593209.0, + "step": 9342 + }, + { + "epoch": 1.0260267955194378, + "grad_norm": 2.1777913570404053, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7381451725959778, + "num_tokens": 239617511.0, + "step": 9343 + }, + { + "epoch": 1.0261366132220513, + "grad_norm": 2.1092417240142822, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7253105044364929, + "num_tokens": 239644597.0, + "step": 9344 + }, + { + "epoch": 1.026246430924665, + "grad_norm": 2.0205042362213135, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7012214660644531, + "num_tokens": 239677239.0, + "step": 9345 + }, + { + "epoch": 1.0263562486272788, + "grad_norm": 2.3680336475372314, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7259511351585388, + "num_tokens": 239701993.0, + "step": 9346 + }, + { + "epoch": 1.0264660663298923, + "grad_norm": 1.9814811944961548, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7223219871520996, + "num_tokens": 239730400.0, + "step": 9347 + }, + { + "epoch": 1.026575884032506, + "grad_norm": 2.53867244720459, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7227733731269836, + "num_tokens": 239750592.0, + "step": 9348 + }, + { + "epoch": 1.0266857017351196, + "grad_norm": 2.1226775646209717, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7018904685974121, + "num_tokens": 239779044.0, + "step": 9349 + }, + { + "epoch": 1.0267955194377334, + "grad_norm": 2.3362882137298584, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.704309344291687, + "num_tokens": 239804345.0, + "step": 9350 + }, + { + "epoch": 1.026905337140347, + "grad_norm": 2.2157187461853027, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7377341985702515, + "num_tokens": 239829280.0, + "step": 9351 + }, + { + "epoch": 1.0270151548429607, + "grad_norm": 2.2065749168395996, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7124173045158386, + "num_tokens": 239856993.0, + "step": 9352 + }, + { + "epoch": 1.0271249725455744, + "grad_norm": 2.2285239696502686, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7409236431121826, + "num_tokens": 239880719.0, + "step": 9353 + }, + { + "epoch": 1.027234790248188, + "grad_norm": 2.003584146499634, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7243432998657227, + "num_tokens": 239908335.0, + "step": 9354 + }, + { + "epoch": 1.0273446079508017, + "grad_norm": 2.273592472076416, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7213540077209473, + "num_tokens": 239934294.0, + "step": 9355 + }, + { + "epoch": 1.0274544256534153, + "grad_norm": 2.51021671295166, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.709571123123169, + "num_tokens": 239956244.0, + "step": 9356 + }, + { + "epoch": 1.027564243356029, + "grad_norm": 2.027576208114624, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7082337141036987, + "num_tokens": 239985536.0, + "step": 9357 + }, + { + "epoch": 1.0276740610586426, + "grad_norm": 2.576284885406494, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7033170461654663, + "num_tokens": 240007115.0, + "step": 9358 + }, + { + "epoch": 1.0277838787612563, + "grad_norm": 2.190483808517456, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7075176239013672, + "num_tokens": 240032202.0, + "step": 9359 + }, + { + "epoch": 1.02789369646387, + "grad_norm": 2.143730401992798, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7159668207168579, + "num_tokens": 240059883.0, + "step": 9360 + }, + { + "epoch": 1.0280035141664836, + "grad_norm": 2.0480823516845703, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7020176649093628, + "num_tokens": 240094177.0, + "step": 9361 + }, + { + "epoch": 1.0281133318690974, + "grad_norm": 2.131622791290283, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7275322079658508, + "num_tokens": 240121828.0, + "step": 9362 + }, + { + "epoch": 1.028223149571711, + "grad_norm": 2.1316354274749756, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6895450949668884, + "num_tokens": 240150126.0, + "step": 9363 + }, + { + "epoch": 1.0283329672743247, + "grad_norm": 2.307598352432251, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7361787557601929, + "num_tokens": 240172928.0, + "step": 9364 + }, + { + "epoch": 1.0284427849769382, + "grad_norm": 2.0973927974700928, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7064049243927002, + "num_tokens": 240203921.0, + "step": 9365 + }, + { + "epoch": 1.028552602679552, + "grad_norm": 2.637706756591797, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7053084373474121, + "num_tokens": 240225074.0, + "step": 9366 + }, + { + "epoch": 1.0286624203821657, + "grad_norm": 2.531123638153076, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7369047999382019, + "num_tokens": 240245381.0, + "step": 9367 + }, + { + "epoch": 1.0287722380847792, + "grad_norm": 2.3740999698638916, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7383050918579102, + "num_tokens": 240270288.0, + "step": 9368 + }, + { + "epoch": 1.028882055787393, + "grad_norm": 2.354977607727051, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7406625151634216, + "num_tokens": 240292789.0, + "step": 9369 + }, + { + "epoch": 1.0289918734900065, + "grad_norm": 2.3514201641082764, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7304143309593201, + "num_tokens": 240319703.0, + "step": 9370 + }, + { + "epoch": 1.0291016911926203, + "grad_norm": 2.1930129528045654, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6864926815032959, + "num_tokens": 240350588.0, + "step": 9371 + }, + { + "epoch": 1.0292115088952338, + "grad_norm": 2.127600908279419, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7197312712669373, + "num_tokens": 240380061.0, + "step": 9372 + }, + { + "epoch": 1.0293213265978476, + "grad_norm": 2.591489315032959, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7333866357803345, + "num_tokens": 240401338.0, + "step": 9373 + }, + { + "epoch": 1.0294311443004613, + "grad_norm": 2.238144636154175, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7197755575180054, + "num_tokens": 240427346.0, + "step": 9374 + }, + { + "epoch": 1.0295409620030749, + "grad_norm": 2.103982925415039, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7055644989013672, + "num_tokens": 240456945.0, + "step": 9375 + }, + { + "epoch": 1.0296507797056886, + "grad_norm": 2.0603504180908203, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.706824779510498, + "num_tokens": 240485236.0, + "step": 9376 + }, + { + "epoch": 1.0297605974083022, + "grad_norm": 2.4119367599487305, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7147602438926697, + "num_tokens": 240508720.0, + "step": 9377 + }, + { + "epoch": 1.029870415110916, + "grad_norm": 2.1886417865753174, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7005782723426819, + "num_tokens": 240537835.0, + "step": 9378 + }, + { + "epoch": 1.0299802328135295, + "grad_norm": 2.5591495037078857, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7291452884674072, + "num_tokens": 240558065.0, + "step": 9379 + }, + { + "epoch": 1.0300900505161432, + "grad_norm": 2.044886589050293, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6964966654777527, + "num_tokens": 240589731.0, + "step": 9380 + }, + { + "epoch": 1.030199868218757, + "grad_norm": 2.024151563644409, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7030059099197388, + "num_tokens": 240617114.0, + "step": 9381 + }, + { + "epoch": 1.0303096859213705, + "grad_norm": 2.346879243850708, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7151801586151123, + "num_tokens": 240640295.0, + "step": 9382 + }, + { + "epoch": 1.0304195036239843, + "grad_norm": 2.1185553073883057, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6925237774848938, + "num_tokens": 240670233.0, + "step": 9383 + }, + { + "epoch": 1.0305293213265978, + "grad_norm": 2.1497268676757812, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7170482873916626, + "num_tokens": 240697344.0, + "step": 9384 + }, + { + "epoch": 1.0306391390292116, + "grad_norm": 2.306621551513672, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7104448676109314, + "num_tokens": 240721949.0, + "step": 9385 + }, + { + "epoch": 1.030748956731825, + "grad_norm": 2.271472215652466, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7093505859375, + "num_tokens": 240747113.0, + "step": 9386 + }, + { + "epoch": 1.0308587744344389, + "grad_norm": 2.27892804145813, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6852196455001831, + "num_tokens": 240773476.0, + "step": 9387 + }, + { + "epoch": 1.0309685921370524, + "grad_norm": 2.3597235679626465, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7122849225997925, + "num_tokens": 240797129.0, + "step": 9388 + }, + { + "epoch": 1.0310784098396661, + "grad_norm": 2.345858573913574, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7099079489707947, + "num_tokens": 240822342.0, + "step": 9389 + }, + { + "epoch": 1.03118822754228, + "grad_norm": 2.173173427581787, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7091407775878906, + "num_tokens": 240851419.0, + "step": 9390 + }, + { + "epoch": 1.0312980452448934, + "grad_norm": 2.194106340408325, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7089076042175293, + "num_tokens": 240877687.0, + "step": 9391 + }, + { + "epoch": 1.0314078629475072, + "grad_norm": 1.9176138639450073, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7202720046043396, + "num_tokens": 240911350.0, + "step": 9392 + }, + { + "epoch": 1.0315176806501207, + "grad_norm": 2.2912991046905518, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7321111559867859, + "num_tokens": 240932876.0, + "step": 9393 + }, + { + "epoch": 1.0316274983527345, + "grad_norm": 2.4102485179901123, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7111148834228516, + "num_tokens": 240955099.0, + "step": 9394 + }, + { + "epoch": 1.031737316055348, + "grad_norm": 2.4969940185546875, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7248246669769287, + "num_tokens": 240977297.0, + "step": 9395 + }, + { + "epoch": 1.0318471337579618, + "grad_norm": 2.665492534637451, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.744678795337677, + "num_tokens": 240996689.0, + "step": 9396 + }, + { + "epoch": 1.0319569514605755, + "grad_norm": 2.1863787174224854, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.705988883972168, + "num_tokens": 241024817.0, + "step": 9397 + }, + { + "epoch": 1.032066769163189, + "grad_norm": 2.387483596801758, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7196328639984131, + "num_tokens": 241047568.0, + "step": 9398 + }, + { + "epoch": 1.0321765868658028, + "grad_norm": 2.42415452003479, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6908462643623352, + "num_tokens": 241072486.0, + "step": 9399 + }, + { + "epoch": 1.0322864045684164, + "grad_norm": 2.1815104484558105, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7242059111595154, + "num_tokens": 241100869.0, + "step": 9400 + }, + { + "epoch": 1.0323962222710301, + "grad_norm": 2.4242470264434814, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6968380808830261, + "num_tokens": 241126720.0, + "step": 9401 + }, + { + "epoch": 1.0325060399736437, + "grad_norm": 2.278822660446167, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7212343811988831, + "num_tokens": 241150978.0, + "step": 9402 + }, + { + "epoch": 1.0326158576762574, + "grad_norm": 2.3398983478546143, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7250423431396484, + "num_tokens": 241176020.0, + "step": 9403 + }, + { + "epoch": 1.0327256753788712, + "grad_norm": 2.016491413116455, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.708061158657074, + "num_tokens": 241205611.0, + "step": 9404 + }, + { + "epoch": 1.0328354930814847, + "grad_norm": 2.259235143661499, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7316556572914124, + "num_tokens": 241231167.0, + "step": 9405 + }, + { + "epoch": 1.0329453107840985, + "grad_norm": 2.286432981491089, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7117128372192383, + "num_tokens": 241254949.0, + "step": 9406 + }, + { + "epoch": 1.033055128486712, + "grad_norm": 2.2222559452056885, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.710707426071167, + "num_tokens": 241280001.0, + "step": 9407 + }, + { + "epoch": 1.0331649461893258, + "grad_norm": 1.9577720165252686, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7172970771789551, + "num_tokens": 241315127.0, + "step": 9408 + }, + { + "epoch": 1.0332747638919393, + "grad_norm": 2.3273978233337402, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7131544351577759, + "num_tokens": 241339835.0, + "step": 9409 + }, + { + "epoch": 1.033384581594553, + "grad_norm": 2.2321436405181885, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7068963050842285, + "num_tokens": 241366217.0, + "step": 9410 + }, + { + "epoch": 1.0334943992971668, + "grad_norm": 2.318917989730835, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7215843200683594, + "num_tokens": 241390077.0, + "step": 9411 + }, + { + "epoch": 1.0336042169997803, + "grad_norm": 2.0523438453674316, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.711959719657898, + "num_tokens": 241418247.0, + "step": 9412 + }, + { + "epoch": 1.033714034702394, + "grad_norm": 2.2985570430755615, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7023163437843323, + "num_tokens": 241443354.0, + "step": 9413 + }, + { + "epoch": 1.0338238524050076, + "grad_norm": 2.2963948249816895, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.715461015701294, + "num_tokens": 241467310.0, + "step": 9414 + }, + { + "epoch": 1.0339336701076214, + "grad_norm": 2.5300583839416504, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7176082134246826, + "num_tokens": 241487543.0, + "step": 9415 + }, + { + "epoch": 1.034043487810235, + "grad_norm": 2.3575282096862793, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7058143615722656, + "num_tokens": 241511184.0, + "step": 9416 + }, + { + "epoch": 1.0341533055128487, + "grad_norm": 2.4312071800231934, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7360966205596924, + "num_tokens": 241531685.0, + "step": 9417 + }, + { + "epoch": 1.0342631232154624, + "grad_norm": 2.310727596282959, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7203561663627625, + "num_tokens": 241555067.0, + "step": 9418 + }, + { + "epoch": 1.034372940918076, + "grad_norm": 2.399688243865967, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7257024645805359, + "num_tokens": 241580036.0, + "step": 9419 + }, + { + "epoch": 1.0344827586206897, + "grad_norm": 2.065160036087036, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.729852557182312, + "num_tokens": 241607654.0, + "step": 9420 + }, + { + "epoch": 1.0345925763233033, + "grad_norm": 2.011777639389038, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.6990317702293396, + "num_tokens": 241637254.0, + "step": 9421 + }, + { + "epoch": 1.034702394025917, + "grad_norm": 2.3530731201171875, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7183674573898315, + "num_tokens": 241662030.0, + "step": 9422 + }, + { + "epoch": 1.0348122117285306, + "grad_norm": 2.4490528106689453, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7119107246398926, + "num_tokens": 241684343.0, + "step": 9423 + }, + { + "epoch": 1.0349220294311443, + "grad_norm": 2.5570719242095947, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7410192489624023, + "num_tokens": 241704082.0, + "step": 9424 + }, + { + "epoch": 1.035031847133758, + "grad_norm": 2.1371610164642334, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6976166367530823, + "num_tokens": 241732643.0, + "step": 9425 + }, + { + "epoch": 1.0351416648363716, + "grad_norm": 2.1247472763061523, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7251272797584534, + "num_tokens": 241758369.0, + "step": 9426 + }, + { + "epoch": 1.0352514825389854, + "grad_norm": 2.0425491333007812, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.694844663143158, + "num_tokens": 241787061.0, + "step": 9427 + }, + { + "epoch": 1.035361300241599, + "grad_norm": 2.472352981567383, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.718119740486145, + "num_tokens": 241809155.0, + "step": 9428 + }, + { + "epoch": 1.0354711179442126, + "grad_norm": 1.9774643182754517, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7020822763442993, + "num_tokens": 241839563.0, + "step": 9429 + }, + { + "epoch": 1.0355809356468262, + "grad_norm": 2.2905778884887695, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7264269590377808, + "num_tokens": 241863552.0, + "step": 9430 + }, + { + "epoch": 1.03569075334944, + "grad_norm": 2.18645977973938, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7084652781486511, + "num_tokens": 241890971.0, + "step": 9431 + }, + { + "epoch": 1.0358005710520537, + "grad_norm": 2.0826148986816406, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6994134187698364, + "num_tokens": 241918529.0, + "step": 9432 + }, + { + "epoch": 1.0359103887546672, + "grad_norm": 2.169672727584839, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6965596079826355, + "num_tokens": 241946538.0, + "step": 9433 + }, + { + "epoch": 1.036020206457281, + "grad_norm": 2.1614694595336914, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.709693193435669, + "num_tokens": 241974338.0, + "step": 9434 + }, + { + "epoch": 1.0361300241598945, + "grad_norm": 2.0587546825408936, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7086184620857239, + "num_tokens": 242007937.0, + "step": 9435 + }, + { + "epoch": 1.0362398418625083, + "grad_norm": 2.25292706489563, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7111314535140991, + "num_tokens": 242034332.0, + "step": 9436 + }, + { + "epoch": 1.0363496595651218, + "grad_norm": 2.201038360595703, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.725115180015564, + "num_tokens": 242061322.0, + "step": 9437 + }, + { + "epoch": 1.0364594772677356, + "grad_norm": 2.2676987648010254, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7468594312667847, + "num_tokens": 242083645.0, + "step": 9438 + }, + { + "epoch": 1.036569294970349, + "grad_norm": 2.1862974166870117, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7156145572662354, + "num_tokens": 242112452.0, + "step": 9439 + }, + { + "epoch": 1.0366791126729629, + "grad_norm": 2.3695380687713623, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.723261833190918, + "num_tokens": 242134353.0, + "step": 9440 + }, + { + "epoch": 1.0367889303755766, + "grad_norm": 2.161623001098633, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7114993333816528, + "num_tokens": 242162567.0, + "step": 9441 + }, + { + "epoch": 1.0368987480781902, + "grad_norm": 2.1730077266693115, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7401471138000488, + "num_tokens": 242186757.0, + "step": 9442 + }, + { + "epoch": 1.037008565780804, + "grad_norm": 2.11491322517395, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7027758359909058, + "num_tokens": 242218173.0, + "step": 9443 + }, + { + "epoch": 1.0371183834834174, + "grad_norm": 1.8967558145523071, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7030035257339478, + "num_tokens": 242253321.0, + "step": 9444 + }, + { + "epoch": 1.0372282011860312, + "grad_norm": 2.1774725914001465, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7219800353050232, + "num_tokens": 242281049.0, + "step": 9445 + }, + { + "epoch": 1.037338018888645, + "grad_norm": 2.4237992763519287, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7150740623474121, + "num_tokens": 242304027.0, + "step": 9446 + }, + { + "epoch": 1.0374478365912585, + "grad_norm": 2.5155913829803467, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7320044636726379, + "num_tokens": 242325247.0, + "step": 9447 + }, + { + "epoch": 1.0375576542938723, + "grad_norm": 2.5191874504089355, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7221840023994446, + "num_tokens": 242345816.0, + "step": 9448 + }, + { + "epoch": 1.0376674719964858, + "grad_norm": 2.297715663909912, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.703662097454071, + "num_tokens": 242370070.0, + "step": 9449 + }, + { + "epoch": 1.0377772896990995, + "grad_norm": 2.214864492416382, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7269067764282227, + "num_tokens": 242394727.0, + "step": 9450 + }, + { + "epoch": 1.037887107401713, + "grad_norm": 2.3330624103546143, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.712165117263794, + "num_tokens": 242417891.0, + "step": 9451 + }, + { + "epoch": 1.0379969251043268, + "grad_norm": 2.3224756717681885, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7173174619674683, + "num_tokens": 242441601.0, + "step": 9452 + }, + { + "epoch": 1.0381067428069404, + "grad_norm": 2.1964902877807617, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7326666116714478, + "num_tokens": 242467263.0, + "step": 9453 + }, + { + "epoch": 1.0382165605095541, + "grad_norm": 2.2602791786193848, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7161226272583008, + "num_tokens": 242493646.0, + "step": 9454 + }, + { + "epoch": 1.0383263782121679, + "grad_norm": 2.557094097137451, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7270928621292114, + "num_tokens": 242512820.0, + "step": 9455 + }, + { + "epoch": 1.0384361959147814, + "grad_norm": 2.3959362506866455, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7373554706573486, + "num_tokens": 242534142.0, + "step": 9456 + }, + { + "epoch": 1.0385460136173952, + "grad_norm": 2.362032175064087, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7052286863327026, + "num_tokens": 242556255.0, + "step": 9457 + }, + { + "epoch": 1.0386558313200087, + "grad_norm": 2.3226425647735596, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7202515006065369, + "num_tokens": 242580431.0, + "step": 9458 + }, + { + "epoch": 1.0387656490226225, + "grad_norm": 2.023104190826416, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7181771993637085, + "num_tokens": 242610981.0, + "step": 9459 + }, + { + "epoch": 1.038875466725236, + "grad_norm": 2.517139434814453, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7100750803947449, + "num_tokens": 242631467.0, + "step": 9460 + }, + { + "epoch": 1.0389852844278498, + "grad_norm": 2.4434423446655273, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7377689480781555, + "num_tokens": 242652511.0, + "step": 9461 + }, + { + "epoch": 1.0390951021304635, + "grad_norm": 2.211460828781128, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7122370600700378, + "num_tokens": 242679889.0, + "step": 9462 + }, + { + "epoch": 1.039204919833077, + "grad_norm": 2.2039365768432617, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7125359773635864, + "num_tokens": 242704867.0, + "step": 9463 + }, + { + "epoch": 1.0393147375356908, + "grad_norm": 2.691771984100342, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7304309606552124, + "num_tokens": 242725531.0, + "step": 9464 + }, + { + "epoch": 1.0394245552383043, + "grad_norm": 2.3402202129364014, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7220571041107178, + "num_tokens": 242750948.0, + "step": 9465 + }, + { + "epoch": 1.039534372940918, + "grad_norm": 2.133387565612793, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7088107466697693, + "num_tokens": 242780040.0, + "step": 9466 + }, + { + "epoch": 1.0396441906435316, + "grad_norm": 2.2924728393554688, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7260634303092957, + "num_tokens": 242804669.0, + "step": 9467 + }, + { + "epoch": 1.0397540083461454, + "grad_norm": 2.2051072120666504, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7304564714431763, + "num_tokens": 242831297.0, + "step": 9468 + }, + { + "epoch": 1.0398638260487592, + "grad_norm": 2.0422801971435547, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6957312822341919, + "num_tokens": 242861249.0, + "step": 9469 + }, + { + "epoch": 1.0399736437513727, + "grad_norm": 2.0618584156036377, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7100350856781006, + "num_tokens": 242892019.0, + "step": 9470 + }, + { + "epoch": 1.0400834614539864, + "grad_norm": 2.720797061920166, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7276870012283325, + "num_tokens": 242911313.0, + "step": 9471 + }, + { + "epoch": 1.0401932791566, + "grad_norm": 2.104442834854126, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7042863368988037, + "num_tokens": 242941008.0, + "step": 9472 + }, + { + "epoch": 1.0403030968592137, + "grad_norm": 2.2772138118743896, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7020639777183533, + "num_tokens": 242966642.0, + "step": 9473 + }, + { + "epoch": 1.0404129145618273, + "grad_norm": 2.3282077312469482, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.73324054479599, + "num_tokens": 242989339.0, + "step": 9474 + }, + { + "epoch": 1.040522732264441, + "grad_norm": 2.374755859375, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7274891138076782, + "num_tokens": 243012350.0, + "step": 9475 + }, + { + "epoch": 1.0406325499670548, + "grad_norm": 2.673931121826172, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7137863636016846, + "num_tokens": 243031463.0, + "step": 9476 + }, + { + "epoch": 1.0407423676696683, + "grad_norm": 2.1360440254211426, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7104992866516113, + "num_tokens": 243058482.0, + "step": 9477 + }, + { + "epoch": 1.040852185372282, + "grad_norm": 1.8827354907989502, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7185652256011963, + "num_tokens": 243092743.0, + "step": 9478 + }, + { + "epoch": 1.0409620030748956, + "grad_norm": 2.240509271621704, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7054398059844971, + "num_tokens": 243118944.0, + "step": 9479 + }, + { + "epoch": 1.0410718207775094, + "grad_norm": 2.26405930519104, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7045897245407104, + "num_tokens": 243145172.0, + "step": 9480 + }, + { + "epoch": 1.041181638480123, + "grad_norm": 2.010009765625, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7260851263999939, + "num_tokens": 243175790.0, + "step": 9481 + }, + { + "epoch": 1.0412914561827367, + "grad_norm": 2.0777268409729004, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7198550701141357, + "num_tokens": 243204908.0, + "step": 9482 + }, + { + "epoch": 1.0414012738853504, + "grad_norm": 2.2223503589630127, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7146264314651489, + "num_tokens": 243230401.0, + "step": 9483 + }, + { + "epoch": 1.041511091587964, + "grad_norm": 2.29935622215271, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.6995949745178223, + "num_tokens": 243254655.0, + "step": 9484 + }, + { + "epoch": 1.0416209092905777, + "grad_norm": 1.9622855186462402, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7154890894889832, + "num_tokens": 243286187.0, + "step": 9485 + }, + { + "epoch": 1.0417307269931912, + "grad_norm": 2.445305109024048, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7161149382591248, + "num_tokens": 243309145.0, + "step": 9486 + }, + { + "epoch": 1.041840544695805, + "grad_norm": 2.2352073192596436, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7462472915649414, + "num_tokens": 243331888.0, + "step": 9487 + }, + { + "epoch": 1.0419503623984185, + "grad_norm": 2.0191988945007324, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7045891284942627, + "num_tokens": 243362183.0, + "step": 9488 + }, + { + "epoch": 1.0420601801010323, + "grad_norm": 2.107260227203369, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7271758913993835, + "num_tokens": 243389554.0, + "step": 9489 + }, + { + "epoch": 1.042169997803646, + "grad_norm": 2.3221359252929688, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7123252153396606, + "num_tokens": 243412478.0, + "step": 9490 + }, + { + "epoch": 1.0422798155062596, + "grad_norm": 2.309246301651001, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7166967391967773, + "num_tokens": 243438319.0, + "step": 9491 + }, + { + "epoch": 1.0423896332088733, + "grad_norm": 2.2764363288879395, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7216057777404785, + "num_tokens": 243462011.0, + "step": 9492 + }, + { + "epoch": 1.0424994509114869, + "grad_norm": 2.4047725200653076, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7103581428527832, + "num_tokens": 243485347.0, + "step": 9493 + }, + { + "epoch": 1.0426092686141006, + "grad_norm": 2.402231454849243, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7012417316436768, + "num_tokens": 243506714.0, + "step": 9494 + }, + { + "epoch": 1.0427190863167142, + "grad_norm": 1.8947012424468994, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7206461429595947, + "num_tokens": 243538499.0, + "step": 9495 + }, + { + "epoch": 1.042828904019328, + "grad_norm": 2.1238746643066406, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7313723564147949, + "num_tokens": 243564914.0, + "step": 9496 + }, + { + "epoch": 1.0429387217219417, + "grad_norm": 2.429309844970703, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.734940767288208, + "num_tokens": 243587544.0, + "step": 9497 + }, + { + "epoch": 1.0430485394245552, + "grad_norm": 2.435418128967285, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7370273470878601, + "num_tokens": 243608687.0, + "step": 9498 + }, + { + "epoch": 1.043158357127169, + "grad_norm": 2.1022586822509766, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7137428522109985, + "num_tokens": 243638344.0, + "step": 9499 + }, + { + "epoch": 1.0432681748297825, + "grad_norm": 2.290325880050659, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7276135087013245, + "num_tokens": 243663840.0, + "step": 9500 + }, + { + "epoch": 1.0433779925323963, + "grad_norm": 2.248243808746338, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7051648497581482, + "num_tokens": 243688049.0, + "step": 9501 + }, + { + "epoch": 1.0434878102350098, + "grad_norm": 2.3150477409362793, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7267657518386841, + "num_tokens": 243711329.0, + "step": 9502 + }, + { + "epoch": 1.0435976279376236, + "grad_norm": 2.3894152641296387, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6992369890213013, + "num_tokens": 243734868.0, + "step": 9503 + }, + { + "epoch": 1.043707445640237, + "grad_norm": 2.243537425994873, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7323812246322632, + "num_tokens": 243759033.0, + "step": 9504 + }, + { + "epoch": 1.0438172633428509, + "grad_norm": 2.3814029693603516, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7198746204376221, + "num_tokens": 243782005.0, + "step": 9505 + }, + { + "epoch": 1.0439270810454646, + "grad_norm": 2.5017449855804443, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.72331702709198, + "num_tokens": 243804798.0, + "step": 9506 + }, + { + "epoch": 1.0440368987480781, + "grad_norm": 2.2996978759765625, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7197417616844177, + "num_tokens": 243829769.0, + "step": 9507 + }, + { + "epoch": 1.044146716450692, + "grad_norm": 2.437419891357422, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7191644906997681, + "num_tokens": 243852706.0, + "step": 9508 + }, + { + "epoch": 1.0442565341533054, + "grad_norm": 2.003716468811035, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7176717519760132, + "num_tokens": 243882826.0, + "step": 9509 + }, + { + "epoch": 1.0443663518559192, + "grad_norm": 2.37060284614563, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7371296286582947, + "num_tokens": 243905614.0, + "step": 9510 + }, + { + "epoch": 1.0444761695585327, + "grad_norm": 2.2270712852478027, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7332437634468079, + "num_tokens": 243929261.0, + "step": 9511 + }, + { + "epoch": 1.0445859872611465, + "grad_norm": 2.6210765838623047, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7400007247924805, + "num_tokens": 243948013.0, + "step": 9512 + }, + { + "epoch": 1.0446958049637602, + "grad_norm": 2.2277841567993164, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7277364730834961, + "num_tokens": 243973196.0, + "step": 9513 + }, + { + "epoch": 1.0448056226663738, + "grad_norm": 2.2485413551330566, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7276136875152588, + "num_tokens": 243996706.0, + "step": 9514 + }, + { + "epoch": 1.0449154403689875, + "grad_norm": 2.1481025218963623, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6914886236190796, + "num_tokens": 244023407.0, + "step": 9515 + }, + { + "epoch": 1.045025258071601, + "grad_norm": 2.4494566917419434, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.736221194267273, + "num_tokens": 244048266.0, + "step": 9516 + }, + { + "epoch": 1.0451350757742148, + "grad_norm": 2.4170796871185303, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7250255346298218, + "num_tokens": 244071515.0, + "step": 9517 + }, + { + "epoch": 1.0452448934768284, + "grad_norm": 2.3065266609191895, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.688675045967102, + "num_tokens": 244098785.0, + "step": 9518 + }, + { + "epoch": 1.0453547111794421, + "grad_norm": 2.429150342941284, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7023121118545532, + "num_tokens": 244122693.0, + "step": 9519 + }, + { + "epoch": 1.0454645288820559, + "grad_norm": 2.064620018005371, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7038860321044922, + "num_tokens": 244152290.0, + "step": 9520 + }, + { + "epoch": 1.0455743465846694, + "grad_norm": 2.1724648475646973, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7016541957855225, + "num_tokens": 244181972.0, + "step": 9521 + }, + { + "epoch": 1.0456841642872832, + "grad_norm": 2.1760053634643555, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7123022079467773, + "num_tokens": 244208533.0, + "step": 9522 + }, + { + "epoch": 1.0457939819898967, + "grad_norm": 2.1365511417388916, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6976911425590515, + "num_tokens": 244237738.0, + "step": 9523 + }, + { + "epoch": 1.0459037996925105, + "grad_norm": 1.939325213432312, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7374973893165588, + "num_tokens": 244268411.0, + "step": 9524 + }, + { + "epoch": 1.046013617395124, + "grad_norm": 2.4846816062927246, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7232649922370911, + "num_tokens": 244289600.0, + "step": 9525 + }, + { + "epoch": 1.0461234350977378, + "grad_norm": 2.152829647064209, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.726224958896637, + "num_tokens": 244316233.0, + "step": 9526 + }, + { + "epoch": 1.0462332528003515, + "grad_norm": 2.360563278198242, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7192206382751465, + "num_tokens": 244341042.0, + "step": 9527 + }, + { + "epoch": 1.046343070502965, + "grad_norm": 2.221590995788574, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7145880460739136, + "num_tokens": 244365898.0, + "step": 9528 + }, + { + "epoch": 1.0464528882055788, + "grad_norm": 2.050710678100586, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7289927005767822, + "num_tokens": 244394424.0, + "step": 9529 + }, + { + "epoch": 1.0465627059081923, + "grad_norm": 2.0930731296539307, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7136136293411255, + "num_tokens": 244423162.0, + "step": 9530 + }, + { + "epoch": 1.046672523610806, + "grad_norm": 2.009791851043701, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7172304391860962, + "num_tokens": 244453509.0, + "step": 9531 + }, + { + "epoch": 1.0467823413134196, + "grad_norm": 2.5936455726623535, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7380591630935669, + "num_tokens": 244472447.0, + "step": 9532 + }, + { + "epoch": 1.0468921590160334, + "grad_norm": 2.453615665435791, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7357022166252136, + "num_tokens": 244495008.0, + "step": 9533 + }, + { + "epoch": 1.0470019767186471, + "grad_norm": 2.21081805229187, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7451010942459106, + "num_tokens": 244519789.0, + "step": 9534 + }, + { + "epoch": 1.0471117944212607, + "grad_norm": 2.456190347671509, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7277743816375732, + "num_tokens": 244542017.0, + "step": 9535 + }, + { + "epoch": 1.0472216121238744, + "grad_norm": 2.245664596557617, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7086898684501648, + "num_tokens": 244567072.0, + "step": 9536 + }, + { + "epoch": 1.047331429826488, + "grad_norm": 2.122633457183838, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7183467149734497, + "num_tokens": 244594079.0, + "step": 9537 + }, + { + "epoch": 1.0474412475291017, + "grad_norm": 2.240877389907837, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7220407724380493, + "num_tokens": 244621824.0, + "step": 9538 + }, + { + "epoch": 1.0475510652317153, + "grad_norm": 2.1971211433410645, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7112948894500732, + "num_tokens": 244649302.0, + "step": 9539 + }, + { + "epoch": 1.047660882934329, + "grad_norm": 2.2435238361358643, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7267776131629944, + "num_tokens": 244676006.0, + "step": 9540 + }, + { + "epoch": 1.0477707006369428, + "grad_norm": 2.4262325763702393, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7256073355674744, + "num_tokens": 244696745.0, + "step": 9541 + }, + { + "epoch": 1.0478805183395563, + "grad_norm": 2.7991154193878174, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7150646448135376, + "num_tokens": 244714436.0, + "step": 9542 + }, + { + "epoch": 1.04799033604217, + "grad_norm": 2.0743207931518555, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6836403608322144, + "num_tokens": 244746598.0, + "step": 9543 + }, + { + "epoch": 1.0481001537447836, + "grad_norm": 2.1824474334716797, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.690072774887085, + "num_tokens": 244774407.0, + "step": 9544 + }, + { + "epoch": 1.0482099714473974, + "grad_norm": 2.3843650817871094, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7450046539306641, + "num_tokens": 244795524.0, + "step": 9545 + }, + { + "epoch": 1.048319789150011, + "grad_norm": 2.2729268074035645, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6988646984100342, + "num_tokens": 244821053.0, + "step": 9546 + }, + { + "epoch": 1.0484296068526247, + "grad_norm": 2.156050443649292, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7042832374572754, + "num_tokens": 244849731.0, + "step": 9547 + }, + { + "epoch": 1.0485394245552384, + "grad_norm": 2.134157657623291, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7054272294044495, + "num_tokens": 244876965.0, + "step": 9548 + }, + { + "epoch": 1.048649242257852, + "grad_norm": 2.1025750637054443, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7370406985282898, + "num_tokens": 244903999.0, + "step": 9549 + }, + { + "epoch": 1.0487590599604657, + "grad_norm": 2.272484064102173, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7123241424560547, + "num_tokens": 244927082.0, + "step": 9550 + }, + { + "epoch": 1.0488688776630792, + "grad_norm": 2.349684476852417, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.71418297290802, + "num_tokens": 244949593.0, + "step": 9551 + }, + { + "epoch": 1.048978695365693, + "grad_norm": 2.2612826824188232, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7095293998718262, + "num_tokens": 244973312.0, + "step": 9552 + }, + { + "epoch": 1.0490885130683065, + "grad_norm": 2.5194921493530273, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7384649515151978, + "num_tokens": 244993661.0, + "step": 9553 + }, + { + "epoch": 1.0491983307709203, + "grad_norm": 2.262244462966919, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7417933940887451, + "num_tokens": 245018264.0, + "step": 9554 + }, + { + "epoch": 1.0493081484735338, + "grad_norm": 2.191229820251465, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7012861967086792, + "num_tokens": 245046027.0, + "step": 9555 + }, + { + "epoch": 1.0494179661761476, + "grad_norm": 2.326355218887329, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.700009286403656, + "num_tokens": 245069560.0, + "step": 9556 + }, + { + "epoch": 1.0495277838787613, + "grad_norm": 1.9719557762145996, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7144532203674316, + "num_tokens": 245101705.0, + "step": 9557 + }, + { + "epoch": 1.0496376015813749, + "grad_norm": 2.2358899116516113, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7011773586273193, + "num_tokens": 245129534.0, + "step": 9558 + }, + { + "epoch": 1.0497474192839886, + "grad_norm": 2.2253048419952393, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.706608772277832, + "num_tokens": 245155845.0, + "step": 9559 + }, + { + "epoch": 1.0498572369866022, + "grad_norm": 2.4533584117889404, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7168275117874146, + "num_tokens": 245177297.0, + "step": 9560 + }, + { + "epoch": 1.049967054689216, + "grad_norm": 2.4324285984039307, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7218424677848816, + "num_tokens": 245200465.0, + "step": 9561 + }, + { + "epoch": 1.0500768723918297, + "grad_norm": 1.8589688539505005, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7111228704452515, + "num_tokens": 245234852.0, + "step": 9562 + }, + { + "epoch": 1.0501866900944432, + "grad_norm": 2.4258196353912354, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7282279133796692, + "num_tokens": 245256113.0, + "step": 9563 + }, + { + "epoch": 1.050296507797057, + "grad_norm": 2.467045307159424, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7140576243400574, + "num_tokens": 245278491.0, + "step": 9564 + }, + { + "epoch": 1.0504063254996705, + "grad_norm": 2.0293469429016113, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7189792990684509, + "num_tokens": 245310916.0, + "step": 9565 + }, + { + "epoch": 1.0505161432022843, + "grad_norm": 2.0063021183013916, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7023677825927734, + "num_tokens": 245343451.0, + "step": 9566 + }, + { + "epoch": 1.0506259609048978, + "grad_norm": 2.295319080352783, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7159200310707092, + "num_tokens": 245367317.0, + "step": 9567 + }, + { + "epoch": 1.0507357786075116, + "grad_norm": 2.486081123352051, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7323113083839417, + "num_tokens": 245389022.0, + "step": 9568 + }, + { + "epoch": 1.050845596310125, + "grad_norm": 2.390627384185791, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7099611759185791, + "num_tokens": 245412151.0, + "step": 9569 + }, + { + "epoch": 1.0509554140127388, + "grad_norm": 2.583047866821289, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7459778785705566, + "num_tokens": 245431057.0, + "step": 9570 + }, + { + "epoch": 1.0510652317153526, + "grad_norm": 2.387899875640869, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6944741010665894, + "num_tokens": 245456519.0, + "step": 9571 + }, + { + "epoch": 1.0511750494179661, + "grad_norm": 2.3402023315429688, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.717303991317749, + "num_tokens": 245479690.0, + "step": 9572 + }, + { + "epoch": 1.05128486712058, + "grad_norm": 2.1712968349456787, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.757552444934845, + "num_tokens": 245506370.0, + "step": 9573 + }, + { + "epoch": 1.0513946848231934, + "grad_norm": 2.1705429553985596, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7149685621261597, + "num_tokens": 245532457.0, + "step": 9574 + }, + { + "epoch": 1.0515045025258072, + "grad_norm": 2.4369215965270996, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7255338430404663, + "num_tokens": 245555202.0, + "step": 9575 + }, + { + "epoch": 1.0516143202284207, + "grad_norm": 2.1908795833587646, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7334249019622803, + "num_tokens": 245580873.0, + "step": 9576 + }, + { + "epoch": 1.0517241379310345, + "grad_norm": 2.2126340866088867, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.712472140789032, + "num_tokens": 245606711.0, + "step": 9577 + }, + { + "epoch": 1.0518339556336482, + "grad_norm": 2.117246627807617, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7229642868041992, + "num_tokens": 245634299.0, + "step": 9578 + }, + { + "epoch": 1.0519437733362618, + "grad_norm": 2.540802240371704, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7378277778625488, + "num_tokens": 245653161.0, + "step": 9579 + }, + { + "epoch": 1.0520535910388755, + "grad_norm": 2.2494142055511475, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7166124582290649, + "num_tokens": 245678191.0, + "step": 9580 + }, + { + "epoch": 1.052163408741489, + "grad_norm": 2.4017257690429688, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7293527126312256, + "num_tokens": 245700138.0, + "step": 9581 + }, + { + "epoch": 1.0522732264441028, + "grad_norm": 2.2025392055511475, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7216317653656006, + "num_tokens": 245726838.0, + "step": 9582 + }, + { + "epoch": 1.0523830441467164, + "grad_norm": 2.161412239074707, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6920859813690186, + "num_tokens": 245754157.0, + "step": 9583 + }, + { + "epoch": 1.05249286184933, + "grad_norm": 2.086172580718994, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6965471506118774, + "num_tokens": 245784256.0, + "step": 9584 + }, + { + "epoch": 1.0526026795519439, + "grad_norm": 2.1595137119293213, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7232716083526611, + "num_tokens": 245810560.0, + "step": 9585 + }, + { + "epoch": 1.0527124972545574, + "grad_norm": 2.4626758098602295, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7117840647697449, + "num_tokens": 245832834.0, + "step": 9586 + }, + { + "epoch": 1.0528223149571712, + "grad_norm": 2.4590985774993896, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.719406247138977, + "num_tokens": 245855253.0, + "step": 9587 + }, + { + "epoch": 1.0529321326597847, + "grad_norm": 2.0174951553344727, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7204475402832031, + "num_tokens": 245885396.0, + "step": 9588 + }, + { + "epoch": 1.0530419503623984, + "grad_norm": 2.5850741863250732, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7416041493415833, + "num_tokens": 245906213.0, + "step": 9589 + }, + { + "epoch": 1.053151768065012, + "grad_norm": 1.945268154144287, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6895394325256348, + "num_tokens": 245939973.0, + "step": 9590 + }, + { + "epoch": 1.0532615857676257, + "grad_norm": 2.4786057472229004, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7235091924667358, + "num_tokens": 245963223.0, + "step": 9591 + }, + { + "epoch": 1.0533714034702395, + "grad_norm": 2.332120895385742, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7075666189193726, + "num_tokens": 245987182.0, + "step": 9592 + }, + { + "epoch": 1.053481221172853, + "grad_norm": 2.123443841934204, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7517609596252441, + "num_tokens": 246013219.0, + "step": 9593 + }, + { + "epoch": 1.0535910388754668, + "grad_norm": 2.1377832889556885, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7017006278038025, + "num_tokens": 246041630.0, + "step": 9594 + }, + { + "epoch": 1.0537008565780803, + "grad_norm": 1.9245644807815552, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6884297132492065, + "num_tokens": 246077480.0, + "step": 9595 + }, + { + "epoch": 1.053810674280694, + "grad_norm": 2.07293438911438, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7319942116737366, + "num_tokens": 246105632.0, + "step": 9596 + }, + { + "epoch": 1.0539204919833076, + "grad_norm": 2.210409164428711, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7293264865875244, + "num_tokens": 246130319.0, + "step": 9597 + }, + { + "epoch": 1.0540303096859214, + "grad_norm": 2.247553586959839, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6966075897216797, + "num_tokens": 246156073.0, + "step": 9598 + }, + { + "epoch": 1.0541401273885351, + "grad_norm": 1.9735690355300903, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7053371667861938, + "num_tokens": 246186885.0, + "step": 9599 + }, + { + "epoch": 1.0542499450911487, + "grad_norm": 1.770219087600708, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6867300868034363, + "num_tokens": 246224448.0, + "step": 9600 + }, + { + "epoch": 1.0543597627937624, + "grad_norm": 2.4765496253967285, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7174884080886841, + "num_tokens": 246245989.0, + "step": 9601 + }, + { + "epoch": 1.054469580496376, + "grad_norm": 2.3949849605560303, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7233589887619019, + "num_tokens": 246269496.0, + "step": 9602 + }, + { + "epoch": 1.0545793981989897, + "grad_norm": 2.4441840648651123, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7167002558708191, + "num_tokens": 246292422.0, + "step": 9603 + }, + { + "epoch": 1.0546892159016032, + "grad_norm": 2.2682297229766846, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7261655926704407, + "num_tokens": 246316676.0, + "step": 9604 + }, + { + "epoch": 1.054799033604217, + "grad_norm": 2.5737485885620117, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7116119861602783, + "num_tokens": 246338495.0, + "step": 9605 + }, + { + "epoch": 1.0549088513068308, + "grad_norm": 2.4648261070251465, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7294247150421143, + "num_tokens": 246360272.0, + "step": 9606 + }, + { + "epoch": 1.0550186690094443, + "grad_norm": 2.183837652206421, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7315418720245361, + "num_tokens": 246387906.0, + "step": 9607 + }, + { + "epoch": 1.055128486712058, + "grad_norm": 2.123169183731079, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6997388601303101, + "num_tokens": 246417402.0, + "step": 9608 + }, + { + "epoch": 1.0552383044146716, + "grad_norm": 2.0913245677948, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7388969659805298, + "num_tokens": 246446359.0, + "step": 9609 + }, + { + "epoch": 1.0553481221172853, + "grad_norm": 2.431751251220703, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7343389987945557, + "num_tokens": 246467362.0, + "step": 9610 + }, + { + "epoch": 1.0554579398198989, + "grad_norm": 2.2729146480560303, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7182849645614624, + "num_tokens": 246493345.0, + "step": 9611 + }, + { + "epoch": 1.0555677575225126, + "grad_norm": 2.374408483505249, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.714693546295166, + "num_tokens": 246516325.0, + "step": 9612 + }, + { + "epoch": 1.0556775752251264, + "grad_norm": 2.136324167251587, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7359974384307861, + "num_tokens": 246542653.0, + "step": 9613 + }, + { + "epoch": 1.05578739292774, + "grad_norm": 2.1629440784454346, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6951266527175903, + "num_tokens": 246570445.0, + "step": 9614 + }, + { + "epoch": 1.0558972106303537, + "grad_norm": 2.6309316158294678, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7101879715919495, + "num_tokens": 246588644.0, + "step": 9615 + }, + { + "epoch": 1.0560070283329672, + "grad_norm": 2.1357309818267822, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7312595248222351, + "num_tokens": 246617819.0, + "step": 9616 + }, + { + "epoch": 1.056116846035581, + "grad_norm": 2.1113948822021484, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7142723202705383, + "num_tokens": 246647323.0, + "step": 9617 + }, + { + "epoch": 1.0562266637381945, + "grad_norm": 2.1165897846221924, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7071002721786499, + "num_tokens": 246676761.0, + "step": 9618 + }, + { + "epoch": 1.0563364814408083, + "grad_norm": 2.26216197013855, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.719699501991272, + "num_tokens": 246699732.0, + "step": 9619 + }, + { + "epoch": 1.0564462991434218, + "grad_norm": 2.0083465576171875, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6914721727371216, + "num_tokens": 246730907.0, + "step": 9620 + }, + { + "epoch": 1.0565561168460356, + "grad_norm": 2.2538092136383057, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7115293145179749, + "num_tokens": 246756107.0, + "step": 9621 + }, + { + "epoch": 1.0566659345486493, + "grad_norm": 2.1463735103607178, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7093935608863831, + "num_tokens": 246784905.0, + "step": 9622 + }, + { + "epoch": 1.0567757522512629, + "grad_norm": 2.2872531414031982, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.730340838432312, + "num_tokens": 246807915.0, + "step": 9623 + }, + { + "epoch": 1.0568855699538766, + "grad_norm": 2.273216485977173, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7092059850692749, + "num_tokens": 246834063.0, + "step": 9624 + }, + { + "epoch": 1.0569953876564901, + "grad_norm": 1.9103707075119019, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6968709826469421, + "num_tokens": 246869334.0, + "step": 9625 + }, + { + "epoch": 1.057105205359104, + "grad_norm": 2.3465306758880615, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.720984697341919, + "num_tokens": 246892992.0, + "step": 9626 + }, + { + "epoch": 1.0572150230617177, + "grad_norm": 2.2351417541503906, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.712687611579895, + "num_tokens": 246919256.0, + "step": 9627 + }, + { + "epoch": 1.0573248407643312, + "grad_norm": 2.431377410888672, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7386056780815125, + "num_tokens": 246941821.0, + "step": 9628 + }, + { + "epoch": 1.057434658466945, + "grad_norm": 2.3910861015319824, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7292305827140808, + "num_tokens": 246964492.0, + "step": 9629 + }, + { + "epoch": 1.0575444761695585, + "grad_norm": 2.1075751781463623, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7412304878234863, + "num_tokens": 246991744.0, + "step": 9630 + }, + { + "epoch": 1.0576542938721722, + "grad_norm": 2.391512155532837, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7421759963035583, + "num_tokens": 247012237.0, + "step": 9631 + }, + { + "epoch": 1.0577641115747858, + "grad_norm": 2.2354488372802734, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7112749814987183, + "num_tokens": 247037767.0, + "step": 9632 + }, + { + "epoch": 1.0578739292773995, + "grad_norm": 2.4251906871795654, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7084256410598755, + "num_tokens": 247061397.0, + "step": 9633 + }, + { + "epoch": 1.057983746980013, + "grad_norm": 2.187150001525879, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7026861906051636, + "num_tokens": 247089605.0, + "step": 9634 + }, + { + "epoch": 1.0580935646826268, + "grad_norm": 2.3198301792144775, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.74953293800354, + "num_tokens": 247111480.0, + "step": 9635 + }, + { + "epoch": 1.0582033823852406, + "grad_norm": 2.3738837242126465, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.6995030045509338, + "num_tokens": 247135921.0, + "step": 9636 + }, + { + "epoch": 1.0583132000878541, + "grad_norm": 2.1582443714141846, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7117946147918701, + "num_tokens": 247162495.0, + "step": 9637 + }, + { + "epoch": 1.0584230177904679, + "grad_norm": 2.278780698776245, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7238434553146362, + "num_tokens": 247185768.0, + "step": 9638 + }, + { + "epoch": 1.0585328354930814, + "grad_norm": 2.2487664222717285, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7104939818382263, + "num_tokens": 247212136.0, + "step": 9639 + }, + { + "epoch": 1.0586426531956952, + "grad_norm": 2.0521109104156494, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7061559557914734, + "num_tokens": 247243356.0, + "step": 9640 + }, + { + "epoch": 1.0587524708983087, + "grad_norm": 2.4111759662628174, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7390937805175781, + "num_tokens": 247265322.0, + "step": 9641 + }, + { + "epoch": 1.0588622886009225, + "grad_norm": 2.224808931350708, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7240529656410217, + "num_tokens": 247291652.0, + "step": 9642 + }, + { + "epoch": 1.0589721063035362, + "grad_norm": 2.437105178833008, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7366822361946106, + "num_tokens": 247315040.0, + "step": 9643 + }, + { + "epoch": 1.0590819240061498, + "grad_norm": 2.56532883644104, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7369237542152405, + "num_tokens": 247335455.0, + "step": 9644 + }, + { + "epoch": 1.0591917417087635, + "grad_norm": 2.1020991802215576, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7356595993041992, + "num_tokens": 247363608.0, + "step": 9645 + }, + { + "epoch": 1.059301559411377, + "grad_norm": 2.2122509479522705, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.699051022529602, + "num_tokens": 247391041.0, + "step": 9646 + }, + { + "epoch": 1.0594113771139908, + "grad_norm": 2.327425003051758, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7218633890151978, + "num_tokens": 247415520.0, + "step": 9647 + }, + { + "epoch": 1.0595211948166043, + "grad_norm": 2.133657693862915, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.715700626373291, + "num_tokens": 247442372.0, + "step": 9648 + }, + { + "epoch": 1.059631012519218, + "grad_norm": 2.1735785007476807, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6931567192077637, + "num_tokens": 247470550.0, + "step": 9649 + }, + { + "epoch": 1.0597408302218319, + "grad_norm": 2.567568302154541, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7066384553909302, + "num_tokens": 247490717.0, + "step": 9650 + }, + { + "epoch": 1.0598506479244454, + "grad_norm": 2.0114879608154297, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.6990700960159302, + "num_tokens": 247524294.0, + "step": 9651 + }, + { + "epoch": 1.0599604656270591, + "grad_norm": 2.271921396255493, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7214282751083374, + "num_tokens": 247550205.0, + "step": 9652 + }, + { + "epoch": 1.0600702833296727, + "grad_norm": 1.8471969366073608, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7044034004211426, + "num_tokens": 247585575.0, + "step": 9653 + }, + { + "epoch": 1.0601801010322864, + "grad_norm": 2.3726696968078613, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7150383591651917, + "num_tokens": 247608990.0, + "step": 9654 + }, + { + "epoch": 1.0602899187349, + "grad_norm": 2.439136505126953, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7106916308403015, + "num_tokens": 247632073.0, + "step": 9655 + }, + { + "epoch": 1.0603997364375137, + "grad_norm": 2.3291995525360107, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7118266224861145, + "num_tokens": 247657539.0, + "step": 9656 + }, + { + "epoch": 1.0605095541401275, + "grad_norm": 2.139611005783081, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7199177742004395, + "num_tokens": 247684991.0, + "step": 9657 + }, + { + "epoch": 1.060619371842741, + "grad_norm": 2.1042113304138184, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7100021839141846, + "num_tokens": 247711994.0, + "step": 9658 + }, + { + "epoch": 1.0607291895453548, + "grad_norm": 2.734126329421997, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.720180094242096, + "num_tokens": 247731889.0, + "step": 9659 + }, + { + "epoch": 1.0608390072479683, + "grad_norm": 2.183988332748413, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7044199705123901, + "num_tokens": 247760461.0, + "step": 9660 + }, + { + "epoch": 1.060948824950582, + "grad_norm": 2.4239115715026855, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7279692888259888, + "num_tokens": 247781724.0, + "step": 9661 + }, + { + "epoch": 1.0610586426531956, + "grad_norm": 2.045356035232544, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7003598213195801, + "num_tokens": 247811631.0, + "step": 9662 + }, + { + "epoch": 1.0611684603558094, + "grad_norm": 2.1975603103637695, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.73086017370224, + "num_tokens": 247835833.0, + "step": 9663 + }, + { + "epoch": 1.0612782780584231, + "grad_norm": 2.2386221885681152, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7281945943832397, + "num_tokens": 247860226.0, + "step": 9664 + }, + { + "epoch": 1.0613880957610367, + "grad_norm": 1.9878835678100586, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7103522419929504, + "num_tokens": 247892098.0, + "step": 9665 + }, + { + "epoch": 1.0614979134636504, + "grad_norm": 2.483316659927368, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7103886604309082, + "num_tokens": 247915154.0, + "step": 9666 + }, + { + "epoch": 1.061607731166264, + "grad_norm": 2.1180152893066406, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7339997291564941, + "num_tokens": 247942432.0, + "step": 9667 + }, + { + "epoch": 1.0617175488688777, + "grad_norm": 2.2317497730255127, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.711006760597229, + "num_tokens": 247969316.0, + "step": 9668 + }, + { + "epoch": 1.0618273665714912, + "grad_norm": 2.112109899520874, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.726392388343811, + "num_tokens": 247995385.0, + "step": 9669 + }, + { + "epoch": 1.061937184274105, + "grad_norm": 2.332716703414917, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7216970920562744, + "num_tokens": 248018368.0, + "step": 9670 + }, + { + "epoch": 1.0620470019767188, + "grad_norm": 2.2238638401031494, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7385895252227783, + "num_tokens": 248042307.0, + "step": 9671 + }, + { + "epoch": 1.0621568196793323, + "grad_norm": 2.1956446170806885, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7315151691436768, + "num_tokens": 248068707.0, + "step": 9672 + }, + { + "epoch": 1.062266637381946, + "grad_norm": 2.0596508979797363, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.717529296875, + "num_tokens": 248096883.0, + "step": 9673 + }, + { + "epoch": 1.0623764550845596, + "grad_norm": 2.1842806339263916, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6923158168792725, + "num_tokens": 248127682.0, + "step": 9674 + }, + { + "epoch": 1.0624862727871733, + "grad_norm": 2.6465189456939697, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7188183665275574, + "num_tokens": 248148980.0, + "step": 9675 + }, + { + "epoch": 1.0625960904897869, + "grad_norm": 2.524217128753662, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7334656715393066, + "num_tokens": 248170479.0, + "step": 9676 + }, + { + "epoch": 1.0627059081924006, + "grad_norm": 2.3383030891418457, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7138234972953796, + "num_tokens": 248194345.0, + "step": 9677 + }, + { + "epoch": 1.0628157258950144, + "grad_norm": 2.2273178100585938, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7087346911430359, + "num_tokens": 248219540.0, + "step": 9678 + }, + { + "epoch": 1.062925543597628, + "grad_norm": 1.8890734910964966, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.713859498500824, + "num_tokens": 248254416.0, + "step": 9679 + }, + { + "epoch": 1.0630353613002417, + "grad_norm": 2.553248882293701, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7364146113395691, + "num_tokens": 248275852.0, + "step": 9680 + }, + { + "epoch": 1.0631451790028552, + "grad_norm": 2.4812304973602295, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7181886434555054, + "num_tokens": 248298087.0, + "step": 9681 + }, + { + "epoch": 1.063254996705469, + "grad_norm": 2.202153444290161, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6880882978439331, + "num_tokens": 248328114.0, + "step": 9682 + }, + { + "epoch": 1.0633648144080825, + "grad_norm": 2.1210100650787354, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7091199159622192, + "num_tokens": 248357623.0, + "step": 9683 + }, + { + "epoch": 1.0634746321106963, + "grad_norm": 2.0762317180633545, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6975834369659424, + "num_tokens": 248385465.0, + "step": 9684 + }, + { + "epoch": 1.0635844498133098, + "grad_norm": 2.3516108989715576, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7182915806770325, + "num_tokens": 248409653.0, + "step": 9685 + }, + { + "epoch": 1.0636942675159236, + "grad_norm": 2.704216241836548, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7265797257423401, + "num_tokens": 248429745.0, + "step": 9686 + }, + { + "epoch": 1.0638040852185373, + "grad_norm": 2.203331708908081, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6842043995857239, + "num_tokens": 248458002.0, + "step": 9687 + }, + { + "epoch": 1.0639139029211508, + "grad_norm": 2.349440813064575, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7199757695198059, + "num_tokens": 248482638.0, + "step": 9688 + }, + { + "epoch": 1.0640237206237646, + "grad_norm": 2.0875051021575928, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7323418259620667, + "num_tokens": 248511245.0, + "step": 9689 + }, + { + "epoch": 1.0641335383263781, + "grad_norm": 2.5268592834472656, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.740185558795929, + "num_tokens": 248531727.0, + "step": 9690 + }, + { + "epoch": 1.064243356028992, + "grad_norm": 2.362677574157715, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7319220304489136, + "num_tokens": 248553324.0, + "step": 9691 + }, + { + "epoch": 1.0643531737316057, + "grad_norm": 2.6640408039093018, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7311537861824036, + "num_tokens": 248573397.0, + "step": 9692 + }, + { + "epoch": 1.0644629914342192, + "grad_norm": 2.4620137214660645, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7056925892829895, + "num_tokens": 248596459.0, + "step": 9693 + }, + { + "epoch": 1.064572809136833, + "grad_norm": 2.143605947494507, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7205619812011719, + "num_tokens": 248625614.0, + "step": 9694 + }, + { + "epoch": 1.0646826268394465, + "grad_norm": 2.6564736366271973, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7454127073287964, + "num_tokens": 248643729.0, + "step": 9695 + }, + { + "epoch": 1.0647924445420602, + "grad_norm": 2.4342660903930664, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7249500751495361, + "num_tokens": 248664931.0, + "step": 9696 + }, + { + "epoch": 1.0649022622446738, + "grad_norm": 2.1709635257720947, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7214705944061279, + "num_tokens": 248691973.0, + "step": 9697 + }, + { + "epoch": 1.0650120799472875, + "grad_norm": 2.591534376144409, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7180531620979309, + "num_tokens": 248712092.0, + "step": 9698 + }, + { + "epoch": 1.065121897649901, + "grad_norm": 2.579861879348755, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7150155305862427, + "num_tokens": 248732336.0, + "step": 9699 + }, + { + "epoch": 1.0652317153525148, + "grad_norm": 2.3109633922576904, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7310708165168762, + "num_tokens": 248758773.0, + "step": 9700 + }, + { + "epoch": 1.0653415330551286, + "grad_norm": 2.0905821323394775, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7025409936904907, + "num_tokens": 248787790.0, + "step": 9701 + }, + { + "epoch": 1.065451350757742, + "grad_norm": 1.995529294013977, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7348456382751465, + "num_tokens": 248818195.0, + "step": 9702 + }, + { + "epoch": 1.0655611684603559, + "grad_norm": 2.1168127059936523, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7456574440002441, + "num_tokens": 248843950.0, + "step": 9703 + }, + { + "epoch": 1.0656709861629694, + "grad_norm": 1.9771040678024292, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7348005771636963, + "num_tokens": 248874484.0, + "step": 9704 + }, + { + "epoch": 1.0657808038655832, + "grad_norm": 2.53155517578125, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7345497012138367, + "num_tokens": 248894453.0, + "step": 9705 + }, + { + "epoch": 1.0658906215681967, + "grad_norm": 2.157238721847534, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.739399790763855, + "num_tokens": 248921403.0, + "step": 9706 + }, + { + "epoch": 1.0660004392708105, + "grad_norm": 2.756378173828125, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7371005415916443, + "num_tokens": 248939408.0, + "step": 9707 + }, + { + "epoch": 1.0661102569734242, + "grad_norm": 1.9222478866577148, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6893308758735657, + "num_tokens": 248977096.0, + "step": 9708 + }, + { + "epoch": 1.0662200746760377, + "grad_norm": 2.0084216594696045, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7170909643173218, + "num_tokens": 249008373.0, + "step": 9709 + }, + { + "epoch": 1.0663298923786515, + "grad_norm": 2.4118027687072754, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7152215838432312, + "num_tokens": 249030362.0, + "step": 9710 + }, + { + "epoch": 1.066439710081265, + "grad_norm": 2.5842063426971436, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7149710059165955, + "num_tokens": 249052198.0, + "step": 9711 + }, + { + "epoch": 1.0665495277838788, + "grad_norm": 2.0571205615997314, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7290546894073486, + "num_tokens": 249078924.0, + "step": 9712 + }, + { + "epoch": 1.0666593454864923, + "grad_norm": 2.263277769088745, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7280095815658569, + "num_tokens": 249103506.0, + "step": 9713 + }, + { + "epoch": 1.066769163189106, + "grad_norm": 2.4590137004852295, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7207933664321899, + "num_tokens": 249125134.0, + "step": 9714 + }, + { + "epoch": 1.0668789808917198, + "grad_norm": 2.49127459526062, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7009652853012085, + "num_tokens": 249148793.0, + "step": 9715 + }, + { + "epoch": 1.0669887985943334, + "grad_norm": 2.0546391010284424, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7101891040802002, + "num_tokens": 249178349.0, + "step": 9716 + }, + { + "epoch": 1.0670986162969471, + "grad_norm": 2.337242841720581, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7150765657424927, + "num_tokens": 249203234.0, + "step": 9717 + }, + { + "epoch": 1.0672084339995607, + "grad_norm": 2.0081329345703125, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7069728970527649, + "num_tokens": 249233263.0, + "step": 9718 + }, + { + "epoch": 1.0673182517021744, + "grad_norm": 2.3899667263031006, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.725409984588623, + "num_tokens": 249254834.0, + "step": 9719 + }, + { + "epoch": 1.067428069404788, + "grad_norm": 2.2261464595794678, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7032761573791504, + "num_tokens": 249279566.0, + "step": 9720 + }, + { + "epoch": 1.0675378871074017, + "grad_norm": 2.2670280933380127, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7190722227096558, + "num_tokens": 249304334.0, + "step": 9721 + }, + { + "epoch": 1.0676477048100155, + "grad_norm": 2.5323541164398193, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7148088812828064, + "num_tokens": 249326965.0, + "step": 9722 + }, + { + "epoch": 1.067757522512629, + "grad_norm": 2.4757447242736816, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7078176736831665, + "num_tokens": 249349809.0, + "step": 9723 + }, + { + "epoch": 1.0678673402152428, + "grad_norm": 2.3655872344970703, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.711668848991394, + "num_tokens": 249373781.0, + "step": 9724 + }, + { + "epoch": 1.0679771579178563, + "grad_norm": 1.9852200746536255, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7024798393249512, + "num_tokens": 249402701.0, + "step": 9725 + }, + { + "epoch": 1.06808697562047, + "grad_norm": 2.313159704208374, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7302955389022827, + "num_tokens": 249426058.0, + "step": 9726 + }, + { + "epoch": 1.0681967933230836, + "grad_norm": 2.2422895431518555, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7158231735229492, + "num_tokens": 249453246.0, + "step": 9727 + }, + { + "epoch": 1.0683066110256974, + "grad_norm": 2.1811745166778564, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.723547637462616, + "num_tokens": 249479944.0, + "step": 9728 + }, + { + "epoch": 1.068416428728311, + "grad_norm": 2.1162068843841553, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7033438086509705, + "num_tokens": 249507749.0, + "step": 9729 + }, + { + "epoch": 1.0685262464309246, + "grad_norm": 2.4307148456573486, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7285347580909729, + "num_tokens": 249529709.0, + "step": 9730 + }, + { + "epoch": 1.0686360641335384, + "grad_norm": 2.218571186065674, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7298247814178467, + "num_tokens": 249553601.0, + "step": 9731 + }, + { + "epoch": 1.068745881836152, + "grad_norm": 2.46842885017395, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7197513580322266, + "num_tokens": 249576305.0, + "step": 9732 + }, + { + "epoch": 1.0688556995387657, + "grad_norm": 2.3557894229888916, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7028409838676453, + "num_tokens": 249602206.0, + "step": 9733 + }, + { + "epoch": 1.0689655172413792, + "grad_norm": 2.339452028274536, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7277183532714844, + "num_tokens": 249625847.0, + "step": 9734 + }, + { + "epoch": 1.069075334943993, + "grad_norm": 2.3514387607574463, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7236117124557495, + "num_tokens": 249650330.0, + "step": 9735 + }, + { + "epoch": 1.0691851526466065, + "grad_norm": 2.3495287895202637, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7048087120056152, + "num_tokens": 249674191.0, + "step": 9736 + }, + { + "epoch": 1.0692949703492203, + "grad_norm": 2.1456334590911865, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7088366746902466, + "num_tokens": 249702192.0, + "step": 9737 + }, + { + "epoch": 1.069404788051834, + "grad_norm": 2.663459539413452, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7234107851982117, + "num_tokens": 249721049.0, + "step": 9738 + }, + { + "epoch": 1.0695146057544476, + "grad_norm": 2.5053629875183105, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7378854155540466, + "num_tokens": 249742193.0, + "step": 9739 + }, + { + "epoch": 1.0696244234570613, + "grad_norm": 2.298919677734375, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7235621809959412, + "num_tokens": 249769920.0, + "step": 9740 + }, + { + "epoch": 1.0697342411596749, + "grad_norm": 2.322310209274292, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7190253138542175, + "num_tokens": 249796212.0, + "step": 9741 + }, + { + "epoch": 1.0698440588622886, + "grad_norm": 2.1260368824005127, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7154114246368408, + "num_tokens": 249824294.0, + "step": 9742 + }, + { + "epoch": 1.0699538765649024, + "grad_norm": 2.0711381435394287, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7231020927429199, + "num_tokens": 249852602.0, + "step": 9743 + }, + { + "epoch": 1.070063694267516, + "grad_norm": 1.9638965129852295, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7077125906944275, + "num_tokens": 249882221.0, + "step": 9744 + }, + { + "epoch": 1.0701735119701297, + "grad_norm": 2.35406756401062, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7112996578216553, + "num_tokens": 249903866.0, + "step": 9745 + }, + { + "epoch": 1.0702833296727432, + "grad_norm": 2.2331173419952393, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7013782858848572, + "num_tokens": 249929981.0, + "step": 9746 + }, + { + "epoch": 1.070393147375357, + "grad_norm": 2.2767457962036133, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7212331295013428, + "num_tokens": 249953541.0, + "step": 9747 + }, + { + "epoch": 1.0705029650779705, + "grad_norm": 2.1725168228149414, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7333132028579712, + "num_tokens": 249978489.0, + "step": 9748 + }, + { + "epoch": 1.0706127827805842, + "grad_norm": 2.3992602825164795, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7356741428375244, + "num_tokens": 250001338.0, + "step": 9749 + }, + { + "epoch": 1.0707226004831978, + "grad_norm": 2.3957555294036865, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7384297251701355, + "num_tokens": 250021759.0, + "step": 9750 + }, + { + "epoch": 1.0708324181858115, + "grad_norm": 2.261434555053711, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7277078628540039, + "num_tokens": 250050206.0, + "step": 9751 + }, + { + "epoch": 1.0709422358884253, + "grad_norm": 2.6924216747283936, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7070520520210266, + "num_tokens": 250069051.0, + "step": 9752 + }, + { + "epoch": 1.0710520535910388, + "grad_norm": 1.9172229766845703, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7060896754264832, + "num_tokens": 250105516.0, + "step": 9753 + }, + { + "epoch": 1.0711618712936526, + "grad_norm": 2.3312361240386963, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7098150253295898, + "num_tokens": 250128592.0, + "step": 9754 + }, + { + "epoch": 1.0712716889962661, + "grad_norm": 2.261307954788208, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7057673931121826, + "num_tokens": 250154212.0, + "step": 9755 + }, + { + "epoch": 1.0713815066988799, + "grad_norm": 2.4643502235412598, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7393155097961426, + "num_tokens": 250174064.0, + "step": 9756 + }, + { + "epoch": 1.0714913244014934, + "grad_norm": 2.083616256713867, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7060743570327759, + "num_tokens": 250204127.0, + "step": 9757 + }, + { + "epoch": 1.0716011421041072, + "grad_norm": 2.4727909564971924, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7192025184631348, + "num_tokens": 250224698.0, + "step": 9758 + }, + { + "epoch": 1.071710959806721, + "grad_norm": 2.310124397277832, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7240900993347168, + "num_tokens": 250247537.0, + "step": 9759 + }, + { + "epoch": 1.0718207775093345, + "grad_norm": 1.9901320934295654, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.712189793586731, + "num_tokens": 250277629.0, + "step": 9760 + }, + { + "epoch": 1.0719305952119482, + "grad_norm": 2.238154649734497, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7381026744842529, + "num_tokens": 250300938.0, + "step": 9761 + }, + { + "epoch": 1.0720404129145618, + "grad_norm": 2.4244937896728516, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7187501192092896, + "num_tokens": 250324696.0, + "step": 9762 + }, + { + "epoch": 1.0721502306171755, + "grad_norm": 2.191776990890503, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7018218040466309, + "num_tokens": 250350653.0, + "step": 9763 + }, + { + "epoch": 1.072260048319789, + "grad_norm": 2.4364449977874756, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7366490364074707, + "num_tokens": 250370526.0, + "step": 9764 + }, + { + "epoch": 1.0723698660224028, + "grad_norm": 2.4273104667663574, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7251882553100586, + "num_tokens": 250391693.0, + "step": 9765 + }, + { + "epoch": 1.0724796837250166, + "grad_norm": 1.9627974033355713, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.716578483581543, + "num_tokens": 250424197.0, + "step": 9766 + }, + { + "epoch": 1.07258950142763, + "grad_norm": 2.186657667160034, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7033225297927856, + "num_tokens": 250450611.0, + "step": 9767 + }, + { + "epoch": 1.0726993191302439, + "grad_norm": 2.108530044555664, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.70135498046875, + "num_tokens": 250478654.0, + "step": 9768 + }, + { + "epoch": 1.0728091368328574, + "grad_norm": 2.124952793121338, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7076839208602905, + "num_tokens": 250508206.0, + "step": 9769 + }, + { + "epoch": 1.0729189545354711, + "grad_norm": 2.412945508956909, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7407904863357544, + "num_tokens": 250527972.0, + "step": 9770 + }, + { + "epoch": 1.0730287722380847, + "grad_norm": 2.086198091506958, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7129664421081543, + "num_tokens": 250556853.0, + "step": 9771 + }, + { + "epoch": 1.0731385899406984, + "grad_norm": 2.4234347343444824, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7004097700119019, + "num_tokens": 250580411.0, + "step": 9772 + }, + { + "epoch": 1.0732484076433122, + "grad_norm": 2.1947484016418457, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7353053092956543, + "num_tokens": 250605358.0, + "step": 9773 + }, + { + "epoch": 1.0733582253459257, + "grad_norm": 2.285454273223877, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7199835181236267, + "num_tokens": 250630229.0, + "step": 9774 + }, + { + "epoch": 1.0734680430485395, + "grad_norm": 2.182873249053955, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7294596433639526, + "num_tokens": 250655220.0, + "step": 9775 + }, + { + "epoch": 1.073577860751153, + "grad_norm": 2.191826105117798, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.723368763923645, + "num_tokens": 250681282.0, + "step": 9776 + }, + { + "epoch": 1.0736876784537668, + "grad_norm": 2.3160765171051025, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7363902926445007, + "num_tokens": 250704713.0, + "step": 9777 + }, + { + "epoch": 1.0737974961563803, + "grad_norm": 2.208716869354248, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7450348138809204, + "num_tokens": 250730075.0, + "step": 9778 + }, + { + "epoch": 1.073907313858994, + "grad_norm": 2.428978443145752, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7558653354644775, + "num_tokens": 250751412.0, + "step": 9779 + }, + { + "epoch": 1.0740171315616078, + "grad_norm": 1.9078413248062134, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.716439962387085, + "num_tokens": 250784420.0, + "step": 9780 + }, + { + "epoch": 1.0741269492642214, + "grad_norm": 2.0545578002929688, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7139196991920471, + "num_tokens": 250815307.0, + "step": 9781 + }, + { + "epoch": 1.0742367669668351, + "grad_norm": 2.1196813583374023, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7278797030448914, + "num_tokens": 250842740.0, + "step": 9782 + }, + { + "epoch": 1.0743465846694487, + "grad_norm": 2.6135647296905518, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.722859263420105, + "num_tokens": 250862330.0, + "step": 9783 + }, + { + "epoch": 1.0744564023720624, + "grad_norm": 2.1960041522979736, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6876740455627441, + "num_tokens": 250891818.0, + "step": 9784 + }, + { + "epoch": 1.074566220074676, + "grad_norm": 2.0963551998138428, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7199349403381348, + "num_tokens": 250918602.0, + "step": 9785 + }, + { + "epoch": 1.0746760377772897, + "grad_norm": 2.600428342819214, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7064903974533081, + "num_tokens": 250939475.0, + "step": 9786 + }, + { + "epoch": 1.0747858554799035, + "grad_norm": 2.2663369178771973, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7163137197494507, + "num_tokens": 250965687.0, + "step": 9787 + }, + { + "epoch": 1.074895673182517, + "grad_norm": 2.3272604942321777, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7039666175842285, + "num_tokens": 250992929.0, + "step": 9788 + }, + { + "epoch": 1.0750054908851308, + "grad_norm": 2.148566722869873, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7095291614532471, + "num_tokens": 251020789.0, + "step": 9789 + }, + { + "epoch": 1.0751153085877443, + "grad_norm": 2.4368371963500977, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7156237959861755, + "num_tokens": 251042839.0, + "step": 9790 + }, + { + "epoch": 1.075225126290358, + "grad_norm": 2.0223398208618164, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7146940231323242, + "num_tokens": 251074720.0, + "step": 9791 + }, + { + "epoch": 1.0753349439929716, + "grad_norm": 2.093524694442749, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7091159820556641, + "num_tokens": 251103076.0, + "step": 9792 + }, + { + "epoch": 1.0754447616955853, + "grad_norm": 2.0148086547851562, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7132428884506226, + "num_tokens": 251132583.0, + "step": 9793 + }, + { + "epoch": 1.075554579398199, + "grad_norm": 2.1320064067840576, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7093064785003662, + "num_tokens": 251158625.0, + "step": 9794 + }, + { + "epoch": 1.0756643971008126, + "grad_norm": 2.368863582611084, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7101595401763916, + "num_tokens": 251182362.0, + "step": 9795 + }, + { + "epoch": 1.0757742148034264, + "grad_norm": 2.3355867862701416, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7072031497955322, + "num_tokens": 251208331.0, + "step": 9796 + }, + { + "epoch": 1.07588403250604, + "grad_norm": 2.4435787200927734, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7178226709365845, + "num_tokens": 251230014.0, + "step": 9797 + }, + { + "epoch": 1.0759938502086537, + "grad_norm": 2.1173315048217773, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7312428951263428, + "num_tokens": 251258105.0, + "step": 9798 + }, + { + "epoch": 1.0761036679112672, + "grad_norm": 2.363482713699341, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.696151077747345, + "num_tokens": 251285010.0, + "step": 9799 + }, + { + "epoch": 1.076213485613881, + "grad_norm": 2.5862464904785156, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7351768016815186, + "num_tokens": 251303504.0, + "step": 9800 + }, + { + "epoch": 1.0763233033164945, + "grad_norm": 2.201735496520996, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7098706960678101, + "num_tokens": 251330064.0, + "step": 9801 + }, + { + "epoch": 1.0764331210191083, + "grad_norm": 2.2701237201690674, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7192062139511108, + "num_tokens": 251355496.0, + "step": 9802 + }, + { + "epoch": 1.076542938721722, + "grad_norm": 2.4988279342651367, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7386850714683533, + "num_tokens": 251376071.0, + "step": 9803 + }, + { + "epoch": 1.0766527564243356, + "grad_norm": 2.0808351039886475, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7076754570007324, + "num_tokens": 251406264.0, + "step": 9804 + }, + { + "epoch": 1.0767625741269493, + "grad_norm": 2.4734694957733154, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7450634241104126, + "num_tokens": 251426304.0, + "step": 9805 + }, + { + "epoch": 1.0768723918295628, + "grad_norm": 2.444659948348999, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7215391993522644, + "num_tokens": 251449863.0, + "step": 9806 + }, + { + "epoch": 1.0769822095321766, + "grad_norm": 2.586829662322998, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7175617814064026, + "num_tokens": 251471037.0, + "step": 9807 + }, + { + "epoch": 1.0770920272347904, + "grad_norm": 2.0317132472991943, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7411260604858398, + "num_tokens": 251498805.0, + "step": 9808 + }, + { + "epoch": 1.077201844937404, + "grad_norm": 1.9670584201812744, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7253141403198242, + "num_tokens": 251530243.0, + "step": 9809 + }, + { + "epoch": 1.0773116626400177, + "grad_norm": 1.9649947881698608, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7387105822563171, + "num_tokens": 251558789.0, + "step": 9810 + }, + { + "epoch": 1.0774214803426312, + "grad_norm": 1.845060110092163, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7218722105026245, + "num_tokens": 251592859.0, + "step": 9811 + }, + { + "epoch": 1.077531298045245, + "grad_norm": 2.152318000793457, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7243431806564331, + "num_tokens": 251618403.0, + "step": 9812 + }, + { + "epoch": 1.0776411157478585, + "grad_norm": 2.1517868041992188, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7316089868545532, + "num_tokens": 251645435.0, + "step": 9813 + }, + { + "epoch": 1.0777509334504722, + "grad_norm": 2.121436357498169, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.709506630897522, + "num_tokens": 251674308.0, + "step": 9814 + }, + { + "epoch": 1.0778607511530858, + "grad_norm": 2.139580011367798, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7271044254302979, + "num_tokens": 251699652.0, + "step": 9815 + }, + { + "epoch": 1.0779705688556995, + "grad_norm": 2.250633955001831, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7162966132164001, + "num_tokens": 251726053.0, + "step": 9816 + }, + { + "epoch": 1.0780803865583133, + "grad_norm": 2.3651773929595947, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7122858762741089, + "num_tokens": 251749942.0, + "step": 9817 + }, + { + "epoch": 1.0781902042609268, + "grad_norm": 2.155219316482544, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7103235721588135, + "num_tokens": 251776709.0, + "step": 9818 + }, + { + "epoch": 1.0783000219635406, + "grad_norm": 2.3116672039031982, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7326339483261108, + "num_tokens": 251800709.0, + "step": 9819 + }, + { + "epoch": 1.0784098396661541, + "grad_norm": 2.1391875743865967, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7160735130310059, + "num_tokens": 251827100.0, + "step": 9820 + }, + { + "epoch": 1.0785196573687679, + "grad_norm": 2.3629279136657715, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7210201025009155, + "num_tokens": 251851268.0, + "step": 9821 + }, + { + "epoch": 1.0786294750713814, + "grad_norm": 2.170844316482544, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.6975930333137512, + "num_tokens": 251878928.0, + "step": 9822 + }, + { + "epoch": 1.0787392927739952, + "grad_norm": 2.1731667518615723, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6919823288917542, + "num_tokens": 251908107.0, + "step": 9823 + }, + { + "epoch": 1.078849110476609, + "grad_norm": 1.881580114364624, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6889559626579285, + "num_tokens": 251942620.0, + "step": 9824 + }, + { + "epoch": 1.0789589281792225, + "grad_norm": 2.2916293144226074, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.71488356590271, + "num_tokens": 251965548.0, + "step": 9825 + }, + { + "epoch": 1.0790687458818362, + "grad_norm": 2.420468330383301, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7302969694137573, + "num_tokens": 251987626.0, + "step": 9826 + }, + { + "epoch": 1.0791785635844497, + "grad_norm": 2.6303176879882812, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.729093074798584, + "num_tokens": 252008187.0, + "step": 9827 + }, + { + "epoch": 1.0792883812870635, + "grad_norm": 1.9830060005187988, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7108479738235474, + "num_tokens": 252040347.0, + "step": 9828 + }, + { + "epoch": 1.079398198989677, + "grad_norm": 2.3130671977996826, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7381678819656372, + "num_tokens": 252066093.0, + "step": 9829 + }, + { + "epoch": 1.0795080166922908, + "grad_norm": 2.0821144580841064, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6989973783493042, + "num_tokens": 252095533.0, + "step": 9830 + }, + { + "epoch": 1.0796178343949046, + "grad_norm": 2.5542397499084473, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7229136228561401, + "num_tokens": 252116906.0, + "step": 9831 + }, + { + "epoch": 1.079727652097518, + "grad_norm": 2.4146010875701904, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7271912097930908, + "num_tokens": 252139205.0, + "step": 9832 + }, + { + "epoch": 1.0798374698001318, + "grad_norm": 2.387923240661621, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7265549898147583, + "num_tokens": 252164220.0, + "step": 9833 + }, + { + "epoch": 1.0799472875027454, + "grad_norm": 2.1295902729034424, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7174298763275146, + "num_tokens": 252192334.0, + "step": 9834 + }, + { + "epoch": 1.0800571052053591, + "grad_norm": 2.2153618335723877, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7248363494873047, + "num_tokens": 252219469.0, + "step": 9835 + }, + { + "epoch": 1.0801669229079727, + "grad_norm": 2.158003568649292, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7102320194244385, + "num_tokens": 252248341.0, + "step": 9836 + }, + { + "epoch": 1.0802767406105864, + "grad_norm": 2.4268152713775635, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7350946664810181, + "num_tokens": 252268568.0, + "step": 9837 + }, + { + "epoch": 1.0803865583132002, + "grad_norm": 2.3580586910247803, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7273635268211365, + "num_tokens": 252292278.0, + "step": 9838 + }, + { + "epoch": 1.0804963760158137, + "grad_norm": 2.2093560695648193, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6988348364830017, + "num_tokens": 252322463.0, + "step": 9839 + }, + { + "epoch": 1.0806061937184275, + "grad_norm": 2.1116092205047607, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7116153240203857, + "num_tokens": 252351445.0, + "step": 9840 + }, + { + "epoch": 1.080716011421041, + "grad_norm": 2.188462018966675, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7436813116073608, + "num_tokens": 252375912.0, + "step": 9841 + }, + { + "epoch": 1.0808258291236548, + "grad_norm": 2.2430214881896973, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7305201292037964, + "num_tokens": 252402106.0, + "step": 9842 + }, + { + "epoch": 1.0809356468262683, + "grad_norm": 2.0972959995269775, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.729629635810852, + "num_tokens": 252430729.0, + "step": 9843 + }, + { + "epoch": 1.081045464528882, + "grad_norm": 2.366926908493042, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7134321928024292, + "num_tokens": 252454175.0, + "step": 9844 + }, + { + "epoch": 1.0811552822314958, + "grad_norm": 2.2701544761657715, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7238980531692505, + "num_tokens": 252480188.0, + "step": 9845 + }, + { + "epoch": 1.0812650999341094, + "grad_norm": 2.5863895416259766, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7392947673797607, + "num_tokens": 252500448.0, + "step": 9846 + }, + { + "epoch": 1.081374917636723, + "grad_norm": 2.2322046756744385, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7317602634429932, + "num_tokens": 252525679.0, + "step": 9847 + }, + { + "epoch": 1.0814847353393366, + "grad_norm": 2.3067848682403564, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7187489867210388, + "num_tokens": 252551284.0, + "step": 9848 + }, + { + "epoch": 1.0815945530419504, + "grad_norm": 2.3849904537200928, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7139871716499329, + "num_tokens": 252574782.0, + "step": 9849 + }, + { + "epoch": 1.081704370744564, + "grad_norm": 2.0239334106445312, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7031983137130737, + "num_tokens": 252605633.0, + "step": 9850 + }, + { + "epoch": 1.0818141884471777, + "grad_norm": 2.4681143760681152, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.733653724193573, + "num_tokens": 252626616.0, + "step": 9851 + }, + { + "epoch": 1.0819240061497912, + "grad_norm": 2.030256509780884, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7050192952156067, + "num_tokens": 252656796.0, + "step": 9852 + }, + { + "epoch": 1.082033823852405, + "grad_norm": 2.235750198364258, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7455475926399231, + "num_tokens": 252685364.0, + "step": 9853 + }, + { + "epoch": 1.0821436415550187, + "grad_norm": 2.8422093391418457, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7255336046218872, + "num_tokens": 252703353.0, + "step": 9854 + }, + { + "epoch": 1.0822534592576323, + "grad_norm": 2.1638829708099365, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7168540358543396, + "num_tokens": 252730532.0, + "step": 9855 + }, + { + "epoch": 1.082363276960246, + "grad_norm": 1.928879737854004, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7121434211730957, + "num_tokens": 252762730.0, + "step": 9856 + }, + { + "epoch": 1.0824730946628596, + "grad_norm": 2.214399576187134, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7186942100524902, + "num_tokens": 252789103.0, + "step": 9857 + }, + { + "epoch": 1.0825829123654733, + "grad_norm": 2.062044620513916, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6909269690513611, + "num_tokens": 252821825.0, + "step": 9858 + }, + { + "epoch": 1.082692730068087, + "grad_norm": 2.462721109390259, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7140754461288452, + "num_tokens": 252843298.0, + "step": 9859 + }, + { + "epoch": 1.0828025477707006, + "grad_norm": 1.9580785036087036, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7376388311386108, + "num_tokens": 252873978.0, + "step": 9860 + }, + { + "epoch": 1.0829123654733144, + "grad_norm": 2.371605396270752, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.704967200756073, + "num_tokens": 252899870.0, + "step": 9861 + }, + { + "epoch": 1.083022183175928, + "grad_norm": 2.079805612564087, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7198693752288818, + "num_tokens": 252926988.0, + "step": 9862 + }, + { + "epoch": 1.0831320008785417, + "grad_norm": 2.4319815635681152, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7333567142486572, + "num_tokens": 252950652.0, + "step": 9863 + }, + { + "epoch": 1.0832418185811552, + "grad_norm": 2.1799447536468506, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7203861474990845, + "num_tokens": 252977361.0, + "step": 9864 + }, + { + "epoch": 1.083351636283769, + "grad_norm": 2.580641746520996, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7381290197372437, + "num_tokens": 252999541.0, + "step": 9865 + }, + { + "epoch": 1.0834614539863825, + "grad_norm": 2.4446864128112793, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7576378583908081, + "num_tokens": 253019067.0, + "step": 9866 + }, + { + "epoch": 1.0835712716889963, + "grad_norm": 2.238830089569092, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7354077100753784, + "num_tokens": 253042941.0, + "step": 9867 + }, + { + "epoch": 1.08368108939161, + "grad_norm": 2.5155506134033203, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.726089596748352, + "num_tokens": 253063179.0, + "step": 9868 + }, + { + "epoch": 1.0837909070942235, + "grad_norm": 2.178333282470703, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7286157608032227, + "num_tokens": 253091448.0, + "step": 9869 + }, + { + "epoch": 1.0839007247968373, + "grad_norm": 2.222780466079712, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.7070571184158325, + "num_tokens": 253116741.0, + "step": 9870 + }, + { + "epoch": 1.0840105424994508, + "grad_norm": 1.9835164546966553, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7419198751449585, + "num_tokens": 253143992.0, + "step": 9871 + }, + { + "epoch": 1.0841203602020646, + "grad_norm": 2.607391834259033, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7041469216346741, + "num_tokens": 253164061.0, + "step": 9872 + }, + { + "epoch": 1.0842301779046783, + "grad_norm": 2.3148045539855957, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6863701343536377, + "num_tokens": 253190333.0, + "step": 9873 + }, + { + "epoch": 1.0843399956072919, + "grad_norm": 2.259334087371826, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7136535048484802, + "num_tokens": 253217174.0, + "step": 9874 + }, + { + "epoch": 1.0844498133099056, + "grad_norm": 2.2355737686157227, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6996397376060486, + "num_tokens": 253242903.0, + "step": 9875 + }, + { + "epoch": 1.0845596310125192, + "grad_norm": 2.2689294815063477, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7039152383804321, + "num_tokens": 253269319.0, + "step": 9876 + }, + { + "epoch": 1.084669448715133, + "grad_norm": 2.0391833782196045, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.699289083480835, + "num_tokens": 253301130.0, + "step": 9877 + }, + { + "epoch": 1.0847792664177465, + "grad_norm": 2.2091567516326904, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6928800344467163, + "num_tokens": 253329031.0, + "step": 9878 + }, + { + "epoch": 1.0848890841203602, + "grad_norm": 2.3029754161834717, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7375937700271606, + "num_tokens": 253353272.0, + "step": 9879 + }, + { + "epoch": 1.0849989018229738, + "grad_norm": 2.072885513305664, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7294505834579468, + "num_tokens": 253381164.0, + "step": 9880 + }, + { + "epoch": 1.0851087195255875, + "grad_norm": 2.429145574569702, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7154339551925659, + "num_tokens": 253403211.0, + "step": 9881 + }, + { + "epoch": 1.0852185372282013, + "grad_norm": 1.8760430812835693, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7038291692733765, + "num_tokens": 253440884.0, + "step": 9882 + }, + { + "epoch": 1.0853283549308148, + "grad_norm": 2.5780115127563477, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.721177339553833, + "num_tokens": 253462935.0, + "step": 9883 + }, + { + "epoch": 1.0854381726334286, + "grad_norm": 2.4157967567443848, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7190952897071838, + "num_tokens": 253486951.0, + "step": 9884 + }, + { + "epoch": 1.085547990336042, + "grad_norm": 2.1038265228271484, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.6987220048904419, + "num_tokens": 253516810.0, + "step": 9885 + }, + { + "epoch": 1.0856578080386559, + "grad_norm": 2.213380813598633, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7141079902648926, + "num_tokens": 253543490.0, + "step": 9886 + }, + { + "epoch": 1.0857676257412694, + "grad_norm": 2.346605062484741, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7435290813446045, + "num_tokens": 253566285.0, + "step": 9887 + }, + { + "epoch": 1.0858774434438832, + "grad_norm": 2.2911276817321777, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7108656167984009, + "num_tokens": 253591125.0, + "step": 9888 + }, + { + "epoch": 1.085987261146497, + "grad_norm": 2.473022699356079, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7420382499694824, + "num_tokens": 253613463.0, + "step": 9889 + }, + { + "epoch": 1.0860970788491104, + "grad_norm": 2.219369649887085, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7278872728347778, + "num_tokens": 253638562.0, + "step": 9890 + }, + { + "epoch": 1.0862068965517242, + "grad_norm": 2.3186891078948975, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7014854550361633, + "num_tokens": 253661996.0, + "step": 9891 + }, + { + "epoch": 1.0863167142543377, + "grad_norm": 2.0589380264282227, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6991106271743774, + "num_tokens": 253692427.0, + "step": 9892 + }, + { + "epoch": 1.0864265319569515, + "grad_norm": 2.1622118949890137, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7077520489692688, + "num_tokens": 253719795.0, + "step": 9893 + }, + { + "epoch": 1.086536349659565, + "grad_norm": 2.2862324714660645, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7287104725837708, + "num_tokens": 253743962.0, + "step": 9894 + }, + { + "epoch": 1.0866461673621788, + "grad_norm": 3.0415005683898926, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7416795492172241, + "num_tokens": 253760258.0, + "step": 9895 + }, + { + "epoch": 1.0867559850647925, + "grad_norm": 2.1599254608154297, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7266412377357483, + "num_tokens": 253786888.0, + "step": 9896 + }, + { + "epoch": 1.086865802767406, + "grad_norm": 2.1377594470977783, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.6979697346687317, + "num_tokens": 253813520.0, + "step": 9897 + }, + { + "epoch": 1.0869756204700198, + "grad_norm": 2.1036574840545654, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.723793625831604, + "num_tokens": 253839191.0, + "step": 9898 + }, + { + "epoch": 1.0870854381726334, + "grad_norm": 2.056535005569458, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7065752148628235, + "num_tokens": 253870779.0, + "step": 9899 + }, + { + "epoch": 1.0871952558752471, + "grad_norm": 2.443824052810669, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7355006337165833, + "num_tokens": 253892626.0, + "step": 9900 + }, + { + "epoch": 1.0873050735778607, + "grad_norm": 2.1717042922973633, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.711768388748169, + "num_tokens": 253921791.0, + "step": 9901 + }, + { + "epoch": 1.0874148912804744, + "grad_norm": 2.2792513370513916, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7209122180938721, + "num_tokens": 253947278.0, + "step": 9902 + }, + { + "epoch": 1.0875247089830882, + "grad_norm": 2.2445197105407715, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7291444540023804, + "num_tokens": 253973108.0, + "step": 9903 + }, + { + "epoch": 1.0876345266857017, + "grad_norm": 2.1423957347869873, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7176468372344971, + "num_tokens": 254002674.0, + "step": 9904 + }, + { + "epoch": 1.0877443443883155, + "grad_norm": 2.1531436443328857, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.71770179271698, + "num_tokens": 254029900.0, + "step": 9905 + }, + { + "epoch": 1.087854162090929, + "grad_norm": 2.250783681869507, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7274969816207886, + "num_tokens": 254055741.0, + "step": 9906 + }, + { + "epoch": 1.0879639797935428, + "grad_norm": 2.2979719638824463, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7246159315109253, + "num_tokens": 254080062.0, + "step": 9907 + }, + { + "epoch": 1.0880737974961563, + "grad_norm": 1.951236605644226, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.72664475440979, + "num_tokens": 254112795.0, + "step": 9908 + }, + { + "epoch": 1.08818361519877, + "grad_norm": 2.0709919929504395, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7361462116241455, + "num_tokens": 254141454.0, + "step": 9909 + }, + { + "epoch": 1.0882934329013838, + "grad_norm": 2.23030948638916, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7040804624557495, + "num_tokens": 254168504.0, + "step": 9910 + }, + { + "epoch": 1.0884032506039973, + "grad_norm": 2.0887057781219482, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7300211191177368, + "num_tokens": 254196407.0, + "step": 9911 + }, + { + "epoch": 1.088513068306611, + "grad_norm": 2.409175395965576, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7304185032844543, + "num_tokens": 254217960.0, + "step": 9912 + }, + { + "epoch": 1.0886228860092246, + "grad_norm": 2.3163998126983643, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7222346067428589, + "num_tokens": 254240542.0, + "step": 9913 + }, + { + "epoch": 1.0887327037118384, + "grad_norm": 2.0935604572296143, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7169179320335388, + "num_tokens": 254269809.0, + "step": 9914 + }, + { + "epoch": 1.088842521414452, + "grad_norm": 2.2237226963043213, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7158105373382568, + "num_tokens": 254296417.0, + "step": 9915 + }, + { + "epoch": 1.0889523391170657, + "grad_norm": 2.1405844688415527, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7115461826324463, + "num_tokens": 254327779.0, + "step": 9916 + }, + { + "epoch": 1.0890621568196792, + "grad_norm": 2.197996139526367, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7224196195602417, + "num_tokens": 254353548.0, + "step": 9917 + }, + { + "epoch": 1.089171974522293, + "grad_norm": 2.552924156188965, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7201245427131653, + "num_tokens": 254374040.0, + "step": 9918 + }, + { + "epoch": 1.0892817922249067, + "grad_norm": 2.304905652999878, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.699174165725708, + "num_tokens": 254400302.0, + "step": 9919 + }, + { + "epoch": 1.0893916099275203, + "grad_norm": 2.2500381469726562, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7091040015220642, + "num_tokens": 254425964.0, + "step": 9920 + }, + { + "epoch": 1.089501427630134, + "grad_norm": 2.172032117843628, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7104005813598633, + "num_tokens": 254452818.0, + "step": 9921 + }, + { + "epoch": 1.0896112453327476, + "grad_norm": 2.023327112197876, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.702544093132019, + "num_tokens": 254485387.0, + "step": 9922 + }, + { + "epoch": 1.0897210630353613, + "grad_norm": 2.28515887260437, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7405792474746704, + "num_tokens": 254508869.0, + "step": 9923 + }, + { + "epoch": 1.089830880737975, + "grad_norm": 2.188627004623413, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7191171646118164, + "num_tokens": 254534499.0, + "step": 9924 + }, + { + "epoch": 1.0899406984405886, + "grad_norm": 2.442521810531616, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7121402621269226, + "num_tokens": 254558264.0, + "step": 9925 + }, + { + "epoch": 1.0900505161432024, + "grad_norm": 2.104402542114258, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7103903293609619, + "num_tokens": 254586465.0, + "step": 9926 + }, + { + "epoch": 1.090160333845816, + "grad_norm": 2.131662130355835, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6782227158546448, + "num_tokens": 254617924.0, + "step": 9927 + }, + { + "epoch": 1.0902701515484297, + "grad_norm": 2.1781294345855713, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6990727186203003, + "num_tokens": 254644995.0, + "step": 9928 + }, + { + "epoch": 1.0903799692510432, + "grad_norm": 2.4311349391937256, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7353459596633911, + "num_tokens": 254666662.0, + "step": 9929 + }, + { + "epoch": 1.090489786953657, + "grad_norm": 2.203563928604126, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7151850461959839, + "num_tokens": 254694820.0, + "step": 9930 + }, + { + "epoch": 1.0905996046562705, + "grad_norm": 2.0749614238739014, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7205294966697693, + "num_tokens": 254721255.0, + "step": 9931 + }, + { + "epoch": 1.0907094223588842, + "grad_norm": 2.3041980266571045, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7029836773872375, + "num_tokens": 254748219.0, + "step": 9932 + }, + { + "epoch": 1.090819240061498, + "grad_norm": 2.188262939453125, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7094428539276123, + "num_tokens": 254776423.0, + "step": 9933 + }, + { + "epoch": 1.0909290577641115, + "grad_norm": 2.1091389656066895, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7396451234817505, + "num_tokens": 254801969.0, + "step": 9934 + }, + { + "epoch": 1.0910388754667253, + "grad_norm": 2.037353992462158, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.6946814060211182, + "num_tokens": 254832335.0, + "step": 9935 + }, + { + "epoch": 1.0911486931693388, + "grad_norm": 2.3398983478546143, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7060530185699463, + "num_tokens": 254855868.0, + "step": 9936 + }, + { + "epoch": 1.0912585108719526, + "grad_norm": 2.2464821338653564, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.728499174118042, + "num_tokens": 254882204.0, + "step": 9937 + }, + { + "epoch": 1.0913683285745663, + "grad_norm": 2.2779154777526855, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.720439076423645, + "num_tokens": 254907617.0, + "step": 9938 + }, + { + "epoch": 1.0914781462771799, + "grad_norm": 2.2933623790740967, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7136920690536499, + "num_tokens": 254933174.0, + "step": 9939 + }, + { + "epoch": 1.0915879639797936, + "grad_norm": 2.3176825046539307, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6918723583221436, + "num_tokens": 254958320.0, + "step": 9940 + }, + { + "epoch": 1.0916977816824072, + "grad_norm": 2.5527191162109375, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.716052770614624, + "num_tokens": 254980059.0, + "step": 9941 + }, + { + "epoch": 1.091807599385021, + "grad_norm": 2.078903913497925, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7054221630096436, + "num_tokens": 255008138.0, + "step": 9942 + }, + { + "epoch": 1.0919174170876345, + "grad_norm": 2.2890067100524902, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7210807800292969, + "num_tokens": 255033147.0, + "step": 9943 + }, + { + "epoch": 1.0920272347902482, + "grad_norm": 2.2601873874664307, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6956683397293091, + "num_tokens": 255060943.0, + "step": 9944 + }, + { + "epoch": 1.0921370524928617, + "grad_norm": 2.181908369064331, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7150561809539795, + "num_tokens": 255086058.0, + "step": 9945 + }, + { + "epoch": 1.0922468701954755, + "grad_norm": 2.2010319232940674, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7230669260025024, + "num_tokens": 255112607.0, + "step": 9946 + }, + { + "epoch": 1.0923566878980893, + "grad_norm": 1.9634053707122803, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7050553560256958, + "num_tokens": 255145097.0, + "step": 9947 + }, + { + "epoch": 1.0924665056007028, + "grad_norm": 2.2036712169647217, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7149715423583984, + "num_tokens": 255172482.0, + "step": 9948 + }, + { + "epoch": 1.0925763233033166, + "grad_norm": 2.226891040802002, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7320868968963623, + "num_tokens": 255197941.0, + "step": 9949 + }, + { + "epoch": 1.09268614100593, + "grad_norm": 2.4910342693328857, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7216867208480835, + "num_tokens": 255218807.0, + "step": 9950 + }, + { + "epoch": 1.0927959587085438, + "grad_norm": 2.4497644901275635, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.724823534488678, + "num_tokens": 255239886.0, + "step": 9951 + }, + { + "epoch": 1.0929057764111574, + "grad_norm": 2.2670416831970215, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7248808145523071, + "num_tokens": 255263651.0, + "step": 9952 + }, + { + "epoch": 1.0930155941137711, + "grad_norm": 2.3582711219787598, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7192937731742859, + "num_tokens": 255287153.0, + "step": 9953 + }, + { + "epoch": 1.093125411816385, + "grad_norm": 2.3365345001220703, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7122129797935486, + "num_tokens": 255311052.0, + "step": 9954 + }, + { + "epoch": 1.0932352295189984, + "grad_norm": 2.7217373847961426, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.744701623916626, + "num_tokens": 255328495.0, + "step": 9955 + }, + { + "epoch": 1.0933450472216122, + "grad_norm": 2.444404125213623, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7135353684425354, + "num_tokens": 255350509.0, + "step": 9956 + }, + { + "epoch": 1.0934548649242257, + "grad_norm": 2.163635492324829, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7078279256820679, + "num_tokens": 255379144.0, + "step": 9957 + }, + { + "epoch": 1.0935646826268395, + "grad_norm": 2.1440248489379883, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7203727960586548, + "num_tokens": 255406407.0, + "step": 9958 + }, + { + "epoch": 1.093674500329453, + "grad_norm": 2.0368220806121826, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7205662131309509, + "num_tokens": 255437903.0, + "step": 9959 + }, + { + "epoch": 1.0937843180320668, + "grad_norm": 2.4719595909118652, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7293723821640015, + "num_tokens": 255460347.0, + "step": 9960 + }, + { + "epoch": 1.0938941357346805, + "grad_norm": 2.165177822113037, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7140014171600342, + "num_tokens": 255488805.0, + "step": 9961 + }, + { + "epoch": 1.094003953437294, + "grad_norm": 2.3644280433654785, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7276928424835205, + "num_tokens": 255511731.0, + "step": 9962 + }, + { + "epoch": 1.0941137711399078, + "grad_norm": 2.277164936065674, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7268700003623962, + "num_tokens": 255536426.0, + "step": 9963 + }, + { + "epoch": 1.0942235888425214, + "grad_norm": 2.3977530002593994, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7232500910758972, + "num_tokens": 255560054.0, + "step": 9964 + }, + { + "epoch": 1.0943334065451351, + "grad_norm": 2.0029189586639404, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7087076902389526, + "num_tokens": 255590277.0, + "step": 9965 + }, + { + "epoch": 1.0944432242477486, + "grad_norm": 2.513282537460327, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7162942886352539, + "num_tokens": 255611938.0, + "step": 9966 + }, + { + "epoch": 1.0945530419503624, + "grad_norm": 2.1809277534484863, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7128002643585205, + "num_tokens": 255638493.0, + "step": 9967 + }, + { + "epoch": 1.0946628596529762, + "grad_norm": 2.0099222660064697, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.713654637336731, + "num_tokens": 255668127.0, + "step": 9968 + }, + { + "epoch": 1.0947726773555897, + "grad_norm": 2.3313236236572266, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7245709300041199, + "num_tokens": 255692589.0, + "step": 9969 + }, + { + "epoch": 1.0948824950582035, + "grad_norm": 2.295341730117798, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7219895124435425, + "num_tokens": 255715307.0, + "step": 9970 + }, + { + "epoch": 1.094992312760817, + "grad_norm": 2.3541886806488037, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7244667410850525, + "num_tokens": 255741201.0, + "step": 9971 + }, + { + "epoch": 1.0951021304634307, + "grad_norm": 2.5838406085968018, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7310701608657837, + "num_tokens": 255761221.0, + "step": 9972 + }, + { + "epoch": 1.0952119481660443, + "grad_norm": 2.356067419052124, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7228542566299438, + "num_tokens": 255785179.0, + "step": 9973 + }, + { + "epoch": 1.095321765868658, + "grad_norm": 2.2567994594573975, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7152987718582153, + "num_tokens": 255810397.0, + "step": 9974 + }, + { + "epoch": 1.0954315835712718, + "grad_norm": 2.3058013916015625, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6970216035842896, + "num_tokens": 255838095.0, + "step": 9975 + }, + { + "epoch": 1.0955414012738853, + "grad_norm": 2.1899571418762207, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7161136865615845, + "num_tokens": 255866888.0, + "step": 9976 + }, + { + "epoch": 1.095651218976499, + "grad_norm": 2.2603445053100586, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7138872742652893, + "num_tokens": 255894395.0, + "step": 9977 + }, + { + "epoch": 1.0957610366791126, + "grad_norm": 2.461074113845825, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7192875146865845, + "num_tokens": 255916869.0, + "step": 9978 + }, + { + "epoch": 1.0958708543817264, + "grad_norm": 2.2622430324554443, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7429298162460327, + "num_tokens": 255942344.0, + "step": 9979 + }, + { + "epoch": 1.09598067208434, + "grad_norm": 2.1524064540863037, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7032352685928345, + "num_tokens": 255971368.0, + "step": 9980 + }, + { + "epoch": 1.0960904897869537, + "grad_norm": 2.0166077613830566, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.706974446773529, + "num_tokens": 256001828.0, + "step": 9981 + }, + { + "epoch": 1.0962003074895672, + "grad_norm": 2.3076860904693604, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.731092631816864, + "num_tokens": 256025372.0, + "step": 9982 + }, + { + "epoch": 1.096310125192181, + "grad_norm": 2.288884401321411, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7273659706115723, + "num_tokens": 256048401.0, + "step": 9983 + }, + { + "epoch": 1.0964199428947947, + "grad_norm": 2.003912925720215, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7121877670288086, + "num_tokens": 256080428.0, + "step": 9984 + }, + { + "epoch": 1.0965297605974083, + "grad_norm": 2.0943944454193115, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7056651711463928, + "num_tokens": 256111968.0, + "step": 9985 + }, + { + "epoch": 1.096639578300022, + "grad_norm": 2.0999958515167236, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7218437194824219, + "num_tokens": 256141042.0, + "step": 9986 + }, + { + "epoch": 1.0967493960026355, + "grad_norm": 2.421933174133301, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7059576511383057, + "num_tokens": 256164547.0, + "step": 9987 + }, + { + "epoch": 1.0968592137052493, + "grad_norm": 1.921393632888794, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7204461097717285, + "num_tokens": 256195267.0, + "step": 9988 + }, + { + "epoch": 1.096969031407863, + "grad_norm": 2.430762529373169, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7190716862678528, + "num_tokens": 256218306.0, + "step": 9989 + }, + { + "epoch": 1.0970788491104766, + "grad_norm": 2.4433467388153076, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7067373991012573, + "num_tokens": 256242525.0, + "step": 9990 + }, + { + "epoch": 1.0971886668130904, + "grad_norm": 2.3958144187927246, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.708285391330719, + "num_tokens": 256268038.0, + "step": 9991 + }, + { + "epoch": 1.0972984845157039, + "grad_norm": 2.108938217163086, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7474148273468018, + "num_tokens": 256296259.0, + "step": 9992 + }, + { + "epoch": 1.0974083022183176, + "grad_norm": 2.6340415477752686, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7119436264038086, + "num_tokens": 256317102.0, + "step": 9993 + }, + { + "epoch": 1.0975181199209312, + "grad_norm": 2.0942482948303223, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.722093403339386, + "num_tokens": 256344300.0, + "step": 9994 + }, + { + "epoch": 1.097627937623545, + "grad_norm": 1.925674557685852, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7284476161003113, + "num_tokens": 256376133.0, + "step": 9995 + }, + { + "epoch": 1.0977377553261585, + "grad_norm": 2.547572612762451, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7374788522720337, + "num_tokens": 256395359.0, + "step": 9996 + }, + { + "epoch": 1.0978475730287722, + "grad_norm": 2.654266595840454, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.72270667552948, + "num_tokens": 256416431.0, + "step": 9997 + }, + { + "epoch": 1.097957390731386, + "grad_norm": 2.195530652999878, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7159240245819092, + "num_tokens": 256442769.0, + "step": 9998 + }, + { + "epoch": 1.0980672084339995, + "grad_norm": 2.0808799266815186, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7174713611602783, + "num_tokens": 256472400.0, + "step": 9999 + }, + { + "epoch": 1.0981770261366133, + "grad_norm": 2.1437556743621826, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7302125096321106, + "num_tokens": 256499670.0, + "step": 10000 + }, + { + "epoch": 1.0982868438392268, + "grad_norm": 2.4369921684265137, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7220544815063477, + "num_tokens": 256521853.0, + "step": 10001 + }, + { + "epoch": 1.0983966615418406, + "grad_norm": 2.352430820465088, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6941393613815308, + "num_tokens": 256548490.0, + "step": 10002 + }, + { + "epoch": 1.098506479244454, + "grad_norm": 1.7567338943481445, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7224198579788208, + "num_tokens": 256587783.0, + "step": 10003 + }, + { + "epoch": 1.0986162969470679, + "grad_norm": 2.5307695865631104, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7218583822250366, + "num_tokens": 256608466.0, + "step": 10004 + }, + { + "epoch": 1.0987261146496816, + "grad_norm": 2.1606688499450684, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7395383715629578, + "num_tokens": 256633929.0, + "step": 10005 + }, + { + "epoch": 1.0988359323522952, + "grad_norm": 2.6297874450683594, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7421330809593201, + "num_tokens": 256652834.0, + "step": 10006 + }, + { + "epoch": 1.098945750054909, + "grad_norm": 2.09617280960083, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7144753932952881, + "num_tokens": 256681390.0, + "step": 10007 + }, + { + "epoch": 1.0990555677575224, + "grad_norm": 2.34004282951355, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7349212169647217, + "num_tokens": 256704321.0, + "step": 10008 + }, + { + "epoch": 1.0991653854601362, + "grad_norm": 2.382603168487549, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7126551270484924, + "num_tokens": 256726914.0, + "step": 10009 + }, + { + "epoch": 1.0992752031627497, + "grad_norm": 2.30202579498291, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7057071924209595, + "num_tokens": 256754318.0, + "step": 10010 + }, + { + "epoch": 1.0993850208653635, + "grad_norm": 2.1856460571289062, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.684346079826355, + "num_tokens": 256785981.0, + "step": 10011 + }, + { + "epoch": 1.0994948385679773, + "grad_norm": 2.440950393676758, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7077631950378418, + "num_tokens": 256808367.0, + "step": 10012 + }, + { + "epoch": 1.0996046562705908, + "grad_norm": 1.9881746768951416, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7165732383728027, + "num_tokens": 256837714.0, + "step": 10013 + }, + { + "epoch": 1.0997144739732045, + "grad_norm": 2.264878749847412, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7124499082565308, + "num_tokens": 256863467.0, + "step": 10014 + }, + { + "epoch": 1.099824291675818, + "grad_norm": 2.623354196548462, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7422471046447754, + "num_tokens": 256882183.0, + "step": 10015 + }, + { + "epoch": 1.0999341093784318, + "grad_norm": 2.2403388023376465, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7019764184951782, + "num_tokens": 256909613.0, + "step": 10016 + }, + { + "epoch": 1.1000439270810454, + "grad_norm": 2.3274636268615723, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7021718621253967, + "num_tokens": 256935892.0, + "step": 10017 + }, + { + "epoch": 1.1001537447836591, + "grad_norm": 2.226726770401001, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7221989631652832, + "num_tokens": 256962301.0, + "step": 10018 + }, + { + "epoch": 1.1002635624862729, + "grad_norm": 2.4536430835723877, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7299160957336426, + "num_tokens": 256984008.0, + "step": 10019 + }, + { + "epoch": 1.1003733801888864, + "grad_norm": 2.2622673511505127, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7202897071838379, + "num_tokens": 257009977.0, + "step": 10020 + }, + { + "epoch": 1.1004831978915002, + "grad_norm": 2.440582513809204, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7108162641525269, + "num_tokens": 257034746.0, + "step": 10021 + }, + { + "epoch": 1.1005930155941137, + "grad_norm": 2.024848222732544, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7240661382675171, + "num_tokens": 257063832.0, + "step": 10022 + }, + { + "epoch": 1.1007028332967275, + "grad_norm": 2.1723575592041016, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.6997312307357788, + "num_tokens": 257089781.0, + "step": 10023 + }, + { + "epoch": 1.100812650999341, + "grad_norm": 2.3485231399536133, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7184730768203735, + "num_tokens": 257112578.0, + "step": 10024 + }, + { + "epoch": 1.1009224687019548, + "grad_norm": 2.487006902694702, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7312801480293274, + "num_tokens": 257132449.0, + "step": 10025 + }, + { + "epoch": 1.1010322864045685, + "grad_norm": 2.5201165676116943, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7173258662223816, + "num_tokens": 257153724.0, + "step": 10026 + }, + { + "epoch": 1.101142104107182, + "grad_norm": 2.3393771648406982, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7229081392288208, + "num_tokens": 257177504.0, + "step": 10027 + }, + { + "epoch": 1.1012519218097958, + "grad_norm": 2.641618251800537, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7136633396148682, + "num_tokens": 257197704.0, + "step": 10028 + }, + { + "epoch": 1.1013617395124093, + "grad_norm": 2.1719155311584473, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7204174995422363, + "num_tokens": 257223850.0, + "step": 10029 + }, + { + "epoch": 1.101471557215023, + "grad_norm": 1.9360724687576294, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.7015741467475891, + "num_tokens": 257256348.0, + "step": 10030 + }, + { + "epoch": 1.1015813749176366, + "grad_norm": 1.8656738996505737, + "learning_rate": 1e-06, + "loss": 1.1145, + "mean_token_accuracy": 0.663912832736969, + "num_tokens": 257291728.0, + "step": 10031 + }, + { + "epoch": 1.1016911926202504, + "grad_norm": 2.1408510208129883, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7173177003860474, + "num_tokens": 257317884.0, + "step": 10032 + }, + { + "epoch": 1.1018010103228641, + "grad_norm": 2.4331915378570557, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7184900045394897, + "num_tokens": 257339860.0, + "step": 10033 + }, + { + "epoch": 1.1019108280254777, + "grad_norm": 2.2522165775299072, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7069107890129089, + "num_tokens": 257366022.0, + "step": 10034 + }, + { + "epoch": 1.1020206457280914, + "grad_norm": 1.9615086317062378, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7086242437362671, + "num_tokens": 257397647.0, + "step": 10035 + }, + { + "epoch": 1.102130463430705, + "grad_norm": 2.725585699081421, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7192555665969849, + "num_tokens": 257416923.0, + "step": 10036 + }, + { + "epoch": 1.1022402811333187, + "grad_norm": 2.3567731380462646, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7186852693557739, + "num_tokens": 257439810.0, + "step": 10037 + }, + { + "epoch": 1.1023500988359323, + "grad_norm": 2.38919997215271, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7324551343917847, + "num_tokens": 257462560.0, + "step": 10038 + }, + { + "epoch": 1.102459916538546, + "grad_norm": 2.043389320373535, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7311628460884094, + "num_tokens": 257489865.0, + "step": 10039 + }, + { + "epoch": 1.1025697342411598, + "grad_norm": 2.27038311958313, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.715795636177063, + "num_tokens": 257515680.0, + "step": 10040 + }, + { + "epoch": 1.1026795519437733, + "grad_norm": 2.143310070037842, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7108265161514282, + "num_tokens": 257544197.0, + "step": 10041 + }, + { + "epoch": 1.102789369646387, + "grad_norm": 2.177508592605591, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7241506576538086, + "num_tokens": 257568481.0, + "step": 10042 + }, + { + "epoch": 1.1028991873490006, + "grad_norm": 2.409320592880249, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7180020809173584, + "num_tokens": 257591556.0, + "step": 10043 + }, + { + "epoch": 1.1030090050516144, + "grad_norm": 2.140563726425171, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7600855231285095, + "num_tokens": 257615846.0, + "step": 10044 + }, + { + "epoch": 1.103118822754228, + "grad_norm": 2.3466477394104004, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.730228066444397, + "num_tokens": 257638611.0, + "step": 10045 + }, + { + "epoch": 1.1032286404568417, + "grad_norm": 2.3168563842773438, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7311030626296997, + "num_tokens": 257663422.0, + "step": 10046 + }, + { + "epoch": 1.1033384581594552, + "grad_norm": 2.3218297958374023, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7181209325790405, + "num_tokens": 257688259.0, + "step": 10047 + }, + { + "epoch": 1.103448275862069, + "grad_norm": 2.2870595455169678, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7031469941139221, + "num_tokens": 257715398.0, + "step": 10048 + }, + { + "epoch": 1.1035580935646827, + "grad_norm": 2.1746103763580322, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7073587775230408, + "num_tokens": 257742823.0, + "step": 10049 + }, + { + "epoch": 1.1036679112672962, + "grad_norm": 2.5273890495300293, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7513065934181213, + "num_tokens": 257763455.0, + "step": 10050 + }, + { + "epoch": 1.10377772896991, + "grad_norm": 2.2952516078948975, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7332211136817932, + "num_tokens": 257786791.0, + "step": 10051 + }, + { + "epoch": 1.1038875466725235, + "grad_norm": 2.299010753631592, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7089627385139465, + "num_tokens": 257811133.0, + "step": 10052 + }, + { + "epoch": 1.1039973643751373, + "grad_norm": 2.51230788230896, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7079629898071289, + "num_tokens": 257832440.0, + "step": 10053 + }, + { + "epoch": 1.104107182077751, + "grad_norm": 1.9908697605133057, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.722819447517395, + "num_tokens": 257862683.0, + "step": 10054 + }, + { + "epoch": 1.1042169997803646, + "grad_norm": 2.1666572093963623, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7136458158493042, + "num_tokens": 257891922.0, + "step": 10055 + }, + { + "epoch": 1.1043268174829783, + "grad_norm": 2.156675100326538, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7181260585784912, + "num_tokens": 257918538.0, + "step": 10056 + }, + { + "epoch": 1.1044366351855919, + "grad_norm": 2.1876347064971924, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.7010555863380432, + "num_tokens": 257946612.0, + "step": 10057 + }, + { + "epoch": 1.1045464528882056, + "grad_norm": 2.2454569339752197, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7105675935745239, + "num_tokens": 257972978.0, + "step": 10058 + }, + { + "epoch": 1.1046562705908192, + "grad_norm": 2.0102856159210205, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.709841251373291, + "num_tokens": 258004068.0, + "step": 10059 + }, + { + "epoch": 1.104766088293433, + "grad_norm": 2.378574848175049, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7291460037231445, + "num_tokens": 258025618.0, + "step": 10060 + }, + { + "epoch": 1.1048759059960465, + "grad_norm": 2.2818446159362793, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7202216386795044, + "num_tokens": 258050178.0, + "step": 10061 + }, + { + "epoch": 1.1049857236986602, + "grad_norm": 2.2816574573516846, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7201765775680542, + "num_tokens": 258073616.0, + "step": 10062 + }, + { + "epoch": 1.105095541401274, + "grad_norm": 1.9980175495147705, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7070962190628052, + "num_tokens": 258104243.0, + "step": 10063 + }, + { + "epoch": 1.1052053591038875, + "grad_norm": 2.2931466102600098, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7257416248321533, + "num_tokens": 258129938.0, + "step": 10064 + }, + { + "epoch": 1.1053151768065013, + "grad_norm": 2.45475435256958, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7471235394477844, + "num_tokens": 258150854.0, + "step": 10065 + }, + { + "epoch": 1.1054249945091148, + "grad_norm": 2.4135448932647705, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7155057191848755, + "num_tokens": 258172998.0, + "step": 10066 + }, + { + "epoch": 1.1055348122117286, + "grad_norm": 2.278097152709961, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7222323417663574, + "num_tokens": 258195412.0, + "step": 10067 + }, + { + "epoch": 1.105644629914342, + "grad_norm": 1.9787389039993286, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7248767614364624, + "num_tokens": 258227833.0, + "step": 10068 + }, + { + "epoch": 1.1057544476169558, + "grad_norm": 2.126615047454834, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7053041458129883, + "num_tokens": 258255369.0, + "step": 10069 + }, + { + "epoch": 1.1058642653195696, + "grad_norm": 2.1270720958709717, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7216459512710571, + "num_tokens": 258283062.0, + "step": 10070 + }, + { + "epoch": 1.1059740830221831, + "grad_norm": 2.159557819366455, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7170010805130005, + "num_tokens": 258311358.0, + "step": 10071 + }, + { + "epoch": 1.106083900724797, + "grad_norm": 2.156419515609741, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7172845602035522, + "num_tokens": 258337225.0, + "step": 10072 + }, + { + "epoch": 1.1061937184274104, + "grad_norm": 1.9883053302764893, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7298693656921387, + "num_tokens": 258368566.0, + "step": 10073 + }, + { + "epoch": 1.1063035361300242, + "grad_norm": 2.067012071609497, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7252292633056641, + "num_tokens": 258398877.0, + "step": 10074 + }, + { + "epoch": 1.1064133538326377, + "grad_norm": 2.1097569465637207, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7120418548583984, + "num_tokens": 258427739.0, + "step": 10075 + }, + { + "epoch": 1.1065231715352515, + "grad_norm": 2.4633331298828125, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7224898338317871, + "num_tokens": 258448336.0, + "step": 10076 + }, + { + "epoch": 1.1066329892378652, + "grad_norm": 2.186472177505493, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7150986194610596, + "num_tokens": 258476125.0, + "step": 10077 + }, + { + "epoch": 1.1067428069404788, + "grad_norm": 2.374079465866089, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7334124445915222, + "num_tokens": 258497929.0, + "step": 10078 + }, + { + "epoch": 1.1068526246430925, + "grad_norm": 2.256282329559326, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7261836528778076, + "num_tokens": 258522875.0, + "step": 10079 + }, + { + "epoch": 1.106962442345706, + "grad_norm": 2.6043283939361572, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7111299633979797, + "num_tokens": 258542734.0, + "step": 10080 + }, + { + "epoch": 1.1070722600483198, + "grad_norm": 2.484151601791382, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7306480407714844, + "num_tokens": 258563933.0, + "step": 10081 + }, + { + "epoch": 1.1071820777509334, + "grad_norm": 2.5364530086517334, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7310663461685181, + "num_tokens": 258585098.0, + "step": 10082 + }, + { + "epoch": 1.1072918954535471, + "grad_norm": 2.2462594509124756, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7208805680274963, + "num_tokens": 258608783.0, + "step": 10083 + }, + { + "epoch": 1.1074017131561609, + "grad_norm": 2.2792911529541016, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7387007474899292, + "num_tokens": 258632872.0, + "step": 10084 + }, + { + "epoch": 1.1075115308587744, + "grad_norm": 2.320791244506836, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7377399206161499, + "num_tokens": 258655989.0, + "step": 10085 + }, + { + "epoch": 1.1076213485613882, + "grad_norm": 2.132718801498413, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7018812894821167, + "num_tokens": 258683067.0, + "step": 10086 + }, + { + "epoch": 1.1077311662640017, + "grad_norm": 2.2887821197509766, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7198686003684998, + "num_tokens": 258707785.0, + "step": 10087 + }, + { + "epoch": 1.1078409839666155, + "grad_norm": 2.1344165802001953, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7392992973327637, + "num_tokens": 258733334.0, + "step": 10088 + }, + { + "epoch": 1.107950801669229, + "grad_norm": 2.1686952114105225, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7196148633956909, + "num_tokens": 258757919.0, + "step": 10089 + }, + { + "epoch": 1.1080606193718427, + "grad_norm": 2.5327768325805664, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.724026620388031, + "num_tokens": 258779209.0, + "step": 10090 + }, + { + "epoch": 1.1081704370744565, + "grad_norm": 2.599465847015381, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.749399721622467, + "num_tokens": 258797300.0, + "step": 10091 + }, + { + "epoch": 1.10828025477707, + "grad_norm": 2.329082727432251, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7368708252906799, + "num_tokens": 258818923.0, + "step": 10092 + }, + { + "epoch": 1.1083900724796838, + "grad_norm": 2.0026817321777344, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7257014513015747, + "num_tokens": 258846498.0, + "step": 10093 + }, + { + "epoch": 1.1084998901822973, + "grad_norm": 2.097552537918091, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7016850709915161, + "num_tokens": 258876092.0, + "step": 10094 + }, + { + "epoch": 1.108609707884911, + "grad_norm": 2.3577895164489746, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.693763256072998, + "num_tokens": 258902804.0, + "step": 10095 + }, + { + "epoch": 1.1087195255875246, + "grad_norm": 2.2081165313720703, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7175902724266052, + "num_tokens": 258928942.0, + "step": 10096 + }, + { + "epoch": 1.1088293432901384, + "grad_norm": 2.350562572479248, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7341499328613281, + "num_tokens": 258951177.0, + "step": 10097 + }, + { + "epoch": 1.108939160992752, + "grad_norm": 2.1970763206481934, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6961983442306519, + "num_tokens": 258979382.0, + "step": 10098 + }, + { + "epoch": 1.1090489786953657, + "grad_norm": 2.2077503204345703, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7267338037490845, + "num_tokens": 259005734.0, + "step": 10099 + }, + { + "epoch": 1.1091587963979794, + "grad_norm": 2.27323317527771, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7217673063278198, + "num_tokens": 259030139.0, + "step": 10100 + }, + { + "epoch": 1.109268614100593, + "grad_norm": 2.235926389694214, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7108629941940308, + "num_tokens": 259055621.0, + "step": 10101 + }, + { + "epoch": 1.1093784318032067, + "grad_norm": 2.0010645389556885, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6881849765777588, + "num_tokens": 259087472.0, + "step": 10102 + }, + { + "epoch": 1.1094882495058203, + "grad_norm": 2.1951048374176025, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7085431814193726, + "num_tokens": 259114616.0, + "step": 10103 + }, + { + "epoch": 1.109598067208434, + "grad_norm": 2.3665997982025146, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7472043037414551, + "num_tokens": 259137579.0, + "step": 10104 + }, + { + "epoch": 1.1097078849110478, + "grad_norm": 2.65720796585083, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7075858116149902, + "num_tokens": 259158880.0, + "step": 10105 + }, + { + "epoch": 1.1098177026136613, + "grad_norm": 2.1420798301696777, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.690712034702301, + "num_tokens": 259186865.0, + "step": 10106 + }, + { + "epoch": 1.109927520316275, + "grad_norm": 2.239631414413452, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7276601791381836, + "num_tokens": 259210895.0, + "step": 10107 + }, + { + "epoch": 1.1100373380188886, + "grad_norm": 2.290297746658325, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7137181758880615, + "num_tokens": 259235433.0, + "step": 10108 + }, + { + "epoch": 1.1101471557215024, + "grad_norm": 2.4292805194854736, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.707533061504364, + "num_tokens": 259263909.0, + "step": 10109 + }, + { + "epoch": 1.110256973424116, + "grad_norm": 2.194577932357788, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7105526924133301, + "num_tokens": 259292945.0, + "step": 10110 + }, + { + "epoch": 1.1103667911267296, + "grad_norm": 2.699481248855591, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7398185729980469, + "num_tokens": 259310702.0, + "step": 10111 + }, + { + "epoch": 1.1104766088293432, + "grad_norm": 2.2123031616210938, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7295975089073181, + "num_tokens": 259337339.0, + "step": 10112 + }, + { + "epoch": 1.110586426531957, + "grad_norm": 2.453348398208618, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7230693101882935, + "num_tokens": 259359888.0, + "step": 10113 + }, + { + "epoch": 1.1106962442345707, + "grad_norm": 2.2569093704223633, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7123174667358398, + "num_tokens": 259385743.0, + "step": 10114 + }, + { + "epoch": 1.1108060619371842, + "grad_norm": 2.402435541152954, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7273945212364197, + "num_tokens": 259409042.0, + "step": 10115 + }, + { + "epoch": 1.110915879639798, + "grad_norm": 2.446115493774414, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7205328941345215, + "num_tokens": 259430946.0, + "step": 10116 + }, + { + "epoch": 1.1110256973424115, + "grad_norm": 2.0168468952178955, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7279490232467651, + "num_tokens": 259461740.0, + "step": 10117 + }, + { + "epoch": 1.1111355150450253, + "grad_norm": 2.2787067890167236, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7331785559654236, + "num_tokens": 259487830.0, + "step": 10118 + }, + { + "epoch": 1.111245332747639, + "grad_norm": 2.1277267932891846, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.727209746837616, + "num_tokens": 259514717.0, + "step": 10119 + }, + { + "epoch": 1.1113551504502526, + "grad_norm": 2.0878913402557373, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7354357242584229, + "num_tokens": 259544452.0, + "step": 10120 + }, + { + "epoch": 1.1114649681528663, + "grad_norm": 1.929813027381897, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7124080657958984, + "num_tokens": 259577122.0, + "step": 10121 + }, + { + "epoch": 1.1115747858554799, + "grad_norm": 2.322110652923584, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6976820826530457, + "num_tokens": 259602432.0, + "step": 10122 + }, + { + "epoch": 1.1116846035580936, + "grad_norm": 2.3322339057922363, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7208063006401062, + "num_tokens": 259625323.0, + "step": 10123 + }, + { + "epoch": 1.1117944212607072, + "grad_norm": 2.195892810821533, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6946709752082825, + "num_tokens": 259652965.0, + "step": 10124 + }, + { + "epoch": 1.111904238963321, + "grad_norm": 1.898138403892517, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.718502402305603, + "num_tokens": 259686245.0, + "step": 10125 + }, + { + "epoch": 1.1120140566659344, + "grad_norm": 2.0321226119995117, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7225135564804077, + "num_tokens": 259714720.0, + "step": 10126 + }, + { + "epoch": 1.1121238743685482, + "grad_norm": 2.2287163734436035, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7294970750808716, + "num_tokens": 259740260.0, + "step": 10127 + }, + { + "epoch": 1.112233692071162, + "grad_norm": 2.3695924282073975, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7265128493309021, + "num_tokens": 259764533.0, + "step": 10128 + }, + { + "epoch": 1.1123435097737755, + "grad_norm": 2.3062398433685303, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6940226554870605, + "num_tokens": 259793470.0, + "step": 10129 + }, + { + "epoch": 1.1124533274763893, + "grad_norm": 2.242924928665161, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7315463423728943, + "num_tokens": 259818276.0, + "step": 10130 + }, + { + "epoch": 1.1125631451790028, + "grad_norm": 2.3451521396636963, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7185314297676086, + "num_tokens": 259842212.0, + "step": 10131 + }, + { + "epoch": 1.1126729628816165, + "grad_norm": 2.3545970916748047, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7322801351547241, + "num_tokens": 259864806.0, + "step": 10132 + }, + { + "epoch": 1.11278278058423, + "grad_norm": 2.4552001953125, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7152615785598755, + "num_tokens": 259886004.0, + "step": 10133 + }, + { + "epoch": 1.1128925982868438, + "grad_norm": 2.436567783355713, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7125034928321838, + "num_tokens": 259908883.0, + "step": 10134 + }, + { + "epoch": 1.1130024159894576, + "grad_norm": 2.3714773654937744, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.73459792137146, + "num_tokens": 259932361.0, + "step": 10135 + }, + { + "epoch": 1.1131122336920711, + "grad_norm": 2.287301778793335, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7017358541488647, + "num_tokens": 259957252.0, + "step": 10136 + }, + { + "epoch": 1.1132220513946849, + "grad_norm": 2.4492292404174805, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7233783602714539, + "num_tokens": 259979142.0, + "step": 10137 + }, + { + "epoch": 1.1133318690972984, + "grad_norm": 2.0784313678741455, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.6863691806793213, + "num_tokens": 260006824.0, + "step": 10138 + }, + { + "epoch": 1.1134416867999122, + "grad_norm": 2.4565658569335938, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7249456644058228, + "num_tokens": 260029679.0, + "step": 10139 + }, + { + "epoch": 1.1135515045025257, + "grad_norm": 2.224055767059326, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7368308305740356, + "num_tokens": 260055590.0, + "step": 10140 + }, + { + "epoch": 1.1136613222051395, + "grad_norm": 2.691333055496216, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7255872488021851, + "num_tokens": 260076793.0, + "step": 10141 + }, + { + "epoch": 1.1137711399077532, + "grad_norm": 2.389492988586426, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7369998693466187, + "num_tokens": 260099142.0, + "step": 10142 + }, + { + "epoch": 1.1138809576103668, + "grad_norm": 2.5852408409118652, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7188559174537659, + "num_tokens": 260121171.0, + "step": 10143 + }, + { + "epoch": 1.1139907753129805, + "grad_norm": 2.3090262413024902, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7266901731491089, + "num_tokens": 260145442.0, + "step": 10144 + }, + { + "epoch": 1.114100593015594, + "grad_norm": 2.767817258834839, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7479395270347595, + "num_tokens": 260162527.0, + "step": 10145 + }, + { + "epoch": 1.1142104107182078, + "grad_norm": 2.196103096008301, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.708996593952179, + "num_tokens": 260191180.0, + "step": 10146 + }, + { + "epoch": 1.1143202284208213, + "grad_norm": 2.195728063583374, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7140375375747681, + "num_tokens": 260215208.0, + "step": 10147 + }, + { + "epoch": 1.114430046123435, + "grad_norm": 2.1598472595214844, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7172881364822388, + "num_tokens": 260243549.0, + "step": 10148 + }, + { + "epoch": 1.1145398638260489, + "grad_norm": 2.263669490814209, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7185395359992981, + "num_tokens": 260268710.0, + "step": 10149 + }, + { + "epoch": 1.1146496815286624, + "grad_norm": 2.262075185775757, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7367666959762573, + "num_tokens": 260291884.0, + "step": 10150 + }, + { + "epoch": 1.1147594992312762, + "grad_norm": 2.2753686904907227, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6897045373916626, + "num_tokens": 260317931.0, + "step": 10151 + }, + { + "epoch": 1.1148693169338897, + "grad_norm": 2.2877678871154785, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7174487113952637, + "num_tokens": 260345279.0, + "step": 10152 + }, + { + "epoch": 1.1149791346365034, + "grad_norm": 2.474302291870117, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7234177589416504, + "num_tokens": 260368285.0, + "step": 10153 + }, + { + "epoch": 1.115088952339117, + "grad_norm": 2.163752317428589, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6897821426391602, + "num_tokens": 260397223.0, + "step": 10154 + }, + { + "epoch": 1.1151987700417307, + "grad_norm": 2.236177444458008, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7133976221084595, + "num_tokens": 260422924.0, + "step": 10155 + }, + { + "epoch": 1.1153085877443445, + "grad_norm": 2.276615858078003, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7053384780883789, + "num_tokens": 260450694.0, + "step": 10156 + }, + { + "epoch": 1.115418405446958, + "grad_norm": 2.381312608718872, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7280139327049255, + "num_tokens": 260473118.0, + "step": 10157 + }, + { + "epoch": 1.1155282231495718, + "grad_norm": 2.3751957416534424, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7335013151168823, + "num_tokens": 260495445.0, + "step": 10158 + }, + { + "epoch": 1.1156380408521853, + "grad_norm": 2.3710529804229736, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7192134261131287, + "num_tokens": 260518269.0, + "step": 10159 + }, + { + "epoch": 1.115747858554799, + "grad_norm": 2.0426480770111084, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7115607261657715, + "num_tokens": 260546347.0, + "step": 10160 + }, + { + "epoch": 1.1158576762574126, + "grad_norm": 2.5732507705688477, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7168221473693848, + "num_tokens": 260567388.0, + "step": 10161 + }, + { + "epoch": 1.1159674939600264, + "grad_norm": 2.552718162536621, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7234100103378296, + "num_tokens": 260589203.0, + "step": 10162 + }, + { + "epoch": 1.11607731166264, + "grad_norm": 2.2030982971191406, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7008383274078369, + "num_tokens": 260616950.0, + "step": 10163 + }, + { + "epoch": 1.1161871293652537, + "grad_norm": 2.056809663772583, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7041211128234863, + "num_tokens": 260646839.0, + "step": 10164 + }, + { + "epoch": 1.1162969470678674, + "grad_norm": 2.205599308013916, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7145947217941284, + "num_tokens": 260673172.0, + "step": 10165 + }, + { + "epoch": 1.116406764770481, + "grad_norm": 2.1913230419158936, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7129871249198914, + "num_tokens": 260700657.0, + "step": 10166 + }, + { + "epoch": 1.1165165824730947, + "grad_norm": 2.372206449508667, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7039780616760254, + "num_tokens": 260721874.0, + "step": 10167 + }, + { + "epoch": 1.1166264001757082, + "grad_norm": 2.369117021560669, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7144173383712769, + "num_tokens": 260746414.0, + "step": 10168 + }, + { + "epoch": 1.116736217878322, + "grad_norm": 2.308206558227539, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6855084896087646, + "num_tokens": 260773944.0, + "step": 10169 + }, + { + "epoch": 1.1168460355809358, + "grad_norm": 2.265380620956421, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7277663350105286, + "num_tokens": 260801564.0, + "step": 10170 + }, + { + "epoch": 1.1169558532835493, + "grad_norm": 2.1707096099853516, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7096045017242432, + "num_tokens": 260829606.0, + "step": 10171 + }, + { + "epoch": 1.117065670986163, + "grad_norm": 2.646963596343994, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7404892444610596, + "num_tokens": 260849344.0, + "step": 10172 + }, + { + "epoch": 1.1171754886887766, + "grad_norm": 2.568526268005371, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7184698581695557, + "num_tokens": 260871816.0, + "step": 10173 + }, + { + "epoch": 1.1172853063913903, + "grad_norm": 2.335365056991577, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7016323208808899, + "num_tokens": 260897546.0, + "step": 10174 + }, + { + "epoch": 1.1173951240940039, + "grad_norm": 2.288372039794922, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7128810882568359, + "num_tokens": 260924332.0, + "step": 10175 + }, + { + "epoch": 1.1175049417966176, + "grad_norm": 2.1223549842834473, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7310776710510254, + "num_tokens": 260951375.0, + "step": 10176 + }, + { + "epoch": 1.1176147594992312, + "grad_norm": 2.643371820449829, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7444236278533936, + "num_tokens": 260972940.0, + "step": 10177 + }, + { + "epoch": 1.117724577201845, + "grad_norm": 2.4782400131225586, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7330077290534973, + "num_tokens": 260995336.0, + "step": 10178 + }, + { + "epoch": 1.1178343949044587, + "grad_norm": 2.3877389430999756, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7390099763870239, + "num_tokens": 261019713.0, + "step": 10179 + }, + { + "epoch": 1.1179442126070722, + "grad_norm": 2.1921894550323486, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7018680572509766, + "num_tokens": 261048607.0, + "step": 10180 + }, + { + "epoch": 1.118054030309686, + "grad_norm": 2.365093231201172, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7065412402153015, + "num_tokens": 261074432.0, + "step": 10181 + }, + { + "epoch": 1.1181638480122995, + "grad_norm": 2.423914909362793, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7137001156806946, + "num_tokens": 261097188.0, + "step": 10182 + }, + { + "epoch": 1.1182736657149133, + "grad_norm": 2.0427052974700928, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7295278310775757, + "num_tokens": 261126067.0, + "step": 10183 + }, + { + "epoch": 1.1183834834175268, + "grad_norm": 2.412224292755127, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7020789980888367, + "num_tokens": 261148977.0, + "step": 10184 + }, + { + "epoch": 1.1184933011201406, + "grad_norm": 2.2750823497772217, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7234015464782715, + "num_tokens": 261172542.0, + "step": 10185 + }, + { + "epoch": 1.1186031188227543, + "grad_norm": 2.389272451400757, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7139878273010254, + "num_tokens": 261196856.0, + "step": 10186 + }, + { + "epoch": 1.1187129365253679, + "grad_norm": 2.3869211673736572, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6924474239349365, + "num_tokens": 261222608.0, + "step": 10187 + }, + { + "epoch": 1.1188227542279816, + "grad_norm": 2.607877016067505, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7319211959838867, + "num_tokens": 261243007.0, + "step": 10188 + }, + { + "epoch": 1.1189325719305951, + "grad_norm": 2.0906107425689697, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7064474821090698, + "num_tokens": 261271317.0, + "step": 10189 + }, + { + "epoch": 1.119042389633209, + "grad_norm": 2.3012938499450684, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7154529094696045, + "num_tokens": 261295192.0, + "step": 10190 + }, + { + "epoch": 1.1191522073358224, + "grad_norm": 2.037355661392212, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7275475263595581, + "num_tokens": 261323058.0, + "step": 10191 + }, + { + "epoch": 1.1192620250384362, + "grad_norm": 2.4159395694732666, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7194203734397888, + "num_tokens": 261343686.0, + "step": 10192 + }, + { + "epoch": 1.11937184274105, + "grad_norm": 2.454908847808838, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7365198731422424, + "num_tokens": 261365577.0, + "step": 10193 + }, + { + "epoch": 1.1194816604436635, + "grad_norm": 2.3395707607269287, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7240468263626099, + "num_tokens": 261388685.0, + "step": 10194 + }, + { + "epoch": 1.1195914781462772, + "grad_norm": 2.13606595993042, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.6985092163085938, + "num_tokens": 261417380.0, + "step": 10195 + }, + { + "epoch": 1.1197012958488908, + "grad_norm": 2.4368367195129395, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7158211469650269, + "num_tokens": 261439675.0, + "step": 10196 + }, + { + "epoch": 1.1198111135515045, + "grad_norm": 2.450929641723633, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.732056200504303, + "num_tokens": 261463419.0, + "step": 10197 + }, + { + "epoch": 1.119920931254118, + "grad_norm": 2.1777665615081787, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.726870059967041, + "num_tokens": 261489795.0, + "step": 10198 + }, + { + "epoch": 1.1200307489567318, + "grad_norm": 2.1742849349975586, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7044270038604736, + "num_tokens": 261518417.0, + "step": 10199 + }, + { + "epoch": 1.1201405666593456, + "grad_norm": 2.4630579948425293, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7405186891555786, + "num_tokens": 261539175.0, + "step": 10200 + }, + { + "epoch": 1.1202503843619591, + "grad_norm": 1.9660850763320923, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6834047436714172, + "num_tokens": 261572986.0, + "step": 10201 + }, + { + "epoch": 1.1203602020645729, + "grad_norm": 2.2167930603027344, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7059716582298279, + "num_tokens": 261601490.0, + "step": 10202 + }, + { + "epoch": 1.1204700197671864, + "grad_norm": 2.3395800590515137, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7081937789916992, + "num_tokens": 261629591.0, + "step": 10203 + }, + { + "epoch": 1.1205798374698002, + "grad_norm": 2.344684362411499, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7244797945022583, + "num_tokens": 261655435.0, + "step": 10204 + }, + { + "epoch": 1.1206896551724137, + "grad_norm": 2.1102042198181152, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6995995044708252, + "num_tokens": 261684687.0, + "step": 10205 + }, + { + "epoch": 1.1207994728750275, + "grad_norm": 2.1245791912078857, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7091654539108276, + "num_tokens": 261714386.0, + "step": 10206 + }, + { + "epoch": 1.1209092905776412, + "grad_norm": 2.5624825954437256, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7433798909187317, + "num_tokens": 261734615.0, + "step": 10207 + }, + { + "epoch": 1.1210191082802548, + "grad_norm": 2.3740830421447754, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.702729344367981, + "num_tokens": 261758557.0, + "step": 10208 + }, + { + "epoch": 1.1211289259828685, + "grad_norm": 2.2362687587738037, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7100133299827576, + "num_tokens": 261786419.0, + "step": 10209 + }, + { + "epoch": 1.121238743685482, + "grad_norm": 2.2919328212738037, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7147117257118225, + "num_tokens": 261811432.0, + "step": 10210 + }, + { + "epoch": 1.1213485613880958, + "grad_norm": 2.250007152557373, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7205328941345215, + "num_tokens": 261837338.0, + "step": 10211 + }, + { + "epoch": 1.1214583790907093, + "grad_norm": 2.12448787689209, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7083345651626587, + "num_tokens": 261865552.0, + "step": 10212 + }, + { + "epoch": 1.121568196793323, + "grad_norm": 2.3469929695129395, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7309500575065613, + "num_tokens": 261887827.0, + "step": 10213 + }, + { + "epoch": 1.1216780144959368, + "grad_norm": 2.051072359085083, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7107644081115723, + "num_tokens": 261917494.0, + "step": 10214 + }, + { + "epoch": 1.1217878321985504, + "grad_norm": 2.4001247882843018, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7212331891059875, + "num_tokens": 261940392.0, + "step": 10215 + }, + { + "epoch": 1.1218976499011641, + "grad_norm": 2.213265895843506, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7000489234924316, + "num_tokens": 261965552.0, + "step": 10216 + }, + { + "epoch": 1.1220074676037777, + "grad_norm": 2.4503512382507324, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7298341989517212, + "num_tokens": 261985574.0, + "step": 10217 + }, + { + "epoch": 1.1221172853063914, + "grad_norm": 2.147864818572998, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6937177777290344, + "num_tokens": 262017064.0, + "step": 10218 + }, + { + "epoch": 1.122227103009005, + "grad_norm": 2.322040319442749, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7349238991737366, + "num_tokens": 262040655.0, + "step": 10219 + }, + { + "epoch": 1.1223369207116187, + "grad_norm": 2.1643333435058594, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7054105997085571, + "num_tokens": 262066630.0, + "step": 10220 + }, + { + "epoch": 1.1224467384142325, + "grad_norm": 2.1584532260894775, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7142514586448669, + "num_tokens": 262092853.0, + "step": 10221 + }, + { + "epoch": 1.122556556116846, + "grad_norm": 2.109494924545288, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.680197536945343, + "num_tokens": 262124089.0, + "step": 10222 + }, + { + "epoch": 1.1226663738194598, + "grad_norm": 1.81577730178833, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7165429592132568, + "num_tokens": 262158092.0, + "step": 10223 + }, + { + "epoch": 1.1227761915220733, + "grad_norm": 2.214040994644165, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7020284533500671, + "num_tokens": 262182840.0, + "step": 10224 + }, + { + "epoch": 1.122886009224687, + "grad_norm": 2.1382148265838623, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.708648681640625, + "num_tokens": 262212811.0, + "step": 10225 + }, + { + "epoch": 1.1229958269273006, + "grad_norm": 2.195537567138672, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.6989225149154663, + "num_tokens": 262243676.0, + "step": 10226 + }, + { + "epoch": 1.1231056446299144, + "grad_norm": 2.5673041343688965, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7230561971664429, + "num_tokens": 262262846.0, + "step": 10227 + }, + { + "epoch": 1.123215462332528, + "grad_norm": 2.163905382156372, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7240380048751831, + "num_tokens": 262290433.0, + "step": 10228 + }, + { + "epoch": 1.1233252800351416, + "grad_norm": 2.1593143939971924, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7179420590400696, + "num_tokens": 262317767.0, + "step": 10229 + }, + { + "epoch": 1.1234350977377554, + "grad_norm": 2.177264928817749, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7242728471755981, + "num_tokens": 262343045.0, + "step": 10230 + }, + { + "epoch": 1.123544915440369, + "grad_norm": 2.150735855102539, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6926244497299194, + "num_tokens": 262372742.0, + "step": 10231 + }, + { + "epoch": 1.1236547331429827, + "grad_norm": 2.7264089584350586, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7220836281776428, + "num_tokens": 262393031.0, + "step": 10232 + }, + { + "epoch": 1.1237645508455962, + "grad_norm": 2.0905184745788574, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7183263301849365, + "num_tokens": 262420554.0, + "step": 10233 + }, + { + "epoch": 1.12387436854821, + "grad_norm": 2.294368028640747, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7161757349967957, + "num_tokens": 262444342.0, + "step": 10234 + }, + { + "epoch": 1.1239841862508237, + "grad_norm": 2.541701316833496, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7247899770736694, + "num_tokens": 262465359.0, + "step": 10235 + }, + { + "epoch": 1.1240940039534373, + "grad_norm": 2.5098202228546143, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7336592674255371, + "num_tokens": 262485527.0, + "step": 10236 + }, + { + "epoch": 1.124203821656051, + "grad_norm": 2.3303780555725098, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7312646508216858, + "num_tokens": 262508946.0, + "step": 10237 + }, + { + "epoch": 1.1243136393586646, + "grad_norm": 2.38227915763855, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7132217884063721, + "num_tokens": 262532045.0, + "step": 10238 + }, + { + "epoch": 1.1244234570612783, + "grad_norm": 2.259619951248169, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.697574257850647, + "num_tokens": 262557547.0, + "step": 10239 + }, + { + "epoch": 1.1245332747638919, + "grad_norm": 2.1514668464660645, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7039267420768738, + "num_tokens": 262587392.0, + "step": 10240 + }, + { + "epoch": 1.1246430924665056, + "grad_norm": 2.094935894012451, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7150477170944214, + "num_tokens": 262615916.0, + "step": 10241 + }, + { + "epoch": 1.1247529101691192, + "grad_norm": 2.121325969696045, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7123926877975464, + "num_tokens": 262646269.0, + "step": 10242 + }, + { + "epoch": 1.124862727871733, + "grad_norm": 2.1632347106933594, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7317020893096924, + "num_tokens": 262673597.0, + "step": 10243 + }, + { + "epoch": 1.1249725455743467, + "grad_norm": 2.2407968044281006, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7245314717292786, + "num_tokens": 262698478.0, + "step": 10244 + }, + { + "epoch": 1.1250823632769602, + "grad_norm": 2.432502508163452, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7113075256347656, + "num_tokens": 262721804.0, + "step": 10245 + }, + { + "epoch": 1.125192180979574, + "grad_norm": 2.214374542236328, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6875032782554626, + "num_tokens": 262751843.0, + "step": 10246 + }, + { + "epoch": 1.1253019986821875, + "grad_norm": 2.4689741134643555, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7259879112243652, + "num_tokens": 262773706.0, + "step": 10247 + }, + { + "epoch": 1.1254118163848013, + "grad_norm": 2.197903871536255, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7174079418182373, + "num_tokens": 262799889.0, + "step": 10248 + }, + { + "epoch": 1.125521634087415, + "grad_norm": 2.247687339782715, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7210152745246887, + "num_tokens": 262826657.0, + "step": 10249 + }, + { + "epoch": 1.1256314517900285, + "grad_norm": 2.410919666290283, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7332766056060791, + "num_tokens": 262848381.0, + "step": 10250 + }, + { + "epoch": 1.1257412694926423, + "grad_norm": 2.2266111373901367, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7002828121185303, + "num_tokens": 262874730.0, + "step": 10251 + }, + { + "epoch": 1.1258510871952558, + "grad_norm": 2.164167881011963, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7182579040527344, + "num_tokens": 262902307.0, + "step": 10252 + }, + { + "epoch": 1.1259609048978696, + "grad_norm": 2.065340995788574, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.729718804359436, + "num_tokens": 262930025.0, + "step": 10253 + }, + { + "epoch": 1.1260707226004831, + "grad_norm": 2.2758607864379883, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7199792861938477, + "num_tokens": 262954801.0, + "step": 10254 + }, + { + "epoch": 1.1261805403030969, + "grad_norm": 2.1356279850006104, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7112231254577637, + "num_tokens": 262984124.0, + "step": 10255 + }, + { + "epoch": 1.1262903580057104, + "grad_norm": 2.366844415664673, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7318472862243652, + "num_tokens": 263008637.0, + "step": 10256 + }, + { + "epoch": 1.1264001757083242, + "grad_norm": 2.206044912338257, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7123650312423706, + "num_tokens": 263035166.0, + "step": 10257 + }, + { + "epoch": 1.126509993410938, + "grad_norm": 2.2308809757232666, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7140874862670898, + "num_tokens": 263061481.0, + "step": 10258 + }, + { + "epoch": 1.1266198111135515, + "grad_norm": 2.1458773612976074, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6908445358276367, + "num_tokens": 263091540.0, + "step": 10259 + }, + { + "epoch": 1.1267296288161652, + "grad_norm": 2.2247772216796875, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7057434320449829, + "num_tokens": 263117929.0, + "step": 10260 + }, + { + "epoch": 1.1268394465187788, + "grad_norm": 2.175715684890747, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7134722471237183, + "num_tokens": 263145325.0, + "step": 10261 + }, + { + "epoch": 1.1269492642213925, + "grad_norm": 2.337167263031006, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.730381190776825, + "num_tokens": 263169290.0, + "step": 10262 + }, + { + "epoch": 1.127059081924006, + "grad_norm": 2.59371018409729, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7270167469978333, + "num_tokens": 263188979.0, + "step": 10263 + }, + { + "epoch": 1.1271688996266198, + "grad_norm": 2.5183403491973877, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7243614792823792, + "num_tokens": 263211142.0, + "step": 10264 + }, + { + "epoch": 1.1272787173292333, + "grad_norm": 2.240304946899414, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7057876586914062, + "num_tokens": 263237750.0, + "step": 10265 + }, + { + "epoch": 1.127388535031847, + "grad_norm": 2.129141092300415, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7515156269073486, + "num_tokens": 263262523.0, + "step": 10266 + }, + { + "epoch": 1.1274983527344609, + "grad_norm": 2.2100493907928467, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7092845439910889, + "num_tokens": 263289768.0, + "step": 10267 + }, + { + "epoch": 1.1276081704370744, + "grad_norm": 2.547614574432373, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7334690690040588, + "num_tokens": 263310133.0, + "step": 10268 + }, + { + "epoch": 1.1277179881396882, + "grad_norm": 2.545095205307007, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7174153327941895, + "num_tokens": 263329924.0, + "step": 10269 + }, + { + "epoch": 1.1278278058423017, + "grad_norm": 2.401067018508911, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7231768369674683, + "num_tokens": 263353103.0, + "step": 10270 + }, + { + "epoch": 1.1279376235449154, + "grad_norm": 2.2397189140319824, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.6971418857574463, + "num_tokens": 263377737.0, + "step": 10271 + }, + { + "epoch": 1.1280474412475292, + "grad_norm": 2.123732328414917, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7164296507835388, + "num_tokens": 263408007.0, + "step": 10272 + }, + { + "epoch": 1.1281572589501427, + "grad_norm": 2.2683236598968506, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7163214683532715, + "num_tokens": 263432353.0, + "step": 10273 + }, + { + "epoch": 1.1282670766527565, + "grad_norm": 2.4035086631774902, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7076301574707031, + "num_tokens": 263456294.0, + "step": 10274 + }, + { + "epoch": 1.12837689435537, + "grad_norm": 2.2421157360076904, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.725206732749939, + "num_tokens": 263481223.0, + "step": 10275 + }, + { + "epoch": 1.1284867120579838, + "grad_norm": 2.011013984680176, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7258048057556152, + "num_tokens": 263509746.0, + "step": 10276 + }, + { + "epoch": 1.1285965297605973, + "grad_norm": 2.6745853424072266, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7208609580993652, + "num_tokens": 263528558.0, + "step": 10277 + }, + { + "epoch": 1.128706347463211, + "grad_norm": 2.4939634799957275, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7077643275260925, + "num_tokens": 263550481.0, + "step": 10278 + }, + { + "epoch": 1.1288161651658246, + "grad_norm": 2.209879159927368, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6976046562194824, + "num_tokens": 263579528.0, + "step": 10279 + }, + { + "epoch": 1.1289259828684384, + "grad_norm": 2.339569330215454, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7180296182632446, + "num_tokens": 263604358.0, + "step": 10280 + }, + { + "epoch": 1.1290358005710521, + "grad_norm": 2.0623691082000732, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7171430587768555, + "num_tokens": 263632494.0, + "step": 10281 + }, + { + "epoch": 1.1291456182736657, + "grad_norm": 2.5221610069274902, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7216258645057678, + "num_tokens": 263654329.0, + "step": 10282 + }, + { + "epoch": 1.1292554359762794, + "grad_norm": 2.1066951751708984, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6991344094276428, + "num_tokens": 263682656.0, + "step": 10283 + }, + { + "epoch": 1.129365253678893, + "grad_norm": 2.0942275524139404, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6931525468826294, + "num_tokens": 263712926.0, + "step": 10284 + }, + { + "epoch": 1.1294750713815067, + "grad_norm": 2.026028871536255, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7257159352302551, + "num_tokens": 263744870.0, + "step": 10285 + }, + { + "epoch": 1.1295848890841205, + "grad_norm": 2.263120412826538, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7242092490196228, + "num_tokens": 263771101.0, + "step": 10286 + }, + { + "epoch": 1.129694706786734, + "grad_norm": 2.299940824508667, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7282341718673706, + "num_tokens": 263795447.0, + "step": 10287 + }, + { + "epoch": 1.1298045244893478, + "grad_norm": 2.2147886753082275, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7318115830421448, + "num_tokens": 263820250.0, + "step": 10288 + }, + { + "epoch": 1.1299143421919613, + "grad_norm": 2.3124866485595703, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7297699451446533, + "num_tokens": 263844403.0, + "step": 10289 + }, + { + "epoch": 1.130024159894575, + "grad_norm": 2.0533766746520996, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.684766948223114, + "num_tokens": 263873927.0, + "step": 10290 + }, + { + "epoch": 1.1301339775971886, + "grad_norm": 2.3632755279541016, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7199426293373108, + "num_tokens": 263896354.0, + "step": 10291 + }, + { + "epoch": 1.1302437952998023, + "grad_norm": 2.0874338150024414, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7199583649635315, + "num_tokens": 263925245.0, + "step": 10292 + }, + { + "epoch": 1.1303536130024159, + "grad_norm": 2.097038745880127, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7170664072036743, + "num_tokens": 263952630.0, + "step": 10293 + }, + { + "epoch": 1.1304634307050296, + "grad_norm": 2.7177045345306396, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7021059989929199, + "num_tokens": 263972679.0, + "step": 10294 + }, + { + "epoch": 1.1305732484076434, + "grad_norm": 2.445258378982544, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7232659459114075, + "num_tokens": 263994361.0, + "step": 10295 + }, + { + "epoch": 1.130683066110257, + "grad_norm": 2.003903388977051, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7103935480117798, + "num_tokens": 264026226.0, + "step": 10296 + }, + { + "epoch": 1.1307928838128707, + "grad_norm": 2.27290678024292, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7362918853759766, + "num_tokens": 264050077.0, + "step": 10297 + }, + { + "epoch": 1.1309027015154842, + "grad_norm": 2.3823225498199463, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7057590484619141, + "num_tokens": 264072168.0, + "step": 10298 + }, + { + "epoch": 1.131012519218098, + "grad_norm": 2.2642388343811035, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7147018313407898, + "num_tokens": 264098583.0, + "step": 10299 + }, + { + "epoch": 1.1311223369207117, + "grad_norm": 2.3181278705596924, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7030097246170044, + "num_tokens": 264122559.0, + "step": 10300 + }, + { + "epoch": 1.1312321546233253, + "grad_norm": 2.29140567779541, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7038781642913818, + "num_tokens": 264147876.0, + "step": 10301 + }, + { + "epoch": 1.131341972325939, + "grad_norm": 2.229235887527466, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7196962833404541, + "num_tokens": 264172498.0, + "step": 10302 + }, + { + "epoch": 1.1314517900285526, + "grad_norm": 2.307321310043335, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7056256532669067, + "num_tokens": 264199463.0, + "step": 10303 + }, + { + "epoch": 1.1315616077311663, + "grad_norm": 2.5837228298187256, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7204521894454956, + "num_tokens": 264221181.0, + "step": 10304 + }, + { + "epoch": 1.1316714254337799, + "grad_norm": 2.3274388313293457, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7305821180343628, + "num_tokens": 264244086.0, + "step": 10305 + }, + { + "epoch": 1.1317812431363936, + "grad_norm": 1.9568105936050415, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7211323976516724, + "num_tokens": 264277491.0, + "step": 10306 + }, + { + "epoch": 1.1318910608390071, + "grad_norm": 2.1276369094848633, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7069342136383057, + "num_tokens": 264305508.0, + "step": 10307 + }, + { + "epoch": 1.132000878541621, + "grad_norm": 2.582160711288452, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7443445324897766, + "num_tokens": 264324399.0, + "step": 10308 + }, + { + "epoch": 1.1321106962442347, + "grad_norm": 2.1014606952667236, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7080926299095154, + "num_tokens": 264352355.0, + "step": 10309 + }, + { + "epoch": 1.1322205139468482, + "grad_norm": 2.0689826011657715, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7368273138999939, + "num_tokens": 264378962.0, + "step": 10310 + }, + { + "epoch": 1.132330331649462, + "grad_norm": 2.2011919021606445, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7082260251045227, + "num_tokens": 264406960.0, + "step": 10311 + }, + { + "epoch": 1.1324401493520755, + "grad_norm": 2.4851388931274414, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7257336974143982, + "num_tokens": 264429393.0, + "step": 10312 + }, + { + "epoch": 1.1325499670546892, + "grad_norm": 2.1049530506134033, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7040159702301025, + "num_tokens": 264458588.0, + "step": 10313 + }, + { + "epoch": 1.132659784757303, + "grad_norm": 2.820380449295044, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7213085889816284, + "num_tokens": 264477247.0, + "step": 10314 + }, + { + "epoch": 1.1327696024599165, + "grad_norm": 2.761077880859375, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.711977481842041, + "num_tokens": 264496866.0, + "step": 10315 + }, + { + "epoch": 1.1328794201625303, + "grad_norm": 2.1577699184417725, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7212929129600525, + "num_tokens": 264523822.0, + "step": 10316 + }, + { + "epoch": 1.1329892378651438, + "grad_norm": 2.3147122859954834, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7133753299713135, + "num_tokens": 264548228.0, + "step": 10317 + }, + { + "epoch": 1.1330990555677576, + "grad_norm": 2.0283172130584717, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.708585262298584, + "num_tokens": 264577617.0, + "step": 10318 + }, + { + "epoch": 1.1332088732703711, + "grad_norm": 2.227816581726074, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7132400274276733, + "num_tokens": 264604093.0, + "step": 10319 + }, + { + "epoch": 1.1333186909729849, + "grad_norm": 2.155735969543457, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7363100647926331, + "num_tokens": 264629600.0, + "step": 10320 + }, + { + "epoch": 1.1334285086755984, + "grad_norm": 2.3501853942871094, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7090119123458862, + "num_tokens": 264656590.0, + "step": 10321 + }, + { + "epoch": 1.1335383263782122, + "grad_norm": 2.7264556884765625, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7271087169647217, + "num_tokens": 264674050.0, + "step": 10322 + }, + { + "epoch": 1.133648144080826, + "grad_norm": 1.9835635423660278, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7104655504226685, + "num_tokens": 264705263.0, + "step": 10323 + }, + { + "epoch": 1.1337579617834395, + "grad_norm": 2.528618335723877, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7103551626205444, + "num_tokens": 264727730.0, + "step": 10324 + }, + { + "epoch": 1.1338677794860532, + "grad_norm": 2.0853283405303955, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7005720138549805, + "num_tokens": 264761245.0, + "step": 10325 + }, + { + "epoch": 1.1339775971886668, + "grad_norm": 2.318373680114746, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7015650868415833, + "num_tokens": 264787960.0, + "step": 10326 + }, + { + "epoch": 1.1340874148912805, + "grad_norm": 2.3937065601348877, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7100708484649658, + "num_tokens": 264812310.0, + "step": 10327 + }, + { + "epoch": 1.134197232593894, + "grad_norm": 2.403864860534668, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7230746746063232, + "num_tokens": 264834767.0, + "step": 10328 + }, + { + "epoch": 1.1343070502965078, + "grad_norm": 2.1975018978118896, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7005000710487366, + "num_tokens": 264863708.0, + "step": 10329 + }, + { + "epoch": 1.1344168679991213, + "grad_norm": 2.4536139965057373, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7285839319229126, + "num_tokens": 264885483.0, + "step": 10330 + }, + { + "epoch": 1.134526685701735, + "grad_norm": 2.2163619995117188, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.723314642906189, + "num_tokens": 264910348.0, + "step": 10331 + }, + { + "epoch": 1.1346365034043489, + "grad_norm": 2.1414363384246826, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7101543545722961, + "num_tokens": 264938434.0, + "step": 10332 + }, + { + "epoch": 1.1347463211069624, + "grad_norm": 2.3205370903015137, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7242904901504517, + "num_tokens": 264962117.0, + "step": 10333 + }, + { + "epoch": 1.1348561388095761, + "grad_norm": 2.310227394104004, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7030712366104126, + "num_tokens": 264988815.0, + "step": 10334 + }, + { + "epoch": 1.1349659565121897, + "grad_norm": 2.1595985889434814, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7003945708274841, + "num_tokens": 265019791.0, + "step": 10335 + }, + { + "epoch": 1.1350757742148034, + "grad_norm": 2.240262985229492, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7104101181030273, + "num_tokens": 265046330.0, + "step": 10336 + }, + { + "epoch": 1.1351855919174172, + "grad_norm": 2.1048240661621094, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7352165579795837, + "num_tokens": 265074315.0, + "step": 10337 + }, + { + "epoch": 1.1352954096200307, + "grad_norm": 2.255542516708374, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7180458307266235, + "num_tokens": 265101814.0, + "step": 10338 + }, + { + "epoch": 1.1354052273226445, + "grad_norm": 2.580869436264038, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7357858419418335, + "num_tokens": 265121034.0, + "step": 10339 + }, + { + "epoch": 1.135515045025258, + "grad_norm": 2.2625315189361572, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7031593918800354, + "num_tokens": 265146696.0, + "step": 10340 + }, + { + "epoch": 1.1356248627278718, + "grad_norm": 2.0766961574554443, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6952306032180786, + "num_tokens": 265176217.0, + "step": 10341 + }, + { + "epoch": 1.1357346804304853, + "grad_norm": 2.700918674468994, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7327708601951599, + "num_tokens": 265196361.0, + "step": 10342 + }, + { + "epoch": 1.135844498133099, + "grad_norm": 2.0007166862487793, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.717190146446228, + "num_tokens": 265227450.0, + "step": 10343 + }, + { + "epoch": 1.1359543158357126, + "grad_norm": 2.6197597980499268, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.705586314201355, + "num_tokens": 265252120.0, + "step": 10344 + }, + { + "epoch": 1.1360641335383264, + "grad_norm": 2.230736494064331, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.716004490852356, + "num_tokens": 265277633.0, + "step": 10345 + }, + { + "epoch": 1.1361739512409401, + "grad_norm": 2.1854465007781982, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7210360169410706, + "num_tokens": 265303011.0, + "step": 10346 + }, + { + "epoch": 1.1362837689435537, + "grad_norm": 2.3446671962738037, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7303285002708435, + "num_tokens": 265325624.0, + "step": 10347 + }, + { + "epoch": 1.1363935866461674, + "grad_norm": 2.1145637035369873, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6960931420326233, + "num_tokens": 265356414.0, + "step": 10348 + }, + { + "epoch": 1.136503404348781, + "grad_norm": 2.3950483798980713, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.750418484210968, + "num_tokens": 265378137.0, + "step": 10349 + }, + { + "epoch": 1.1366132220513947, + "grad_norm": 2.3935017585754395, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7158696055412292, + "num_tokens": 265401210.0, + "step": 10350 + }, + { + "epoch": 1.1367230397540085, + "grad_norm": 2.4977219104766846, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7248060703277588, + "num_tokens": 265421630.0, + "step": 10351 + }, + { + "epoch": 1.136832857456622, + "grad_norm": 2.365600347518921, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7183372974395752, + "num_tokens": 265445349.0, + "step": 10352 + }, + { + "epoch": 1.1369426751592357, + "grad_norm": 2.2432425022125244, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7150017023086548, + "num_tokens": 265470333.0, + "step": 10353 + }, + { + "epoch": 1.1370524928618493, + "grad_norm": 2.483734369277954, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6976069808006287, + "num_tokens": 265493597.0, + "step": 10354 + }, + { + "epoch": 1.137162310564463, + "grad_norm": 2.1822564601898193, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7378838062286377, + "num_tokens": 265517152.0, + "step": 10355 + }, + { + "epoch": 1.1372721282670766, + "grad_norm": 2.3191092014312744, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7350083589553833, + "num_tokens": 265540149.0, + "step": 10356 + }, + { + "epoch": 1.1373819459696903, + "grad_norm": 1.9673854112625122, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7311591506004333, + "num_tokens": 265570371.0, + "step": 10357 + }, + { + "epoch": 1.1374917636723039, + "grad_norm": 2.2763583660125732, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7154964208602905, + "num_tokens": 265594227.0, + "step": 10358 + }, + { + "epoch": 1.1376015813749176, + "grad_norm": 2.041567087173462, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7110334634780884, + "num_tokens": 265623769.0, + "step": 10359 + }, + { + "epoch": 1.1377113990775314, + "grad_norm": 2.118361234664917, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7095490097999573, + "num_tokens": 265653098.0, + "step": 10360 + }, + { + "epoch": 1.137821216780145, + "grad_norm": 2.278916835784912, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.700141191482544, + "num_tokens": 265679858.0, + "step": 10361 + }, + { + "epoch": 1.1379310344827587, + "grad_norm": 2.4010727405548096, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7387622594833374, + "num_tokens": 265703135.0, + "step": 10362 + }, + { + "epoch": 1.1380408521853722, + "grad_norm": 2.6492397785186768, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7284234166145325, + "num_tokens": 265721633.0, + "step": 10363 + }, + { + "epoch": 1.138150669887986, + "grad_norm": 2.198213815689087, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7207942008972168, + "num_tokens": 265746524.0, + "step": 10364 + }, + { + "epoch": 1.1382604875905997, + "grad_norm": 2.1165552139282227, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7102755308151245, + "num_tokens": 265776271.0, + "step": 10365 + }, + { + "epoch": 1.1383703052932133, + "grad_norm": 2.198230504989624, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7197470664978027, + "num_tokens": 265802694.0, + "step": 10366 + }, + { + "epoch": 1.138480122995827, + "grad_norm": 1.9442058801651, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7175764441490173, + "num_tokens": 265835558.0, + "step": 10367 + }, + { + "epoch": 1.1385899406984406, + "grad_norm": 2.170820951461792, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7124760150909424, + "num_tokens": 265862399.0, + "step": 10368 + }, + { + "epoch": 1.1386997584010543, + "grad_norm": 2.276669502258301, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.708664059638977, + "num_tokens": 265887005.0, + "step": 10369 + }, + { + "epoch": 1.1388095761036678, + "grad_norm": 2.1114232540130615, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7279056310653687, + "num_tokens": 265914410.0, + "step": 10370 + }, + { + "epoch": 1.1389193938062816, + "grad_norm": 2.328003168106079, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7065658569335938, + "num_tokens": 265938505.0, + "step": 10371 + }, + { + "epoch": 1.1390292115088951, + "grad_norm": 2.2302122116088867, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7337123155593872, + "num_tokens": 265964198.0, + "step": 10372 + }, + { + "epoch": 1.139139029211509, + "grad_norm": 2.0646066665649414, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6983734369277954, + "num_tokens": 265993479.0, + "step": 10373 + }, + { + "epoch": 1.1392488469141226, + "grad_norm": 2.1241753101348877, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6916826367378235, + "num_tokens": 266022484.0, + "step": 10374 + }, + { + "epoch": 1.1393586646167362, + "grad_norm": 2.166844606399536, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7159426212310791, + "num_tokens": 266050367.0, + "step": 10375 + }, + { + "epoch": 1.13946848231935, + "grad_norm": 2.3238649368286133, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7221605181694031, + "num_tokens": 266074305.0, + "step": 10376 + }, + { + "epoch": 1.1395783000219635, + "grad_norm": 2.3418996334075928, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7242274284362793, + "num_tokens": 266096775.0, + "step": 10377 + }, + { + "epoch": 1.1396881177245772, + "grad_norm": 2.0493693351745605, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7051211595535278, + "num_tokens": 266127634.0, + "step": 10378 + }, + { + "epoch": 1.1397979354271908, + "grad_norm": 2.248399257659912, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.762441098690033, + "num_tokens": 266152107.0, + "step": 10379 + }, + { + "epoch": 1.1399077531298045, + "grad_norm": 2.475050926208496, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7298803329467773, + "num_tokens": 266174233.0, + "step": 10380 + }, + { + "epoch": 1.1400175708324183, + "grad_norm": 2.127782106399536, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7142846584320068, + "num_tokens": 266201662.0, + "step": 10381 + }, + { + "epoch": 1.1401273885350318, + "grad_norm": 2.2023069858551025, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7148729562759399, + "num_tokens": 266227412.0, + "step": 10382 + }, + { + "epoch": 1.1402372062376456, + "grad_norm": 2.409369707107544, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7139707207679749, + "num_tokens": 266252629.0, + "step": 10383 + }, + { + "epoch": 1.140347023940259, + "grad_norm": 2.0482826232910156, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7048110365867615, + "num_tokens": 266284465.0, + "step": 10384 + }, + { + "epoch": 1.1404568416428729, + "grad_norm": 2.544999361038208, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.726359486579895, + "num_tokens": 266305754.0, + "step": 10385 + }, + { + "epoch": 1.1405666593454864, + "grad_norm": 2.633430004119873, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7445220947265625, + "num_tokens": 266324712.0, + "step": 10386 + }, + { + "epoch": 1.1406764770481002, + "grad_norm": 2.1935484409332275, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7144659757614136, + "num_tokens": 266351185.0, + "step": 10387 + }, + { + "epoch": 1.140786294750714, + "grad_norm": 2.1843433380126953, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7132452726364136, + "num_tokens": 266377280.0, + "step": 10388 + }, + { + "epoch": 1.1408961124533274, + "grad_norm": 2.336603879928589, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7365396022796631, + "num_tokens": 266398857.0, + "step": 10389 + }, + { + "epoch": 1.1410059301559412, + "grad_norm": 2.2372121810913086, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7365825176239014, + "num_tokens": 266423439.0, + "step": 10390 + }, + { + "epoch": 1.1411157478585547, + "grad_norm": 2.0125246047973633, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7064509987831116, + "num_tokens": 266456799.0, + "step": 10391 + }, + { + "epoch": 1.1412255655611685, + "grad_norm": 2.1317057609558105, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.692612886428833, + "num_tokens": 266485009.0, + "step": 10392 + }, + { + "epoch": 1.141335383263782, + "grad_norm": 2.2191388607025146, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6965989470481873, + "num_tokens": 266515531.0, + "step": 10393 + }, + { + "epoch": 1.1414452009663958, + "grad_norm": 2.4310827255249023, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7329559326171875, + "num_tokens": 266537785.0, + "step": 10394 + }, + { + "epoch": 1.1415550186690093, + "grad_norm": 2.478245973587036, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7168052196502686, + "num_tokens": 266560306.0, + "step": 10395 + }, + { + "epoch": 1.141664836371623, + "grad_norm": 2.0818052291870117, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7171808481216431, + "num_tokens": 266590285.0, + "step": 10396 + }, + { + "epoch": 1.1417746540742368, + "grad_norm": 2.470435380935669, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7631404995918274, + "num_tokens": 266611378.0, + "step": 10397 + }, + { + "epoch": 1.1418844717768504, + "grad_norm": 2.1953086853027344, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7271963357925415, + "num_tokens": 266638166.0, + "step": 10398 + }, + { + "epoch": 1.1419942894794641, + "grad_norm": 2.2208499908447266, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7194226384162903, + "num_tokens": 266663461.0, + "step": 10399 + }, + { + "epoch": 1.1421041071820777, + "grad_norm": 2.2941386699676514, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7117632627487183, + "num_tokens": 266692729.0, + "step": 10400 + }, + { + "epoch": 1.1422139248846914, + "grad_norm": 2.5340030193328857, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7131450772285461, + "num_tokens": 266719054.0, + "step": 10401 + }, + { + "epoch": 1.1423237425873052, + "grad_norm": 2.2258331775665283, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6878868937492371, + "num_tokens": 266749023.0, + "step": 10402 + }, + { + "epoch": 1.1424335602899187, + "grad_norm": 2.411107301712036, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7010794878005981, + "num_tokens": 266772417.0, + "step": 10403 + }, + { + "epoch": 1.1425433779925325, + "grad_norm": 2.2880859375, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7063254117965698, + "num_tokens": 266798200.0, + "step": 10404 + }, + { + "epoch": 1.142653195695146, + "grad_norm": 2.377542018890381, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7152787446975708, + "num_tokens": 266820383.0, + "step": 10405 + }, + { + "epoch": 1.1427630133977598, + "grad_norm": 2.1637656688690186, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7254784107208252, + "num_tokens": 266846978.0, + "step": 10406 + }, + { + "epoch": 1.1428728311003733, + "grad_norm": 2.2356505393981934, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7124817371368408, + "num_tokens": 266874871.0, + "step": 10407 + }, + { + "epoch": 1.142982648802987, + "grad_norm": 2.5788097381591797, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.712489664554596, + "num_tokens": 266897201.0, + "step": 10408 + }, + { + "epoch": 1.1430924665056006, + "grad_norm": 2.180304765701294, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7117149829864502, + "num_tokens": 266924145.0, + "step": 10409 + }, + { + "epoch": 1.1432022842082143, + "grad_norm": 2.2777531147003174, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7220567464828491, + "num_tokens": 266949409.0, + "step": 10410 + }, + { + "epoch": 1.143312101910828, + "grad_norm": 2.521000862121582, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7179679870605469, + "num_tokens": 266971414.0, + "step": 10411 + }, + { + "epoch": 1.1434219196134416, + "grad_norm": 2.478808641433716, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7145935893058777, + "num_tokens": 266994903.0, + "step": 10412 + }, + { + "epoch": 1.1435317373160554, + "grad_norm": 2.3380563259124756, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.730291485786438, + "num_tokens": 267018647.0, + "step": 10413 + }, + { + "epoch": 1.143641555018669, + "grad_norm": 2.271345853805542, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7327774167060852, + "num_tokens": 267045245.0, + "step": 10414 + }, + { + "epoch": 1.1437513727212827, + "grad_norm": 2.2876651287078857, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7235033512115479, + "num_tokens": 267069628.0, + "step": 10415 + }, + { + "epoch": 1.1438611904238964, + "grad_norm": 2.279066562652588, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7018772959709167, + "num_tokens": 267097712.0, + "step": 10416 + }, + { + "epoch": 1.14397100812651, + "grad_norm": 2.08174729347229, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7281903028488159, + "num_tokens": 267125354.0, + "step": 10417 + }, + { + "epoch": 1.1440808258291237, + "grad_norm": 2.200308322906494, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6963424682617188, + "num_tokens": 267150947.0, + "step": 10418 + }, + { + "epoch": 1.1441906435317373, + "grad_norm": 2.3639862537384033, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7379498481750488, + "num_tokens": 267173496.0, + "step": 10419 + }, + { + "epoch": 1.144300461234351, + "grad_norm": 2.0367157459259033, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7007601261138916, + "num_tokens": 267204816.0, + "step": 10420 + }, + { + "epoch": 1.1444102789369646, + "grad_norm": 2.3677656650543213, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7235854864120483, + "num_tokens": 267228052.0, + "step": 10421 + }, + { + "epoch": 1.1445200966395783, + "grad_norm": 2.0663325786590576, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7162156105041504, + "num_tokens": 267259068.0, + "step": 10422 + }, + { + "epoch": 1.1446299143421919, + "grad_norm": 2.2844629287719727, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7211258411407471, + "num_tokens": 267284026.0, + "step": 10423 + }, + { + "epoch": 1.1447397320448056, + "grad_norm": 2.210416078567505, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7241469025611877, + "num_tokens": 267311369.0, + "step": 10424 + }, + { + "epoch": 1.1448495497474194, + "grad_norm": 2.495835065841675, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7216746807098389, + "num_tokens": 267333658.0, + "step": 10425 + }, + { + "epoch": 1.144959367450033, + "grad_norm": 2.5268073081970215, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7309708595275879, + "num_tokens": 267356053.0, + "step": 10426 + }, + { + "epoch": 1.1450691851526467, + "grad_norm": 2.3720273971557617, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7111518979072571, + "num_tokens": 267380914.0, + "step": 10427 + }, + { + "epoch": 1.1451790028552602, + "grad_norm": 2.5899507999420166, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7055933475494385, + "num_tokens": 267400314.0, + "step": 10428 + }, + { + "epoch": 1.145288820557874, + "grad_norm": 2.2440109252929688, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7417311072349548, + "num_tokens": 267426373.0, + "step": 10429 + }, + { + "epoch": 1.1453986382604877, + "grad_norm": 2.2817578315734863, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.711764931678772, + "num_tokens": 267452447.0, + "step": 10430 + }, + { + "epoch": 1.1455084559631012, + "grad_norm": 2.214372396469116, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7180677652359009, + "num_tokens": 267478807.0, + "step": 10431 + }, + { + "epoch": 1.145618273665715, + "grad_norm": 2.2639918327331543, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7141199111938477, + "num_tokens": 267504678.0, + "step": 10432 + }, + { + "epoch": 1.1457280913683285, + "grad_norm": 2.458987236022949, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7271190285682678, + "num_tokens": 267525475.0, + "step": 10433 + }, + { + "epoch": 1.1458379090709423, + "grad_norm": 2.4747893810272217, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7346278429031372, + "num_tokens": 267546765.0, + "step": 10434 + }, + { + "epoch": 1.1459477267735558, + "grad_norm": 2.240251064300537, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7232413291931152, + "num_tokens": 267572559.0, + "step": 10435 + }, + { + "epoch": 1.1460575444761696, + "grad_norm": 2.0825555324554443, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7164719700813293, + "num_tokens": 267601198.0, + "step": 10436 + }, + { + "epoch": 1.1461673621787831, + "grad_norm": 2.3713583946228027, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7106871008872986, + "num_tokens": 267625336.0, + "step": 10437 + }, + { + "epoch": 1.1462771798813969, + "grad_norm": 2.413148880004883, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7219989895820618, + "num_tokens": 267648795.0, + "step": 10438 + }, + { + "epoch": 1.1463869975840106, + "grad_norm": 2.429309606552124, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7138886451721191, + "num_tokens": 267671786.0, + "step": 10439 + }, + { + "epoch": 1.1464968152866242, + "grad_norm": 2.375852584838867, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7277445793151855, + "num_tokens": 267694277.0, + "step": 10440 + }, + { + "epoch": 1.146606632989238, + "grad_norm": 2.372687816619873, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7121754884719849, + "num_tokens": 267719470.0, + "step": 10441 + }, + { + "epoch": 1.1467164506918515, + "grad_norm": 2.4842417240142822, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7142468094825745, + "num_tokens": 267743961.0, + "step": 10442 + }, + { + "epoch": 1.1468262683944652, + "grad_norm": 2.372251510620117, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7048820853233337, + "num_tokens": 267767348.0, + "step": 10443 + }, + { + "epoch": 1.1469360860970788, + "grad_norm": 2.143259286880493, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.753400444984436, + "num_tokens": 267793109.0, + "step": 10444 + }, + { + "epoch": 1.1470459037996925, + "grad_norm": 2.122781276702881, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7289871573448181, + "num_tokens": 267818303.0, + "step": 10445 + }, + { + "epoch": 1.147155721502306, + "grad_norm": 1.9738073348999023, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7038161754608154, + "num_tokens": 267851121.0, + "step": 10446 + }, + { + "epoch": 1.1472655392049198, + "grad_norm": 2.0336639881134033, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7275908589363098, + "num_tokens": 267880543.0, + "step": 10447 + }, + { + "epoch": 1.1473753569075336, + "grad_norm": 2.3803741931915283, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7055642008781433, + "num_tokens": 267904550.0, + "step": 10448 + }, + { + "epoch": 1.147485174610147, + "grad_norm": 2.3286685943603516, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7285544276237488, + "num_tokens": 267927635.0, + "step": 10449 + }, + { + "epoch": 1.1475949923127609, + "grad_norm": 2.311683416366577, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7119089365005493, + "num_tokens": 267952274.0, + "step": 10450 + }, + { + "epoch": 1.1477048100153744, + "grad_norm": 2.2045705318450928, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7197949290275574, + "num_tokens": 267978675.0, + "step": 10451 + }, + { + "epoch": 1.1478146277179881, + "grad_norm": 2.1118080615997314, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.6991543173789978, + "num_tokens": 268006944.0, + "step": 10452 + }, + { + "epoch": 1.147924445420602, + "grad_norm": 2.657512664794922, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7192331552505493, + "num_tokens": 268027023.0, + "step": 10453 + }, + { + "epoch": 1.1480342631232154, + "grad_norm": 2.400655746459961, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7299736738204956, + "num_tokens": 268051534.0, + "step": 10454 + }, + { + "epoch": 1.1481440808258292, + "grad_norm": 2.4205379486083984, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7356142401695251, + "num_tokens": 268074302.0, + "step": 10455 + }, + { + "epoch": 1.1482538985284427, + "grad_norm": 2.1900711059570312, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7251712679862976, + "num_tokens": 268100026.0, + "step": 10456 + }, + { + "epoch": 1.1483637162310565, + "grad_norm": 2.1479291915893555, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7070032358169556, + "num_tokens": 268130377.0, + "step": 10457 + }, + { + "epoch": 1.14847353393367, + "grad_norm": 2.3864548206329346, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6959975957870483, + "num_tokens": 268155058.0, + "step": 10458 + }, + { + "epoch": 1.1485833516362838, + "grad_norm": 2.1616008281707764, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6950732469558716, + "num_tokens": 268184011.0, + "step": 10459 + }, + { + "epoch": 1.1486931693388973, + "grad_norm": 2.4200735092163086, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7285767793655396, + "num_tokens": 268207252.0, + "step": 10460 + }, + { + "epoch": 1.148802987041511, + "grad_norm": 2.691884756088257, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7257257103919983, + "num_tokens": 268227105.0, + "step": 10461 + }, + { + "epoch": 1.1489128047441248, + "grad_norm": 2.3013482093811035, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7251006364822388, + "num_tokens": 268249801.0, + "step": 10462 + }, + { + "epoch": 1.1490226224467384, + "grad_norm": 2.2640774250030518, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7393143773078918, + "num_tokens": 268273552.0, + "step": 10463 + }, + { + "epoch": 1.1491324401493521, + "grad_norm": 2.367767572402954, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7118873596191406, + "num_tokens": 268296256.0, + "step": 10464 + }, + { + "epoch": 1.1492422578519657, + "grad_norm": 2.46641206741333, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7405648231506348, + "num_tokens": 268318241.0, + "step": 10465 + }, + { + "epoch": 1.1493520755545794, + "grad_norm": 2.1013882160186768, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7165408134460449, + "num_tokens": 268346236.0, + "step": 10466 + }, + { + "epoch": 1.1494618932571932, + "grad_norm": 2.3053205013275146, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7322381734848022, + "num_tokens": 268372357.0, + "step": 10467 + }, + { + "epoch": 1.1495717109598067, + "grad_norm": 2.2781388759613037, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7372094392776489, + "num_tokens": 268395357.0, + "step": 10468 + }, + { + "epoch": 1.1496815286624205, + "grad_norm": 2.3579108715057373, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7545838952064514, + "num_tokens": 268418521.0, + "step": 10469 + }, + { + "epoch": 1.149791346365034, + "grad_norm": 2.2522659301757812, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7359544038772583, + "num_tokens": 268442246.0, + "step": 10470 + }, + { + "epoch": 1.1499011640676478, + "grad_norm": 2.341876268386841, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7207553386688232, + "num_tokens": 268466353.0, + "step": 10471 + }, + { + "epoch": 1.1500109817702613, + "grad_norm": 2.3131988048553467, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7258867621421814, + "num_tokens": 268489618.0, + "step": 10472 + }, + { + "epoch": 1.150120799472875, + "grad_norm": 2.3155946731567383, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7235965728759766, + "num_tokens": 268514189.0, + "step": 10473 + }, + { + "epoch": 1.1502306171754886, + "grad_norm": 2.0466675758361816, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7176710367202759, + "num_tokens": 268544940.0, + "step": 10474 + }, + { + "epoch": 1.1503404348781023, + "grad_norm": 2.312425136566162, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7220961451530457, + "num_tokens": 268570861.0, + "step": 10475 + }, + { + "epoch": 1.150450252580716, + "grad_norm": 2.1683192253112793, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.718683123588562, + "num_tokens": 268597355.0, + "step": 10476 + }, + { + "epoch": 1.1505600702833296, + "grad_norm": 2.168771505355835, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6877126693725586, + "num_tokens": 268624872.0, + "step": 10477 + }, + { + "epoch": 1.1506698879859434, + "grad_norm": 2.1393253803253174, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.736990749835968, + "num_tokens": 268652083.0, + "step": 10478 + }, + { + "epoch": 1.150779705688557, + "grad_norm": 2.0303595066070557, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7212942838668823, + "num_tokens": 268681892.0, + "step": 10479 + }, + { + "epoch": 1.1508895233911707, + "grad_norm": 2.199742555618286, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7097539901733398, + "num_tokens": 268708335.0, + "step": 10480 + }, + { + "epoch": 1.1509993410937844, + "grad_norm": 2.111337184906006, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7105540633201599, + "num_tokens": 268738909.0, + "step": 10481 + }, + { + "epoch": 1.151109158796398, + "grad_norm": 2.2690160274505615, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7192548513412476, + "num_tokens": 268762841.0, + "step": 10482 + }, + { + "epoch": 1.1512189764990117, + "grad_norm": 2.343461275100708, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7330913543701172, + "num_tokens": 268786307.0, + "step": 10483 + }, + { + "epoch": 1.1513287942016253, + "grad_norm": 2.287526845932007, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7161146402359009, + "num_tokens": 268810581.0, + "step": 10484 + }, + { + "epoch": 1.151438611904239, + "grad_norm": 2.1122119426727295, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7440241575241089, + "num_tokens": 268836557.0, + "step": 10485 + }, + { + "epoch": 1.1515484296068526, + "grad_norm": 2.0034759044647217, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7046200037002563, + "num_tokens": 268868877.0, + "step": 10486 + }, + { + "epoch": 1.1516582473094663, + "grad_norm": 2.682311534881592, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7263726592063904, + "num_tokens": 268888197.0, + "step": 10487 + }, + { + "epoch": 1.1517680650120798, + "grad_norm": 2.333019495010376, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.7007240056991577, + "num_tokens": 268913168.0, + "step": 10488 + }, + { + "epoch": 1.1518778827146936, + "grad_norm": 2.0939018726348877, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7250195741653442, + "num_tokens": 268940155.0, + "step": 10489 + }, + { + "epoch": 1.1519877004173074, + "grad_norm": 2.2489984035491943, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6931179761886597, + "num_tokens": 268966783.0, + "step": 10490 + }, + { + "epoch": 1.152097518119921, + "grad_norm": 2.5254836082458496, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7364885807037354, + "num_tokens": 268986707.0, + "step": 10491 + }, + { + "epoch": 1.1522073358225347, + "grad_norm": 2.1252782344818115, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7099659442901611, + "num_tokens": 269014423.0, + "step": 10492 + }, + { + "epoch": 1.1523171535251482, + "grad_norm": 2.3912525177001953, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7149308919906616, + "num_tokens": 269037669.0, + "step": 10493 + }, + { + "epoch": 1.152426971227762, + "grad_norm": 2.272170305252075, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7004218101501465, + "num_tokens": 269062797.0, + "step": 10494 + }, + { + "epoch": 1.1525367889303757, + "grad_norm": 2.2187182903289795, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7147671580314636, + "num_tokens": 269087503.0, + "step": 10495 + }, + { + "epoch": 1.1526466066329892, + "grad_norm": 2.1358730792999268, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7083138227462769, + "num_tokens": 269118452.0, + "step": 10496 + }, + { + "epoch": 1.152756424335603, + "grad_norm": 2.2637593746185303, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7128949165344238, + "num_tokens": 269142236.0, + "step": 10497 + }, + { + "epoch": 1.1528662420382165, + "grad_norm": 2.277498960494995, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7124066352844238, + "num_tokens": 269166982.0, + "step": 10498 + }, + { + "epoch": 1.1529760597408303, + "grad_norm": 2.426755905151367, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7222532033920288, + "num_tokens": 269190386.0, + "step": 10499 + }, + { + "epoch": 1.1530858774434438, + "grad_norm": 2.2619380950927734, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.719218909740448, + "num_tokens": 269214581.0, + "step": 10500 + }, + { + "epoch": 1.1531956951460576, + "grad_norm": 2.2816121578216553, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7110919952392578, + "num_tokens": 269240069.0, + "step": 10501 + }, + { + "epoch": 1.153305512848671, + "grad_norm": 2.138533592224121, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7155020236968994, + "num_tokens": 269267775.0, + "step": 10502 + }, + { + "epoch": 1.1534153305512849, + "grad_norm": 2.0793440341949463, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7076547741889954, + "num_tokens": 269297006.0, + "step": 10503 + }, + { + "epoch": 1.1535251482538986, + "grad_norm": 2.4022135734558105, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7174695730209351, + "num_tokens": 269323130.0, + "step": 10504 + }, + { + "epoch": 1.1536349659565122, + "grad_norm": 2.1629645824432373, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7157151103019714, + "num_tokens": 269349667.0, + "step": 10505 + }, + { + "epoch": 1.153744783659126, + "grad_norm": 2.3023219108581543, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7436568737030029, + "num_tokens": 269372326.0, + "step": 10506 + }, + { + "epoch": 1.1538546013617395, + "grad_norm": 2.349738121032715, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7118059396743774, + "num_tokens": 269395129.0, + "step": 10507 + }, + { + "epoch": 1.1539644190643532, + "grad_norm": 2.272268772125244, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7192481756210327, + "num_tokens": 269419474.0, + "step": 10508 + }, + { + "epoch": 1.1540742367669667, + "grad_norm": 2.1303811073303223, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7203544974327087, + "num_tokens": 269445992.0, + "step": 10509 + }, + { + "epoch": 1.1541840544695805, + "grad_norm": 2.5469181537628174, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7208125591278076, + "num_tokens": 269466307.0, + "step": 10510 + }, + { + "epoch": 1.154293872172194, + "grad_norm": 2.110157012939453, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7321836948394775, + "num_tokens": 269494248.0, + "step": 10511 + }, + { + "epoch": 1.1544036898748078, + "grad_norm": 2.337923526763916, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7125807404518127, + "num_tokens": 269518686.0, + "step": 10512 + }, + { + "epoch": 1.1545135075774215, + "grad_norm": 1.9965234994888306, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6921822428703308, + "num_tokens": 269550687.0, + "step": 10513 + }, + { + "epoch": 1.154623325280035, + "grad_norm": 2.565084457397461, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7357485890388489, + "num_tokens": 269572024.0, + "step": 10514 + }, + { + "epoch": 1.1547331429826488, + "grad_norm": 2.1891331672668457, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7160409688949585, + "num_tokens": 269598638.0, + "step": 10515 + }, + { + "epoch": 1.1548429606852624, + "grad_norm": 2.0222716331481934, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7123024463653564, + "num_tokens": 269629496.0, + "step": 10516 + }, + { + "epoch": 1.1549527783878761, + "grad_norm": 2.174391269683838, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7275335788726807, + "num_tokens": 269656269.0, + "step": 10517 + }, + { + "epoch": 1.15506259609049, + "grad_norm": 2.22772479057312, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7264285087585449, + "num_tokens": 269680532.0, + "step": 10518 + }, + { + "epoch": 1.1551724137931034, + "grad_norm": 2.1136951446533203, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.7013781666755676, + "num_tokens": 269712766.0, + "step": 10519 + }, + { + "epoch": 1.1552822314957172, + "grad_norm": 2.273984909057617, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7522488832473755, + "num_tokens": 269737049.0, + "step": 10520 + }, + { + "epoch": 1.1553920491983307, + "grad_norm": 2.5560269355773926, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7284091114997864, + "num_tokens": 269756839.0, + "step": 10521 + }, + { + "epoch": 1.1555018669009445, + "grad_norm": 2.1469199657440186, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7346100807189941, + "num_tokens": 269783588.0, + "step": 10522 + }, + { + "epoch": 1.155611684603558, + "grad_norm": 2.0716938972473145, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7121261358261108, + "num_tokens": 269813501.0, + "step": 10523 + }, + { + "epoch": 1.1557215023061718, + "grad_norm": 2.2894368171691895, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7111952304840088, + "num_tokens": 269838763.0, + "step": 10524 + }, + { + "epoch": 1.1558313200087853, + "grad_norm": 2.328418731689453, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7168315649032593, + "num_tokens": 269862949.0, + "step": 10525 + }, + { + "epoch": 1.155941137711399, + "grad_norm": 2.115553855895996, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7148923873901367, + "num_tokens": 269892567.0, + "step": 10526 + }, + { + "epoch": 1.1560509554140128, + "grad_norm": 2.3674492835998535, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.6991485357284546, + "num_tokens": 269915391.0, + "step": 10527 + }, + { + "epoch": 1.1561607731166264, + "grad_norm": 2.550657033920288, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7213811874389648, + "num_tokens": 269936841.0, + "step": 10528 + }, + { + "epoch": 1.15627059081924, + "grad_norm": 2.3183677196502686, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.704603910446167, + "num_tokens": 269960657.0, + "step": 10529 + }, + { + "epoch": 1.1563804085218536, + "grad_norm": 2.372997999191284, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7251073122024536, + "num_tokens": 269983409.0, + "step": 10530 + }, + { + "epoch": 1.1564902262244674, + "grad_norm": 2.401839017868042, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7207563519477844, + "num_tokens": 270006515.0, + "step": 10531 + }, + { + "epoch": 1.1566000439270812, + "grad_norm": 2.327871322631836, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.702903151512146, + "num_tokens": 270031107.0, + "step": 10532 + }, + { + "epoch": 1.1567098616296947, + "grad_norm": 2.330962896347046, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7358630895614624, + "num_tokens": 270054197.0, + "step": 10533 + }, + { + "epoch": 1.1568196793323084, + "grad_norm": 2.097357988357544, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7048172354698181, + "num_tokens": 270082695.0, + "step": 10534 + }, + { + "epoch": 1.156929497034922, + "grad_norm": 2.259256362915039, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7390273809432983, + "num_tokens": 270106098.0, + "step": 10535 + }, + { + "epoch": 1.1570393147375357, + "grad_norm": 2.1767194271087646, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.707207977771759, + "num_tokens": 270134756.0, + "step": 10536 + }, + { + "epoch": 1.1571491324401493, + "grad_norm": 2.254173755645752, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7103621959686279, + "num_tokens": 270160926.0, + "step": 10537 + }, + { + "epoch": 1.157258950142763, + "grad_norm": 2.2752552032470703, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7522963285446167, + "num_tokens": 270184163.0, + "step": 10538 + }, + { + "epoch": 1.1573687678453766, + "grad_norm": 2.412203550338745, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.726702868938446, + "num_tokens": 270208160.0, + "step": 10539 + }, + { + "epoch": 1.1574785855479903, + "grad_norm": 2.4642906188964844, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7400146126747131, + "num_tokens": 270229702.0, + "step": 10540 + }, + { + "epoch": 1.157588403250604, + "grad_norm": 2.4711296558380127, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7403592467308044, + "num_tokens": 270250417.0, + "step": 10541 + }, + { + "epoch": 1.1576982209532176, + "grad_norm": 2.325563669204712, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7183559536933899, + "num_tokens": 270273970.0, + "step": 10542 + }, + { + "epoch": 1.1578080386558314, + "grad_norm": 2.4424145221710205, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7334849834442139, + "num_tokens": 270298460.0, + "step": 10543 + }, + { + "epoch": 1.157917856358445, + "grad_norm": 2.090040445327759, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7087136507034302, + "num_tokens": 270328634.0, + "step": 10544 + }, + { + "epoch": 1.1580276740610587, + "grad_norm": 2.1191465854644775, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7006304264068604, + "num_tokens": 270356099.0, + "step": 10545 + }, + { + "epoch": 1.1581374917636724, + "grad_norm": 2.100555896759033, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7471736669540405, + "num_tokens": 270382719.0, + "step": 10546 + }, + { + "epoch": 1.158247309466286, + "grad_norm": 2.160284996032715, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7198022603988647, + "num_tokens": 270410819.0, + "step": 10547 + }, + { + "epoch": 1.1583571271688997, + "grad_norm": 2.2878177165985107, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7099900245666504, + "num_tokens": 270435556.0, + "step": 10548 + }, + { + "epoch": 1.1584669448715132, + "grad_norm": 2.1587564945220947, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.721961498260498, + "num_tokens": 270463176.0, + "step": 10549 + }, + { + "epoch": 1.158576762574127, + "grad_norm": 2.2583584785461426, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7183641791343689, + "num_tokens": 270489043.0, + "step": 10550 + }, + { + "epoch": 1.1586865802767405, + "grad_norm": 2.366478443145752, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7271679639816284, + "num_tokens": 270513870.0, + "step": 10551 + }, + { + "epoch": 1.1587963979793543, + "grad_norm": 2.326023817062378, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.741916298866272, + "num_tokens": 270537327.0, + "step": 10552 + }, + { + "epoch": 1.1589062156819678, + "grad_norm": 2.2047665119171143, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7143973112106323, + "num_tokens": 270564642.0, + "step": 10553 + }, + { + "epoch": 1.1590160333845816, + "grad_norm": 2.213078737258911, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7208260297775269, + "num_tokens": 270589923.0, + "step": 10554 + }, + { + "epoch": 1.1591258510871953, + "grad_norm": 2.5509462356567383, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7382392883300781, + "num_tokens": 270609289.0, + "step": 10555 + }, + { + "epoch": 1.1592356687898089, + "grad_norm": 2.1194090843200684, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7294692993164062, + "num_tokens": 270637299.0, + "step": 10556 + }, + { + "epoch": 1.1593454864924226, + "grad_norm": 2.327087163925171, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7123448252677917, + "num_tokens": 270661894.0, + "step": 10557 + }, + { + "epoch": 1.1594553041950362, + "grad_norm": 2.105039596557617, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6999741792678833, + "num_tokens": 270689547.0, + "step": 10558 + }, + { + "epoch": 1.15956512189765, + "grad_norm": 2.4134981632232666, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.744879961013794, + "num_tokens": 270712477.0, + "step": 10559 + }, + { + "epoch": 1.1596749396002635, + "grad_norm": 2.36002254486084, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7215486764907837, + "num_tokens": 270737223.0, + "step": 10560 + }, + { + "epoch": 1.1597847573028772, + "grad_norm": 2.4853639602661133, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.743324875831604, + "num_tokens": 270759344.0, + "step": 10561 + }, + { + "epoch": 1.159894575005491, + "grad_norm": 2.388707399368286, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7434086799621582, + "num_tokens": 270782537.0, + "step": 10562 + }, + { + "epoch": 1.1600043927081045, + "grad_norm": 2.371558427810669, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7229034900665283, + "num_tokens": 270805988.0, + "step": 10563 + }, + { + "epoch": 1.1601142104107183, + "grad_norm": 2.2445716857910156, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7049854397773743, + "num_tokens": 270831894.0, + "step": 10564 + }, + { + "epoch": 1.1602240281133318, + "grad_norm": 2.2002456188201904, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7196788787841797, + "num_tokens": 270859123.0, + "step": 10565 + }, + { + "epoch": 1.1603338458159456, + "grad_norm": 2.2004568576812744, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6939135789871216, + "num_tokens": 270887038.0, + "step": 10566 + }, + { + "epoch": 1.160443663518559, + "grad_norm": 2.218780040740967, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.721710205078125, + "num_tokens": 270915910.0, + "step": 10567 + }, + { + "epoch": 1.1605534812211729, + "grad_norm": 2.5865395069122314, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7178324460983276, + "num_tokens": 270937235.0, + "step": 10568 + }, + { + "epoch": 1.1606632989237866, + "grad_norm": 2.174347162246704, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7189809083938599, + "num_tokens": 270964245.0, + "step": 10569 + }, + { + "epoch": 1.1607731166264001, + "grad_norm": 2.5211074352264404, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7135874629020691, + "num_tokens": 270983580.0, + "step": 10570 + }, + { + "epoch": 1.160882934329014, + "grad_norm": 2.1147446632385254, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7466155886650085, + "num_tokens": 271010500.0, + "step": 10571 + }, + { + "epoch": 1.1609927520316274, + "grad_norm": 2.2786550521850586, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6947685480117798, + "num_tokens": 271035047.0, + "step": 10572 + }, + { + "epoch": 1.1611025697342412, + "grad_norm": 2.19486927986145, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7213600277900696, + "num_tokens": 271061331.0, + "step": 10573 + }, + { + "epoch": 1.1612123874368547, + "grad_norm": 2.359104871749878, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7457277774810791, + "num_tokens": 271083899.0, + "step": 10574 + }, + { + "epoch": 1.1613222051394685, + "grad_norm": 2.153468370437622, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.72954261302948, + "num_tokens": 271110954.0, + "step": 10575 + }, + { + "epoch": 1.161432022842082, + "grad_norm": 1.956050992012024, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7126224040985107, + "num_tokens": 271144649.0, + "step": 10576 + }, + { + "epoch": 1.1615418405446958, + "grad_norm": 2.289785146713257, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7419599294662476, + "num_tokens": 271167639.0, + "step": 10577 + }, + { + "epoch": 1.1616516582473095, + "grad_norm": 2.423403739929199, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.744183361530304, + "num_tokens": 271189680.0, + "step": 10578 + }, + { + "epoch": 1.161761475949923, + "grad_norm": 2.071906328201294, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7045156955718994, + "num_tokens": 271220095.0, + "step": 10579 + }, + { + "epoch": 1.1618712936525368, + "grad_norm": 2.2454872131347656, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7140718698501587, + "num_tokens": 271247393.0, + "step": 10580 + }, + { + "epoch": 1.1619811113551504, + "grad_norm": 2.0354790687561035, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7092256546020508, + "num_tokens": 271278256.0, + "step": 10581 + }, + { + "epoch": 1.1620909290577641, + "grad_norm": 2.1690590381622314, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7220715284347534, + "num_tokens": 271303130.0, + "step": 10582 + }, + { + "epoch": 1.1622007467603779, + "grad_norm": 2.5818090438842773, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7501195073127747, + "num_tokens": 271321431.0, + "step": 10583 + }, + { + "epoch": 1.1623105644629914, + "grad_norm": 2.5016322135925293, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7273617386817932, + "num_tokens": 271343465.0, + "step": 10584 + }, + { + "epoch": 1.1624203821656052, + "grad_norm": 2.4903836250305176, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7390211820602417, + "num_tokens": 271364825.0, + "step": 10585 + }, + { + "epoch": 1.1625301998682187, + "grad_norm": 2.2842681407928467, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.690211296081543, + "num_tokens": 271392833.0, + "step": 10586 + }, + { + "epoch": 1.1626400175708325, + "grad_norm": 2.2750821113586426, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.736099123954773, + "num_tokens": 271415456.0, + "step": 10587 + }, + { + "epoch": 1.162749835273446, + "grad_norm": 2.27587628364563, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7117850184440613, + "num_tokens": 271442538.0, + "step": 10588 + }, + { + "epoch": 1.1628596529760598, + "grad_norm": 2.062223196029663, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7287466526031494, + "num_tokens": 271471751.0, + "step": 10589 + }, + { + "epoch": 1.1629694706786733, + "grad_norm": 2.3182709217071533, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7014955282211304, + "num_tokens": 271496417.0, + "step": 10590 + }, + { + "epoch": 1.163079288381287, + "grad_norm": 2.2465710639953613, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7090132832527161, + "num_tokens": 271521361.0, + "step": 10591 + }, + { + "epoch": 1.1631891060839008, + "grad_norm": 2.2072088718414307, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7194603681564331, + "num_tokens": 271547076.0, + "step": 10592 + }, + { + "epoch": 1.1632989237865143, + "grad_norm": 2.1451027393341064, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7203690409660339, + "num_tokens": 271574117.0, + "step": 10593 + }, + { + "epoch": 1.163408741489128, + "grad_norm": 2.2325279712677, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.730559229850769, + "num_tokens": 271598417.0, + "step": 10594 + }, + { + "epoch": 1.1635185591917416, + "grad_norm": 2.3594093322753906, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7352471351623535, + "num_tokens": 271620124.0, + "step": 10595 + }, + { + "epoch": 1.1636283768943554, + "grad_norm": 2.2737135887145996, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7044613361358643, + "num_tokens": 271644553.0, + "step": 10596 + }, + { + "epoch": 1.1637381945969691, + "grad_norm": 2.6951546669006348, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7452268004417419, + "num_tokens": 271663622.0, + "step": 10597 + }, + { + "epoch": 1.1638480122995827, + "grad_norm": 2.428283452987671, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7134625911712646, + "num_tokens": 271685062.0, + "step": 10598 + }, + { + "epoch": 1.1639578300021964, + "grad_norm": 2.391171455383301, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7169330716133118, + "num_tokens": 271708431.0, + "step": 10599 + }, + { + "epoch": 1.16406764770481, + "grad_norm": 2.4638001918792725, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7347690463066101, + "num_tokens": 271730395.0, + "step": 10600 + }, + { + "epoch": 1.1641774654074237, + "grad_norm": 2.4730260372161865, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7064937353134155, + "num_tokens": 271752676.0, + "step": 10601 + }, + { + "epoch": 1.1642872831100373, + "grad_norm": 2.2546515464782715, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.743545651435852, + "num_tokens": 271778350.0, + "step": 10602 + }, + { + "epoch": 1.164397100812651, + "grad_norm": 2.104469060897827, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7071533203125, + "num_tokens": 271807629.0, + "step": 10603 + }, + { + "epoch": 1.1645069185152646, + "grad_norm": 2.167541980743408, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7201834917068481, + "num_tokens": 271835815.0, + "step": 10604 + }, + { + "epoch": 1.1646167362178783, + "grad_norm": 2.1925415992736816, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.725960373878479, + "num_tokens": 271861403.0, + "step": 10605 + }, + { + "epoch": 1.164726553920492, + "grad_norm": 1.9776136875152588, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7494412660598755, + "num_tokens": 271888178.0, + "step": 10606 + }, + { + "epoch": 1.1648363716231056, + "grad_norm": 2.520066022872925, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7414882183074951, + "num_tokens": 271909647.0, + "step": 10607 + }, + { + "epoch": 1.1649461893257194, + "grad_norm": 2.1716151237487793, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6878313422203064, + "num_tokens": 271937573.0, + "step": 10608 + }, + { + "epoch": 1.165056007028333, + "grad_norm": 2.236370801925659, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7048810720443726, + "num_tokens": 271962580.0, + "step": 10609 + }, + { + "epoch": 1.1651658247309467, + "grad_norm": 2.385937452316284, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7262080907821655, + "num_tokens": 271984299.0, + "step": 10610 + }, + { + "epoch": 1.1652756424335604, + "grad_norm": 2.5377197265625, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7462203502655029, + "num_tokens": 272005046.0, + "step": 10611 + }, + { + "epoch": 1.165385460136174, + "grad_norm": 2.346827268600464, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7196030020713806, + "num_tokens": 272029871.0, + "step": 10612 + }, + { + "epoch": 1.1654952778387877, + "grad_norm": 2.587202310562134, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7398282885551453, + "num_tokens": 272049585.0, + "step": 10613 + }, + { + "epoch": 1.1656050955414012, + "grad_norm": 2.1518959999084473, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7354713082313538, + "num_tokens": 272075936.0, + "step": 10614 + }, + { + "epoch": 1.165714913244015, + "grad_norm": 2.409804582595825, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7278963923454285, + "num_tokens": 272098779.0, + "step": 10615 + }, + { + "epoch": 1.1658247309466285, + "grad_norm": 2.325861692428589, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7398606538772583, + "num_tokens": 272122248.0, + "step": 10616 + }, + { + "epoch": 1.1659345486492423, + "grad_norm": 2.4741742610931396, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7342867851257324, + "num_tokens": 272143188.0, + "step": 10617 + }, + { + "epoch": 1.1660443663518558, + "grad_norm": 2.512437105178833, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7315598130226135, + "num_tokens": 272163963.0, + "step": 10618 + }, + { + "epoch": 1.1661541840544696, + "grad_norm": 2.206796646118164, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.741612434387207, + "num_tokens": 272190778.0, + "step": 10619 + }, + { + "epoch": 1.1662640017570833, + "grad_norm": 2.5611398220062256, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7063496708869934, + "num_tokens": 272214825.0, + "step": 10620 + }, + { + "epoch": 1.1663738194596969, + "grad_norm": 2.2595345973968506, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7191221714019775, + "num_tokens": 272240462.0, + "step": 10621 + }, + { + "epoch": 1.1664836371623106, + "grad_norm": 2.0637922286987305, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7299009561538696, + "num_tokens": 272268818.0, + "step": 10622 + }, + { + "epoch": 1.1665934548649242, + "grad_norm": 2.3206982612609863, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7365521192550659, + "num_tokens": 272292266.0, + "step": 10623 + }, + { + "epoch": 1.166703272567538, + "grad_norm": 2.411349058151245, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7203869223594666, + "num_tokens": 272317612.0, + "step": 10624 + }, + { + "epoch": 1.1668130902701515, + "grad_norm": 2.330131769180298, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7305823564529419, + "num_tokens": 272341167.0, + "step": 10625 + }, + { + "epoch": 1.1669229079727652, + "grad_norm": 2.261153221130371, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6960370540618896, + "num_tokens": 272367115.0, + "step": 10626 + }, + { + "epoch": 1.167032725675379, + "grad_norm": 2.182359457015991, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7090394496917725, + "num_tokens": 272395402.0, + "step": 10627 + }, + { + "epoch": 1.1671425433779925, + "grad_norm": 2.0377631187438965, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.723313570022583, + "num_tokens": 272425293.0, + "step": 10628 + }, + { + "epoch": 1.1672523610806063, + "grad_norm": 1.9544942378997803, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7111223936080933, + "num_tokens": 272456838.0, + "step": 10629 + }, + { + "epoch": 1.1673621787832198, + "grad_norm": 2.697901725769043, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7181694507598877, + "num_tokens": 272476845.0, + "step": 10630 + }, + { + "epoch": 1.1674719964858336, + "grad_norm": 2.2310597896575928, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7072079181671143, + "num_tokens": 272503562.0, + "step": 10631 + }, + { + "epoch": 1.167581814188447, + "grad_norm": 2.183924674987793, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7255766987800598, + "num_tokens": 272529842.0, + "step": 10632 + }, + { + "epoch": 1.1676916318910608, + "grad_norm": 2.078718423843384, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7182604670524597, + "num_tokens": 272557899.0, + "step": 10633 + }, + { + "epoch": 1.1678014495936746, + "grad_norm": 2.3327693939208984, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7288690209388733, + "num_tokens": 272580960.0, + "step": 10634 + }, + { + "epoch": 1.1679112672962881, + "grad_norm": 2.452153444290161, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7205663919448853, + "num_tokens": 272602535.0, + "step": 10635 + }, + { + "epoch": 1.168021084998902, + "grad_norm": 2.2351865768432617, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.703265905380249, + "num_tokens": 272629767.0, + "step": 10636 + }, + { + "epoch": 1.1681309027015154, + "grad_norm": 2.0669851303100586, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7206365466117859, + "num_tokens": 272657997.0, + "step": 10637 + }, + { + "epoch": 1.1682407204041292, + "grad_norm": 2.075428009033203, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7142855525016785, + "num_tokens": 272685880.0, + "step": 10638 + }, + { + "epoch": 1.1683505381067427, + "grad_norm": 2.449777126312256, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7430485486984253, + "num_tokens": 272706544.0, + "step": 10639 + }, + { + "epoch": 1.1684603558093565, + "grad_norm": 2.3288466930389404, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7028853893280029, + "num_tokens": 272731398.0, + "step": 10640 + }, + { + "epoch": 1.16857017351197, + "grad_norm": 2.3146073818206787, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7089400291442871, + "num_tokens": 272757761.0, + "step": 10641 + }, + { + "epoch": 1.1686799912145838, + "grad_norm": 2.653829574584961, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7167624831199646, + "num_tokens": 272776737.0, + "step": 10642 + }, + { + "epoch": 1.1687898089171975, + "grad_norm": 2.08821439743042, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7099325060844421, + "num_tokens": 272807596.0, + "step": 10643 + }, + { + "epoch": 1.168899626619811, + "grad_norm": 2.234157085418701, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7252634763717651, + "num_tokens": 272831673.0, + "step": 10644 + }, + { + "epoch": 1.1690094443224248, + "grad_norm": 2.1881296634674072, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6838080883026123, + "num_tokens": 272862143.0, + "step": 10645 + }, + { + "epoch": 1.1691192620250384, + "grad_norm": 2.4937002658843994, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7228745818138123, + "num_tokens": 272882913.0, + "step": 10646 + }, + { + "epoch": 1.169229079727652, + "grad_norm": 1.9375218152999878, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6936769485473633, + "num_tokens": 272916771.0, + "step": 10647 + }, + { + "epoch": 1.1693388974302659, + "grad_norm": 2.2456538677215576, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7222775816917419, + "num_tokens": 272942180.0, + "step": 10648 + }, + { + "epoch": 1.1694487151328794, + "grad_norm": 2.412078857421875, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7147965431213379, + "num_tokens": 272965908.0, + "step": 10649 + }, + { + "epoch": 1.1695585328354932, + "grad_norm": 2.613961935043335, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7541897892951965, + "num_tokens": 272985478.0, + "step": 10650 + }, + { + "epoch": 1.1696683505381067, + "grad_norm": 2.0995101928710938, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.713530421257019, + "num_tokens": 273015893.0, + "step": 10651 + }, + { + "epoch": 1.1697781682407205, + "grad_norm": 2.0367586612701416, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7288357019424438, + "num_tokens": 273046307.0, + "step": 10652 + }, + { + "epoch": 1.169887985943334, + "grad_norm": 1.8676286935806274, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.731250524520874, + "num_tokens": 273080205.0, + "step": 10653 + }, + { + "epoch": 1.1699978036459477, + "grad_norm": 2.2597413063049316, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7174189686775208, + "num_tokens": 273107198.0, + "step": 10654 + }, + { + "epoch": 1.1701076213485613, + "grad_norm": 2.4389851093292236, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7397317886352539, + "num_tokens": 273129183.0, + "step": 10655 + }, + { + "epoch": 1.170217439051175, + "grad_norm": 2.178593873977661, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7164981365203857, + "num_tokens": 273156475.0, + "step": 10656 + }, + { + "epoch": 1.1703272567537888, + "grad_norm": 2.359970808029175, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7281311750411987, + "num_tokens": 273180567.0, + "step": 10657 + }, + { + "epoch": 1.1704370744564023, + "grad_norm": 2.360964775085449, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7349770069122314, + "num_tokens": 273202797.0, + "step": 10658 + }, + { + "epoch": 1.170546892159016, + "grad_norm": 2.6424434185028076, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7008458971977234, + "num_tokens": 273225205.0, + "step": 10659 + }, + { + "epoch": 1.1706567098616296, + "grad_norm": 2.527453899383545, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7280448079109192, + "num_tokens": 273246212.0, + "step": 10660 + }, + { + "epoch": 1.1707665275642434, + "grad_norm": 2.2340619564056396, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7280009984970093, + "num_tokens": 273271489.0, + "step": 10661 + }, + { + "epoch": 1.1708763452668571, + "grad_norm": 2.1884920597076416, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.724040687084198, + "num_tokens": 273296898.0, + "step": 10662 + }, + { + "epoch": 1.1709861629694707, + "grad_norm": 2.22330379486084, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7124395966529846, + "num_tokens": 273322128.0, + "step": 10663 + }, + { + "epoch": 1.1710959806720844, + "grad_norm": 2.7242043018341064, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7236106991767883, + "num_tokens": 273341820.0, + "step": 10664 + }, + { + "epoch": 1.171205798374698, + "grad_norm": 2.4295082092285156, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7287318706512451, + "num_tokens": 273362994.0, + "step": 10665 + }, + { + "epoch": 1.1713156160773117, + "grad_norm": 2.1556334495544434, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6865462064743042, + "num_tokens": 273391828.0, + "step": 10666 + }, + { + "epoch": 1.1714254337799253, + "grad_norm": 2.0414485931396484, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6911938190460205, + "num_tokens": 273426416.0, + "step": 10667 + }, + { + "epoch": 1.171535251482539, + "grad_norm": 2.247605085372925, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7073448300361633, + "num_tokens": 273453733.0, + "step": 10668 + }, + { + "epoch": 1.1716450691851525, + "grad_norm": 2.2053892612457275, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7326310873031616, + "num_tokens": 273480196.0, + "step": 10669 + }, + { + "epoch": 1.1717548868877663, + "grad_norm": 2.145692825317383, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7331297397613525, + "num_tokens": 273505683.0, + "step": 10670 + }, + { + "epoch": 1.17186470459038, + "grad_norm": 2.533519983291626, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7416313886642456, + "num_tokens": 273526254.0, + "step": 10671 + }, + { + "epoch": 1.1719745222929936, + "grad_norm": 2.321328639984131, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6976650953292847, + "num_tokens": 273552750.0, + "step": 10672 + }, + { + "epoch": 1.1720843399956073, + "grad_norm": 2.1273393630981445, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7071771621704102, + "num_tokens": 273580499.0, + "step": 10673 + }, + { + "epoch": 1.1721941576982209, + "grad_norm": 2.010838747024536, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7263428568840027, + "num_tokens": 273611035.0, + "step": 10674 + }, + { + "epoch": 1.1723039754008346, + "grad_norm": 2.315887451171875, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7101792097091675, + "num_tokens": 273635402.0, + "step": 10675 + }, + { + "epoch": 1.1724137931034484, + "grad_norm": 2.3752102851867676, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7508714199066162, + "num_tokens": 273656388.0, + "step": 10676 + }, + { + "epoch": 1.172523610806062, + "grad_norm": 2.044578790664673, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.713700532913208, + "num_tokens": 273688428.0, + "step": 10677 + }, + { + "epoch": 1.1726334285086757, + "grad_norm": 2.362133741378784, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7091438174247742, + "num_tokens": 273712924.0, + "step": 10678 + }, + { + "epoch": 1.1727432462112892, + "grad_norm": 2.39547061920166, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7204118371009827, + "num_tokens": 273736788.0, + "step": 10679 + }, + { + "epoch": 1.172853063913903, + "grad_norm": 2.160799026489258, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7043063640594482, + "num_tokens": 273765986.0, + "step": 10680 + }, + { + "epoch": 1.1729628816165165, + "grad_norm": 2.4118380546569824, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7319983243942261, + "num_tokens": 273790835.0, + "step": 10681 + }, + { + "epoch": 1.1730726993191303, + "grad_norm": 2.2481441497802734, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7363120317459106, + "num_tokens": 273814140.0, + "step": 10682 + }, + { + "epoch": 1.1731825170217438, + "grad_norm": 2.154346227645874, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7188265323638916, + "num_tokens": 273840685.0, + "step": 10683 + }, + { + "epoch": 1.1732923347243576, + "grad_norm": 2.267042636871338, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7116882801055908, + "num_tokens": 273866084.0, + "step": 10684 + }, + { + "epoch": 1.1734021524269713, + "grad_norm": 2.322908878326416, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7210723161697388, + "num_tokens": 273892647.0, + "step": 10685 + }, + { + "epoch": 1.1735119701295849, + "grad_norm": 2.40352725982666, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7057772874832153, + "num_tokens": 273916831.0, + "step": 10686 + }, + { + "epoch": 1.1736217878321986, + "grad_norm": 2.2846832275390625, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7332888245582581, + "num_tokens": 273939960.0, + "step": 10687 + }, + { + "epoch": 1.1737316055348122, + "grad_norm": 2.1842870712280273, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6967739462852478, + "num_tokens": 273966762.0, + "step": 10688 + }, + { + "epoch": 1.173841423237426, + "grad_norm": 2.110954999923706, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7520574331283569, + "num_tokens": 273991345.0, + "step": 10689 + }, + { + "epoch": 1.1739512409400394, + "grad_norm": 2.350368022918701, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7305325269699097, + "num_tokens": 274014494.0, + "step": 10690 + }, + { + "epoch": 1.1740610586426532, + "grad_norm": 2.060774803161621, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7074140310287476, + "num_tokens": 274042336.0, + "step": 10691 + }, + { + "epoch": 1.1741708763452667, + "grad_norm": 2.331368923187256, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7367104887962341, + "num_tokens": 274066398.0, + "step": 10692 + }, + { + "epoch": 1.1742806940478805, + "grad_norm": 2.474536657333374, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7407616376876831, + "num_tokens": 274088749.0, + "step": 10693 + }, + { + "epoch": 1.1743905117504942, + "grad_norm": 2.2707958221435547, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7117066979408264, + "num_tokens": 274114877.0, + "step": 10694 + }, + { + "epoch": 1.1745003294531078, + "grad_norm": 2.2639453411102295, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7035921812057495, + "num_tokens": 274142193.0, + "step": 10695 + }, + { + "epoch": 1.1746101471557215, + "grad_norm": 2.411203622817993, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7070446014404297, + "num_tokens": 274166242.0, + "step": 10696 + }, + { + "epoch": 1.174719964858335, + "grad_norm": 1.987337589263916, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6918594837188721, + "num_tokens": 274197833.0, + "step": 10697 + }, + { + "epoch": 1.1748297825609488, + "grad_norm": 1.9076939821243286, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7168930768966675, + "num_tokens": 274233020.0, + "step": 10698 + }, + { + "epoch": 1.1749396002635626, + "grad_norm": 2.0411126613616943, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.71653151512146, + "num_tokens": 274264698.0, + "step": 10699 + }, + { + "epoch": 1.1750494179661761, + "grad_norm": 1.9926108121871948, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7037119269371033, + "num_tokens": 274294559.0, + "step": 10700 + }, + { + "epoch": 1.1751592356687899, + "grad_norm": 2.3409125804901123, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7276521921157837, + "num_tokens": 274318071.0, + "step": 10701 + }, + { + "epoch": 1.1752690533714034, + "grad_norm": 2.3713011741638184, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7459189891815186, + "num_tokens": 274339464.0, + "step": 10702 + }, + { + "epoch": 1.1753788710740172, + "grad_norm": 2.1042354106903076, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7292795181274414, + "num_tokens": 274366120.0, + "step": 10703 + }, + { + "epoch": 1.1754886887766307, + "grad_norm": 2.2039287090301514, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.6979331374168396, + "num_tokens": 274393201.0, + "step": 10704 + }, + { + "epoch": 1.1755985064792445, + "grad_norm": 2.459273338317871, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7256793975830078, + "num_tokens": 274416353.0, + "step": 10705 + }, + { + "epoch": 1.175708324181858, + "grad_norm": 2.1467416286468506, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7390954494476318, + "num_tokens": 274443715.0, + "step": 10706 + }, + { + "epoch": 1.1758181418844718, + "grad_norm": 2.130774974822998, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7211441397666931, + "num_tokens": 274471358.0, + "step": 10707 + }, + { + "epoch": 1.1759279595870855, + "grad_norm": 2.3493809700012207, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7443044185638428, + "num_tokens": 274493056.0, + "step": 10708 + }, + { + "epoch": 1.176037777289699, + "grad_norm": 2.6276133060455322, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7227470874786377, + "num_tokens": 274513030.0, + "step": 10709 + }, + { + "epoch": 1.1761475949923128, + "grad_norm": 2.1195476055145264, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7022116780281067, + "num_tokens": 274541420.0, + "step": 10710 + }, + { + "epoch": 1.1762574126949263, + "grad_norm": 1.9318755865097046, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7234519720077515, + "num_tokens": 274571993.0, + "step": 10711 + }, + { + "epoch": 1.17636723039754, + "grad_norm": 2.5536725521087646, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7040767669677734, + "num_tokens": 274592703.0, + "step": 10712 + }, + { + "epoch": 1.1764770481001539, + "grad_norm": 2.0989441871643066, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7283369302749634, + "num_tokens": 274622069.0, + "step": 10713 + }, + { + "epoch": 1.1765868658027674, + "grad_norm": 2.753873109817505, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7206268906593323, + "num_tokens": 274640775.0, + "step": 10714 + }, + { + "epoch": 1.1766966835053811, + "grad_norm": 2.037982702255249, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7102935314178467, + "num_tokens": 274669966.0, + "step": 10715 + }, + { + "epoch": 1.1768065012079947, + "grad_norm": 2.322648525238037, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7195755243301392, + "num_tokens": 274694855.0, + "step": 10716 + }, + { + "epoch": 1.1769163189106084, + "grad_norm": 2.1814472675323486, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7163423895835876, + "num_tokens": 274721077.0, + "step": 10717 + }, + { + "epoch": 1.177026136613222, + "grad_norm": 2.6277639865875244, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7275238037109375, + "num_tokens": 274741654.0, + "step": 10718 + }, + { + "epoch": 1.1771359543158357, + "grad_norm": 2.413905382156372, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7154556512832642, + "num_tokens": 274765090.0, + "step": 10719 + }, + { + "epoch": 1.1772457720184493, + "grad_norm": 2.268338203430176, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7063947916030884, + "num_tokens": 274790720.0, + "step": 10720 + }, + { + "epoch": 1.177355589721063, + "grad_norm": 2.3811187744140625, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7214257717132568, + "num_tokens": 274812717.0, + "step": 10721 + }, + { + "epoch": 1.1774654074236768, + "grad_norm": 2.3157429695129395, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6987180113792419, + "num_tokens": 274838793.0, + "step": 10722 + }, + { + "epoch": 1.1775752251262903, + "grad_norm": 2.2798843383789062, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7412408590316772, + "num_tokens": 274865571.0, + "step": 10723 + }, + { + "epoch": 1.177685042828904, + "grad_norm": 2.145202875137329, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7281795740127563, + "num_tokens": 274893643.0, + "step": 10724 + }, + { + "epoch": 1.1777948605315176, + "grad_norm": 2.3623909950256348, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7272675037384033, + "num_tokens": 274917768.0, + "step": 10725 + }, + { + "epoch": 1.1779046782341314, + "grad_norm": 2.3496694564819336, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7181425094604492, + "num_tokens": 274940786.0, + "step": 10726 + }, + { + "epoch": 1.1780144959367451, + "grad_norm": 2.340470552444458, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7047426104545593, + "num_tokens": 274967601.0, + "step": 10727 + }, + { + "epoch": 1.1781243136393587, + "grad_norm": 2.104424476623535, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.713586688041687, + "num_tokens": 274996353.0, + "step": 10728 + }, + { + "epoch": 1.1782341313419724, + "grad_norm": 2.2348647117614746, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7618271112442017, + "num_tokens": 275021722.0, + "step": 10729 + }, + { + "epoch": 1.178343949044586, + "grad_norm": 2.1177265644073486, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7528131604194641, + "num_tokens": 275046142.0, + "step": 10730 + }, + { + "epoch": 1.1784537667471997, + "grad_norm": 2.198669672012329, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7261455059051514, + "num_tokens": 275072632.0, + "step": 10731 + }, + { + "epoch": 1.1785635844498132, + "grad_norm": 2.310483932495117, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7124561071395874, + "num_tokens": 275097819.0, + "step": 10732 + }, + { + "epoch": 1.178673402152427, + "grad_norm": 2.222897529602051, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7043644785881042, + "num_tokens": 275126191.0, + "step": 10733 + }, + { + "epoch": 1.1787832198550405, + "grad_norm": 2.2983503341674805, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7147904634475708, + "num_tokens": 275151470.0, + "step": 10734 + }, + { + "epoch": 1.1788930375576543, + "grad_norm": 2.3574910163879395, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7063189744949341, + "num_tokens": 275176257.0, + "step": 10735 + }, + { + "epoch": 1.179002855260268, + "grad_norm": 2.3675601482391357, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7287667989730835, + "num_tokens": 275201758.0, + "step": 10736 + }, + { + "epoch": 1.1791126729628816, + "grad_norm": 2.27927303314209, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.721890926361084, + "num_tokens": 275225646.0, + "step": 10737 + }, + { + "epoch": 1.1792224906654953, + "grad_norm": 2.757385492324829, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7212051153182983, + "num_tokens": 275244599.0, + "step": 10738 + }, + { + "epoch": 1.1793323083681089, + "grad_norm": 1.9937793016433716, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7167103290557861, + "num_tokens": 275274916.0, + "step": 10739 + }, + { + "epoch": 1.1794421260707226, + "grad_norm": 1.9535105228424072, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.717182457447052, + "num_tokens": 275305057.0, + "step": 10740 + }, + { + "epoch": 1.1795519437733364, + "grad_norm": 1.8993816375732422, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6962383389472961, + "num_tokens": 275340810.0, + "step": 10741 + }, + { + "epoch": 1.17966176147595, + "grad_norm": 2.4183032512664795, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7008649110794067, + "num_tokens": 275366288.0, + "step": 10742 + }, + { + "epoch": 1.1797715791785637, + "grad_norm": 2.355116367340088, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7415921688079834, + "num_tokens": 275388785.0, + "step": 10743 + }, + { + "epoch": 1.1798813968811772, + "grad_norm": 2.411397933959961, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7116189002990723, + "num_tokens": 275412330.0, + "step": 10744 + }, + { + "epoch": 1.179991214583791, + "grad_norm": 2.565232276916504, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7256435751914978, + "num_tokens": 275431746.0, + "step": 10745 + }, + { + "epoch": 1.1801010322864045, + "grad_norm": 2.4335215091705322, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7348588705062866, + "num_tokens": 275453640.0, + "step": 10746 + }, + { + "epoch": 1.1802108499890183, + "grad_norm": 1.9740968942642212, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7012665271759033, + "num_tokens": 275487012.0, + "step": 10747 + }, + { + "epoch": 1.1803206676916318, + "grad_norm": 1.9294136762619019, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.723078727722168, + "num_tokens": 275518564.0, + "step": 10748 + }, + { + "epoch": 1.1804304853942456, + "grad_norm": 2.49255108833313, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7546166777610779, + "num_tokens": 275536952.0, + "step": 10749 + }, + { + "epoch": 1.1805403030968593, + "grad_norm": 2.2560460567474365, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7173746228218079, + "num_tokens": 275561976.0, + "step": 10750 + }, + { + "epoch": 1.1806501207994728, + "grad_norm": 1.975098967552185, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.6989997029304504, + "num_tokens": 275594552.0, + "step": 10751 + }, + { + "epoch": 1.1807599385020866, + "grad_norm": 2.4154889583587646, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7342686653137207, + "num_tokens": 275616962.0, + "step": 10752 + }, + { + "epoch": 1.1808697562047001, + "grad_norm": 2.298023223876953, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7095397710800171, + "num_tokens": 275641010.0, + "step": 10753 + }, + { + "epoch": 1.180979573907314, + "grad_norm": 2.217442512512207, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7023447751998901, + "num_tokens": 275668522.0, + "step": 10754 + }, + { + "epoch": 1.1810893916099274, + "grad_norm": 2.1081020832061768, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7075530886650085, + "num_tokens": 275697474.0, + "step": 10755 + }, + { + "epoch": 1.1811992093125412, + "grad_norm": 2.3353748321533203, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7267971038818359, + "num_tokens": 275723141.0, + "step": 10756 + }, + { + "epoch": 1.1813090270151547, + "grad_norm": 2.1476526260375977, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7254799008369446, + "num_tokens": 275751033.0, + "step": 10757 + }, + { + "epoch": 1.1814188447177685, + "grad_norm": 2.2921974658966064, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7373976707458496, + "num_tokens": 275775928.0, + "step": 10758 + }, + { + "epoch": 1.1815286624203822, + "grad_norm": 1.954993724822998, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.73567795753479, + "num_tokens": 275805554.0, + "step": 10759 + }, + { + "epoch": 1.1816384801229958, + "grad_norm": 2.167572021484375, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6941405534744263, + "num_tokens": 275833278.0, + "step": 10760 + }, + { + "epoch": 1.1817482978256095, + "grad_norm": 2.2137153148651123, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7052557468414307, + "num_tokens": 275860632.0, + "step": 10761 + }, + { + "epoch": 1.181858115528223, + "grad_norm": 1.964198350906372, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7165538668632507, + "num_tokens": 275892590.0, + "step": 10762 + }, + { + "epoch": 1.1819679332308368, + "grad_norm": 2.4912967681884766, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7207788228988647, + "num_tokens": 275915509.0, + "step": 10763 + }, + { + "epoch": 1.1820777509334506, + "grad_norm": 2.2421226501464844, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7252161502838135, + "num_tokens": 275939230.0, + "step": 10764 + }, + { + "epoch": 1.1821875686360641, + "grad_norm": 2.2050633430480957, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7041350603103638, + "num_tokens": 275966848.0, + "step": 10765 + }, + { + "epoch": 1.1822973863386779, + "grad_norm": 2.2942423820495605, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.735222339630127, + "num_tokens": 275990104.0, + "step": 10766 + }, + { + "epoch": 1.1824072040412914, + "grad_norm": 2.5085203647613525, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7318929433822632, + "num_tokens": 276010543.0, + "step": 10767 + }, + { + "epoch": 1.1825170217439052, + "grad_norm": 2.231513738632202, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7111012935638428, + "num_tokens": 276036021.0, + "step": 10768 + }, + { + "epoch": 1.1826268394465187, + "grad_norm": 2.254107713699341, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7300869822502136, + "num_tokens": 276059182.0, + "step": 10769 + }, + { + "epoch": 1.1827366571491325, + "grad_norm": 2.131375312805176, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7175384163856506, + "num_tokens": 276086221.0, + "step": 10770 + }, + { + "epoch": 1.182846474851746, + "grad_norm": 2.3032851219177246, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7299296259880066, + "num_tokens": 276110852.0, + "step": 10771 + }, + { + "epoch": 1.1829562925543597, + "grad_norm": 2.0119776725769043, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7292962670326233, + "num_tokens": 276140240.0, + "step": 10772 + }, + { + "epoch": 1.1830661102569735, + "grad_norm": 2.5567049980163574, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7107507586479187, + "num_tokens": 276159960.0, + "step": 10773 + }, + { + "epoch": 1.183175927959587, + "grad_norm": 2.289653778076172, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7153652906417847, + "num_tokens": 276188806.0, + "step": 10774 + }, + { + "epoch": 1.1832857456622008, + "grad_norm": 2.409180164337158, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7058587670326233, + "num_tokens": 276211486.0, + "step": 10775 + }, + { + "epoch": 1.1833955633648143, + "grad_norm": 1.930799961090088, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6943042278289795, + "num_tokens": 276246692.0, + "step": 10776 + }, + { + "epoch": 1.183505381067428, + "grad_norm": 2.3738210201263428, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7099136114120483, + "num_tokens": 276270463.0, + "step": 10777 + }, + { + "epoch": 1.1836151987700418, + "grad_norm": 2.3875861167907715, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7018676400184631, + "num_tokens": 276293924.0, + "step": 10778 + }, + { + "epoch": 1.1837250164726554, + "grad_norm": 2.222421407699585, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7302809953689575, + "num_tokens": 276319670.0, + "step": 10779 + }, + { + "epoch": 1.1838348341752691, + "grad_norm": 2.0019373893737793, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7207332849502563, + "num_tokens": 276350907.0, + "step": 10780 + }, + { + "epoch": 1.1839446518778827, + "grad_norm": 2.3507680892944336, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7039047479629517, + "num_tokens": 276375900.0, + "step": 10781 + }, + { + "epoch": 1.1840544695804964, + "grad_norm": 2.002798318862915, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6898694634437561, + "num_tokens": 276409563.0, + "step": 10782 + }, + { + "epoch": 1.18416428728311, + "grad_norm": 2.3240866661071777, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6957002878189087, + "num_tokens": 276435936.0, + "step": 10783 + }, + { + "epoch": 1.1842741049857237, + "grad_norm": 2.582996129989624, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7087599635124207, + "num_tokens": 276456583.0, + "step": 10784 + }, + { + "epoch": 1.1843839226883373, + "grad_norm": 2.175967216491699, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7238895893096924, + "num_tokens": 276482986.0, + "step": 10785 + }, + { + "epoch": 1.184493740390951, + "grad_norm": 2.3935327529907227, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7206109762191772, + "num_tokens": 276505305.0, + "step": 10786 + }, + { + "epoch": 1.1846035580935648, + "grad_norm": 1.9738848209381104, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7263187766075134, + "num_tokens": 276538494.0, + "step": 10787 + }, + { + "epoch": 1.1847133757961783, + "grad_norm": 2.266040086746216, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7131223678588867, + "num_tokens": 276564026.0, + "step": 10788 + }, + { + "epoch": 1.184823193498792, + "grad_norm": 2.3485677242279053, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7049423456192017, + "num_tokens": 276588901.0, + "step": 10789 + }, + { + "epoch": 1.1849330112014056, + "grad_norm": 2.418177604675293, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7351118922233582, + "num_tokens": 276611408.0, + "step": 10790 + }, + { + "epoch": 1.1850428289040194, + "grad_norm": 2.074100971221924, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7162736058235168, + "num_tokens": 276640561.0, + "step": 10791 + }, + { + "epoch": 1.185152646606633, + "grad_norm": 2.301140308380127, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7282614707946777, + "num_tokens": 276663787.0, + "step": 10792 + }, + { + "epoch": 1.1852624643092466, + "grad_norm": 2.231599807739258, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7074645161628723, + "num_tokens": 276688989.0, + "step": 10793 + }, + { + "epoch": 1.1853722820118604, + "grad_norm": 2.3432464599609375, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7145236134529114, + "num_tokens": 276712784.0, + "step": 10794 + }, + { + "epoch": 1.185482099714474, + "grad_norm": 2.3627431392669678, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7254693508148193, + "num_tokens": 276738227.0, + "step": 10795 + }, + { + "epoch": 1.1855919174170877, + "grad_norm": 2.176806688308716, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.714394748210907, + "num_tokens": 276763647.0, + "step": 10796 + }, + { + "epoch": 1.1857017351197012, + "grad_norm": 2.352367639541626, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7308595180511475, + "num_tokens": 276788141.0, + "step": 10797 + }, + { + "epoch": 1.185811552822315, + "grad_norm": 2.133344888687134, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7244126796722412, + "num_tokens": 276815283.0, + "step": 10798 + }, + { + "epoch": 1.1859213705249285, + "grad_norm": 2.2541730403900146, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7056266069412231, + "num_tokens": 276841547.0, + "step": 10799 + }, + { + "epoch": 1.1860311882275423, + "grad_norm": 2.239328384399414, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7164562940597534, + "num_tokens": 276869342.0, + "step": 10800 + }, + { + "epoch": 1.186141005930156, + "grad_norm": 2.203387498855591, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7317608594894409, + "num_tokens": 276896230.0, + "step": 10801 + }, + { + "epoch": 1.1862508236327696, + "grad_norm": 2.2014527320861816, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6979629993438721, + "num_tokens": 276926275.0, + "step": 10802 + }, + { + "epoch": 1.1863606413353833, + "grad_norm": 2.3971705436706543, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7226501703262329, + "num_tokens": 276949184.0, + "step": 10803 + }, + { + "epoch": 1.1864704590379969, + "grad_norm": 2.063756227493286, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6970953345298767, + "num_tokens": 276979608.0, + "step": 10804 + }, + { + "epoch": 1.1865802767406106, + "grad_norm": 2.312260150909424, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7361494302749634, + "num_tokens": 277004634.0, + "step": 10805 + }, + { + "epoch": 1.1866900944432242, + "grad_norm": 2.3740973472595215, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7353707551956177, + "num_tokens": 277026575.0, + "step": 10806 + }, + { + "epoch": 1.186799912145838, + "grad_norm": 2.051548719406128, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7345911264419556, + "num_tokens": 277057422.0, + "step": 10807 + }, + { + "epoch": 1.1869097298484517, + "grad_norm": 2.1335346698760986, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7113794088363647, + "num_tokens": 277086064.0, + "step": 10808 + }, + { + "epoch": 1.1870195475510652, + "grad_norm": 2.4503884315490723, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7446746826171875, + "num_tokens": 277108757.0, + "step": 10809 + }, + { + "epoch": 1.187129365253679, + "grad_norm": 2.2134501934051514, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7201231718063354, + "num_tokens": 277135184.0, + "step": 10810 + }, + { + "epoch": 1.1872391829562925, + "grad_norm": 2.0931437015533447, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7183544635772705, + "num_tokens": 277163860.0, + "step": 10811 + }, + { + "epoch": 1.1873490006589063, + "grad_norm": 2.229046583175659, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.704032838344574, + "num_tokens": 277191538.0, + "step": 10812 + }, + { + "epoch": 1.1874588183615198, + "grad_norm": 2.3701298236846924, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7330735921859741, + "num_tokens": 277215292.0, + "step": 10813 + }, + { + "epoch": 1.1875686360641335, + "grad_norm": 2.27187442779541, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7354834675788879, + "num_tokens": 277239206.0, + "step": 10814 + }, + { + "epoch": 1.1876784537667473, + "grad_norm": 2.3880720138549805, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7130579352378845, + "num_tokens": 277262076.0, + "step": 10815 + }, + { + "epoch": 1.1877882714693608, + "grad_norm": 2.425119638442993, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7414892315864563, + "num_tokens": 277284044.0, + "step": 10816 + }, + { + "epoch": 1.1878980891719746, + "grad_norm": 2.512087821960449, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7132655382156372, + "num_tokens": 277305067.0, + "step": 10817 + }, + { + "epoch": 1.1880079068745881, + "grad_norm": 2.2391250133514404, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7117516398429871, + "num_tokens": 277331535.0, + "step": 10818 + }, + { + "epoch": 1.1881177245772019, + "grad_norm": 2.3753345012664795, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7243423461914062, + "num_tokens": 277355453.0, + "step": 10819 + }, + { + "epoch": 1.1882275422798154, + "grad_norm": 2.3257646560668945, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7419459819793701, + "num_tokens": 277377966.0, + "step": 10820 + }, + { + "epoch": 1.1883373599824292, + "grad_norm": 1.900540828704834, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7077502012252808, + "num_tokens": 277412367.0, + "step": 10821 + }, + { + "epoch": 1.1884471776850427, + "grad_norm": 2.1590843200683594, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7227928042411804, + "num_tokens": 277437802.0, + "step": 10822 + }, + { + "epoch": 1.1885569953876565, + "grad_norm": 2.4624624252319336, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7382292747497559, + "num_tokens": 277459359.0, + "step": 10823 + }, + { + "epoch": 1.1886668130902702, + "grad_norm": 2.0412533283233643, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7097392082214355, + "num_tokens": 277490185.0, + "step": 10824 + }, + { + "epoch": 1.1887766307928838, + "grad_norm": 2.247713804244995, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7431318759918213, + "num_tokens": 277513921.0, + "step": 10825 + }, + { + "epoch": 1.1888864484954975, + "grad_norm": 2.433337688446045, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7453951835632324, + "num_tokens": 277533666.0, + "step": 10826 + }, + { + "epoch": 1.188996266198111, + "grad_norm": 2.402496337890625, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7024514675140381, + "num_tokens": 277556290.0, + "step": 10827 + }, + { + "epoch": 1.1891060839007248, + "grad_norm": 2.689100742340088, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7405721545219421, + "num_tokens": 277574743.0, + "step": 10828 + }, + { + "epoch": 1.1892159016033386, + "grad_norm": 2.35927152633667, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7441284656524658, + "num_tokens": 277597034.0, + "step": 10829 + }, + { + "epoch": 1.189325719305952, + "grad_norm": 1.9363237619400024, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7042825818061829, + "num_tokens": 277629425.0, + "step": 10830 + }, + { + "epoch": 1.1894355370085659, + "grad_norm": 1.9727140665054321, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.711479663848877, + "num_tokens": 277662480.0, + "step": 10831 + }, + { + "epoch": 1.1895453547111794, + "grad_norm": 2.3574352264404297, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6915256977081299, + "num_tokens": 277686724.0, + "step": 10832 + }, + { + "epoch": 1.1896551724137931, + "grad_norm": 1.9905807971954346, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7159320116043091, + "num_tokens": 277720308.0, + "step": 10833 + }, + { + "epoch": 1.1897649901164067, + "grad_norm": 2.3114964962005615, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7166767120361328, + "num_tokens": 277743801.0, + "step": 10834 + }, + { + "epoch": 1.1898748078190204, + "grad_norm": 2.38775897026062, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7137221693992615, + "num_tokens": 277767106.0, + "step": 10835 + }, + { + "epoch": 1.189984625521634, + "grad_norm": 2.0677263736724854, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6885198354721069, + "num_tokens": 277800087.0, + "step": 10836 + }, + { + "epoch": 1.1900944432242477, + "grad_norm": 2.2968838214874268, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7234121561050415, + "num_tokens": 277827021.0, + "step": 10837 + }, + { + "epoch": 1.1902042609268615, + "grad_norm": 2.522794485092163, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7261830568313599, + "num_tokens": 277848001.0, + "step": 10838 + }, + { + "epoch": 1.190314078629475, + "grad_norm": 2.1834754943847656, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7068924903869629, + "num_tokens": 277874822.0, + "step": 10839 + }, + { + "epoch": 1.1904238963320888, + "grad_norm": 2.226813793182373, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.707897424697876, + "num_tokens": 277904128.0, + "step": 10840 + }, + { + "epoch": 1.1905337140347023, + "grad_norm": 2.145111560821533, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7152189016342163, + "num_tokens": 277931376.0, + "step": 10841 + }, + { + "epoch": 1.190643531737316, + "grad_norm": 2.3292224407196045, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7187329530715942, + "num_tokens": 277954814.0, + "step": 10842 + }, + { + "epoch": 1.1907533494399298, + "grad_norm": 2.2872443199157715, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7067673206329346, + "num_tokens": 277978952.0, + "step": 10843 + }, + { + "epoch": 1.1908631671425434, + "grad_norm": 2.1472084522247314, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7007615566253662, + "num_tokens": 278007885.0, + "step": 10844 + }, + { + "epoch": 1.1909729848451571, + "grad_norm": 2.0401499271392822, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.718195915222168, + "num_tokens": 278037623.0, + "step": 10845 + }, + { + "epoch": 1.1910828025477707, + "grad_norm": 2.0444445610046387, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7473068237304688, + "num_tokens": 278064558.0, + "step": 10846 + }, + { + "epoch": 1.1911926202503844, + "grad_norm": 2.0921053886413574, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7247958183288574, + "num_tokens": 278094151.0, + "step": 10847 + }, + { + "epoch": 1.191302437952998, + "grad_norm": 2.187553644180298, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7353711724281311, + "num_tokens": 278120763.0, + "step": 10848 + }, + { + "epoch": 1.1914122556556117, + "grad_norm": 2.3689815998077393, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.714231014251709, + "num_tokens": 278144525.0, + "step": 10849 + }, + { + "epoch": 1.1915220733582252, + "grad_norm": 2.1908533573150635, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7417503595352173, + "num_tokens": 278169374.0, + "step": 10850 + }, + { + "epoch": 1.191631891060839, + "grad_norm": 2.1972594261169434, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.706180214881897, + "num_tokens": 278198575.0, + "step": 10851 + }, + { + "epoch": 1.1917417087634528, + "grad_norm": 2.339982271194458, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7088262438774109, + "num_tokens": 278223747.0, + "step": 10852 + }, + { + "epoch": 1.1918515264660663, + "grad_norm": 2.365314245223999, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7315047979354858, + "num_tokens": 278247397.0, + "step": 10853 + }, + { + "epoch": 1.19196134416868, + "grad_norm": 2.179208278656006, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7160513401031494, + "num_tokens": 278275034.0, + "step": 10854 + }, + { + "epoch": 1.1920711618712936, + "grad_norm": 1.9593291282653809, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.726794958114624, + "num_tokens": 278303551.0, + "step": 10855 + }, + { + "epoch": 1.1921809795739073, + "grad_norm": 2.2327914237976074, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7262866497039795, + "num_tokens": 278330443.0, + "step": 10856 + }, + { + "epoch": 1.192290797276521, + "grad_norm": 2.0507309436798096, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6833810806274414, + "num_tokens": 278361243.0, + "step": 10857 + }, + { + "epoch": 1.1924006149791346, + "grad_norm": 2.392033576965332, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7142744660377502, + "num_tokens": 278384348.0, + "step": 10858 + }, + { + "epoch": 1.1925104326817484, + "grad_norm": 2.4481093883514404, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7509496212005615, + "num_tokens": 278404434.0, + "step": 10859 + }, + { + "epoch": 1.192620250384362, + "grad_norm": 2.254033088684082, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6816997528076172, + "num_tokens": 278432906.0, + "step": 10860 + }, + { + "epoch": 1.1927300680869757, + "grad_norm": 2.305274486541748, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7175928950309753, + "num_tokens": 278457357.0, + "step": 10861 + }, + { + "epoch": 1.1928398857895892, + "grad_norm": 2.106452226638794, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6997824907302856, + "num_tokens": 278484548.0, + "step": 10862 + }, + { + "epoch": 1.192949703492203, + "grad_norm": 2.351900100708008, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7057162523269653, + "num_tokens": 278509807.0, + "step": 10863 + }, + { + "epoch": 1.1930595211948165, + "grad_norm": 2.3495054244995117, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7399593591690063, + "num_tokens": 278531853.0, + "step": 10864 + }, + { + "epoch": 1.1931693388974303, + "grad_norm": 2.131819486618042, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7277071475982666, + "num_tokens": 278558337.0, + "step": 10865 + }, + { + "epoch": 1.193279156600044, + "grad_norm": 2.048788547515869, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7102788686752319, + "num_tokens": 278589001.0, + "step": 10866 + }, + { + "epoch": 1.1933889743026576, + "grad_norm": 2.064467430114746, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.69452303647995, + "num_tokens": 278619350.0, + "step": 10867 + }, + { + "epoch": 1.1934987920052713, + "grad_norm": 2.2233104705810547, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7113711833953857, + "num_tokens": 278648353.0, + "step": 10868 + }, + { + "epoch": 1.1936086097078848, + "grad_norm": 2.4568727016448975, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7235727310180664, + "num_tokens": 278669104.0, + "step": 10869 + }, + { + "epoch": 1.1937184274104986, + "grad_norm": 2.1979217529296875, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.723527193069458, + "num_tokens": 278696293.0, + "step": 10870 + }, + { + "epoch": 1.1938282451131121, + "grad_norm": 2.2364161014556885, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6882123947143555, + "num_tokens": 278722194.0, + "step": 10871 + }, + { + "epoch": 1.193938062815726, + "grad_norm": 2.2812082767486572, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7181926965713501, + "num_tokens": 278744609.0, + "step": 10872 + }, + { + "epoch": 1.1940478805183394, + "grad_norm": 2.175577163696289, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7192922830581665, + "num_tokens": 278771690.0, + "step": 10873 + }, + { + "epoch": 1.1941576982209532, + "grad_norm": 2.0493454933166504, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.709061861038208, + "num_tokens": 278802700.0, + "step": 10874 + }, + { + "epoch": 1.194267515923567, + "grad_norm": 2.3246920108795166, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7080610990524292, + "num_tokens": 278828812.0, + "step": 10875 + }, + { + "epoch": 1.1943773336261805, + "grad_norm": 2.577448606491089, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7079695463180542, + "num_tokens": 278849239.0, + "step": 10876 + }, + { + "epoch": 1.1944871513287942, + "grad_norm": 2.190103769302368, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7243270874023438, + "num_tokens": 278874568.0, + "step": 10877 + }, + { + "epoch": 1.1945969690314078, + "grad_norm": 2.3518598079681396, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7177494168281555, + "num_tokens": 278898332.0, + "step": 10878 + }, + { + "epoch": 1.1947067867340215, + "grad_norm": 2.1416335105895996, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7112342715263367, + "num_tokens": 278926955.0, + "step": 10879 + }, + { + "epoch": 1.1948166044366353, + "grad_norm": 2.3996312618255615, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.717888593673706, + "num_tokens": 278951077.0, + "step": 10880 + }, + { + "epoch": 1.1949264221392488, + "grad_norm": 2.2075750827789307, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6946988105773926, + "num_tokens": 278979962.0, + "step": 10881 + }, + { + "epoch": 1.1950362398418626, + "grad_norm": 2.240384578704834, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.73395836353302, + "num_tokens": 279003008.0, + "step": 10882 + }, + { + "epoch": 1.1951460575444761, + "grad_norm": 2.541280508041382, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7228756546974182, + "num_tokens": 279023942.0, + "step": 10883 + }, + { + "epoch": 1.1952558752470899, + "grad_norm": 2.3233683109283447, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7166445851325989, + "num_tokens": 279046911.0, + "step": 10884 + }, + { + "epoch": 1.1953656929497034, + "grad_norm": 2.4817521572113037, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7127329111099243, + "num_tokens": 279068954.0, + "step": 10885 + }, + { + "epoch": 1.1954755106523172, + "grad_norm": 2.1200597286224365, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7259702086448669, + "num_tokens": 279095308.0, + "step": 10886 + }, + { + "epoch": 1.1955853283549307, + "grad_norm": 2.8354268074035645, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7346216440200806, + "num_tokens": 279112505.0, + "step": 10887 + }, + { + "epoch": 1.1956951460575445, + "grad_norm": 2.1882569789886475, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.711465060710907, + "num_tokens": 279140031.0, + "step": 10888 + }, + { + "epoch": 1.1958049637601582, + "grad_norm": 2.1483192443847656, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7199202179908752, + "num_tokens": 279166027.0, + "step": 10889 + }, + { + "epoch": 1.1959147814627717, + "grad_norm": 2.0933170318603516, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7234346270561218, + "num_tokens": 279193107.0, + "step": 10890 + }, + { + "epoch": 1.1960245991653855, + "grad_norm": 2.1837565898895264, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7514157295227051, + "num_tokens": 279220809.0, + "step": 10891 + }, + { + "epoch": 1.196134416867999, + "grad_norm": 2.2793056964874268, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7495867013931274, + "num_tokens": 279243246.0, + "step": 10892 + }, + { + "epoch": 1.1962442345706128, + "grad_norm": 2.3049404621124268, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7192038893699646, + "num_tokens": 279267050.0, + "step": 10893 + }, + { + "epoch": 1.1963540522732266, + "grad_norm": 2.125359535217285, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.707897961139679, + "num_tokens": 279295368.0, + "step": 10894 + }, + { + "epoch": 1.19646386997584, + "grad_norm": 2.3652632236480713, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7152766585350037, + "num_tokens": 279319587.0, + "step": 10895 + }, + { + "epoch": 1.1965736876784538, + "grad_norm": 2.1927683353424072, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7172432541847229, + "num_tokens": 279343616.0, + "step": 10896 + }, + { + "epoch": 1.1966835053810674, + "grad_norm": 2.2284910678863525, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7172643542289734, + "num_tokens": 279368542.0, + "step": 10897 + }, + { + "epoch": 1.1967933230836811, + "grad_norm": 2.205312490463257, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.728118896484375, + "num_tokens": 279392133.0, + "step": 10898 + }, + { + "epoch": 1.1969031407862947, + "grad_norm": 1.9912837743759155, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7028301954269409, + "num_tokens": 279424439.0, + "step": 10899 + }, + { + "epoch": 1.1970129584889084, + "grad_norm": 2.2440829277038574, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7272019386291504, + "num_tokens": 279451392.0, + "step": 10900 + }, + { + "epoch": 1.197122776191522, + "grad_norm": 2.158320903778076, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7028185129165649, + "num_tokens": 279476900.0, + "step": 10901 + }, + { + "epoch": 1.1972325938941357, + "grad_norm": 2.3216888904571533, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7566144466400146, + "num_tokens": 279498564.0, + "step": 10902 + }, + { + "epoch": 1.1973424115967495, + "grad_norm": 2.3233206272125244, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7390789985656738, + "num_tokens": 279522015.0, + "step": 10903 + }, + { + "epoch": 1.197452229299363, + "grad_norm": 2.3853108882904053, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.707260251045227, + "num_tokens": 279545325.0, + "step": 10904 + }, + { + "epoch": 1.1975620470019768, + "grad_norm": 2.4135279655456543, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7254347205162048, + "num_tokens": 279567236.0, + "step": 10905 + }, + { + "epoch": 1.1976718647045903, + "grad_norm": 2.6789357662200928, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7202129364013672, + "num_tokens": 279587123.0, + "step": 10906 + }, + { + "epoch": 1.197781682407204, + "grad_norm": 2.0660908222198486, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7318628430366516, + "num_tokens": 279615096.0, + "step": 10907 + }, + { + "epoch": 1.1978915001098178, + "grad_norm": 2.5216996669769287, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7241815328598022, + "num_tokens": 279636970.0, + "step": 10908 + }, + { + "epoch": 1.1980013178124314, + "grad_norm": 2.268458127975464, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7222139239311218, + "num_tokens": 279663571.0, + "step": 10909 + }, + { + "epoch": 1.1981111355150451, + "grad_norm": 2.3404691219329834, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.6970610618591309, + "num_tokens": 279687184.0, + "step": 10910 + }, + { + "epoch": 1.1982209532176586, + "grad_norm": 2.2465100288391113, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7341426014900208, + "num_tokens": 279713126.0, + "step": 10911 + }, + { + "epoch": 1.1983307709202724, + "grad_norm": 1.9478027820587158, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7168790102005005, + "num_tokens": 279745651.0, + "step": 10912 + }, + { + "epoch": 1.198440588622886, + "grad_norm": 2.419179916381836, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7276079654693604, + "num_tokens": 279768691.0, + "step": 10913 + }, + { + "epoch": 1.1985504063254997, + "grad_norm": 1.957931637763977, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6989607810974121, + "num_tokens": 279800175.0, + "step": 10914 + }, + { + "epoch": 1.1986602240281132, + "grad_norm": 1.935455083847046, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7046968340873718, + "num_tokens": 279834567.0, + "step": 10915 + }, + { + "epoch": 1.198770041730727, + "grad_norm": 2.255150079727173, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7273268699645996, + "num_tokens": 279859324.0, + "step": 10916 + }, + { + "epoch": 1.1988798594333407, + "grad_norm": 2.165205955505371, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7267405986785889, + "num_tokens": 279885228.0, + "step": 10917 + }, + { + "epoch": 1.1989896771359543, + "grad_norm": 2.148733377456665, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7131757140159607, + "num_tokens": 279913523.0, + "step": 10918 + }, + { + "epoch": 1.199099494838568, + "grad_norm": 2.7293317317962646, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7201160192489624, + "num_tokens": 279932536.0, + "step": 10919 + }, + { + "epoch": 1.1992093125411816, + "grad_norm": 2.1642560958862305, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7035182118415833, + "num_tokens": 279960199.0, + "step": 10920 + }, + { + "epoch": 1.1993191302437953, + "grad_norm": 2.080886125564575, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7205735445022583, + "num_tokens": 279988974.0, + "step": 10921 + }, + { + "epoch": 1.199428947946409, + "grad_norm": 2.2848193645477295, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.721354067325592, + "num_tokens": 280013782.0, + "step": 10922 + }, + { + "epoch": 1.1995387656490226, + "grad_norm": 2.127579689025879, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7187187671661377, + "num_tokens": 280041482.0, + "step": 10923 + }, + { + "epoch": 1.1996485833516364, + "grad_norm": 2.2824227809906006, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7365038394927979, + "num_tokens": 280064838.0, + "step": 10924 + }, + { + "epoch": 1.19975840105425, + "grad_norm": 2.440120220184326, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7322490215301514, + "num_tokens": 280088091.0, + "step": 10925 + }, + { + "epoch": 1.1998682187568637, + "grad_norm": 2.0553348064422607, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7360440492630005, + "num_tokens": 280116861.0, + "step": 10926 + }, + { + "epoch": 1.1999780364594772, + "grad_norm": 2.1456329822540283, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7454166412353516, + "num_tokens": 280144232.0, + "step": 10927 + }, + { + "epoch": 1.200087854162091, + "grad_norm": 2.2993459701538086, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7056010961532593, + "num_tokens": 280170427.0, + "step": 10928 + }, + { + "epoch": 1.2001976718647045, + "grad_norm": 2.0539541244506836, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7099159955978394, + "num_tokens": 280200556.0, + "step": 10929 + }, + { + "epoch": 1.2003074895673183, + "grad_norm": 2.199018955230713, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.6996821761131287, + "num_tokens": 280226741.0, + "step": 10930 + }, + { + "epoch": 1.200417307269932, + "grad_norm": 2.6576478481292725, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7298847436904907, + "num_tokens": 280246288.0, + "step": 10931 + }, + { + "epoch": 1.2005271249725455, + "grad_norm": 2.367781162261963, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7282360792160034, + "num_tokens": 280269178.0, + "step": 10932 + }, + { + "epoch": 1.2006369426751593, + "grad_norm": 2.1999497413635254, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7159186601638794, + "num_tokens": 280296350.0, + "step": 10933 + }, + { + "epoch": 1.2007467603777728, + "grad_norm": 2.20224928855896, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7137147188186646, + "num_tokens": 280323617.0, + "step": 10934 + }, + { + "epoch": 1.2008565780803866, + "grad_norm": 2.5755720138549805, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6978417634963989, + "num_tokens": 280349625.0, + "step": 10935 + }, + { + "epoch": 1.2009663957830001, + "grad_norm": 2.1017444133758545, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7018414735794067, + "num_tokens": 280379517.0, + "step": 10936 + }, + { + "epoch": 1.2010762134856139, + "grad_norm": 2.5376229286193848, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7299720048904419, + "num_tokens": 280400265.0, + "step": 10937 + }, + { + "epoch": 1.2011860311882274, + "grad_norm": 2.341500759124756, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.717549204826355, + "num_tokens": 280424155.0, + "step": 10938 + }, + { + "epoch": 1.2012958488908412, + "grad_norm": 2.1191580295562744, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7332444190979004, + "num_tokens": 280450767.0, + "step": 10939 + }, + { + "epoch": 1.201405666593455, + "grad_norm": 2.060594081878662, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6975198984146118, + "num_tokens": 280482638.0, + "step": 10940 + }, + { + "epoch": 1.2015154842960685, + "grad_norm": 2.461785316467285, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7067317962646484, + "num_tokens": 280506463.0, + "step": 10941 + }, + { + "epoch": 1.2016253019986822, + "grad_norm": 2.024322271347046, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7098166346549988, + "num_tokens": 280538802.0, + "step": 10942 + }, + { + "epoch": 1.2017351197012958, + "grad_norm": 2.210521936416626, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.722599983215332, + "num_tokens": 280563800.0, + "step": 10943 + }, + { + "epoch": 1.2018449374039095, + "grad_norm": 2.176238775253296, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7301515340805054, + "num_tokens": 280590362.0, + "step": 10944 + }, + { + "epoch": 1.2019547551065233, + "grad_norm": 2.688265085220337, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7177853584289551, + "num_tokens": 280610534.0, + "step": 10945 + }, + { + "epoch": 1.2020645728091368, + "grad_norm": 2.3254270553588867, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7243942022323608, + "num_tokens": 280633369.0, + "step": 10946 + }, + { + "epoch": 1.2021743905117506, + "grad_norm": 2.3359127044677734, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7415189743041992, + "num_tokens": 280657246.0, + "step": 10947 + }, + { + "epoch": 1.202284208214364, + "grad_norm": 2.559272050857544, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7242798805236816, + "num_tokens": 280679632.0, + "step": 10948 + }, + { + "epoch": 1.2023940259169779, + "grad_norm": 2.262390613555908, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.709884524345398, + "num_tokens": 280705230.0, + "step": 10949 + }, + { + "epoch": 1.2025038436195914, + "grad_norm": 2.19150710105896, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7288327217102051, + "num_tokens": 280730760.0, + "step": 10950 + }, + { + "epoch": 1.2026136613222052, + "grad_norm": 2.1156411170959473, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7327330112457275, + "num_tokens": 280758565.0, + "step": 10951 + }, + { + "epoch": 1.2027234790248187, + "grad_norm": 2.0386905670166016, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6804418563842773, + "num_tokens": 280790179.0, + "step": 10952 + }, + { + "epoch": 1.2028332967274324, + "grad_norm": 2.3251566886901855, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7202234268188477, + "num_tokens": 280815214.0, + "step": 10953 + }, + { + "epoch": 1.2029431144300462, + "grad_norm": 2.2505648136138916, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7382757067680359, + "num_tokens": 280840262.0, + "step": 10954 + }, + { + "epoch": 1.2030529321326597, + "grad_norm": 2.1940174102783203, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7254295349121094, + "num_tokens": 280869235.0, + "step": 10955 + }, + { + "epoch": 1.2031627498352735, + "grad_norm": 2.1940696239471436, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7182140350341797, + "num_tokens": 280896392.0, + "step": 10956 + }, + { + "epoch": 1.203272567537887, + "grad_norm": 2.040489673614502, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7089913487434387, + "num_tokens": 280926294.0, + "step": 10957 + }, + { + "epoch": 1.2033823852405008, + "grad_norm": 2.0635643005371094, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7113629579544067, + "num_tokens": 280957815.0, + "step": 10958 + }, + { + "epoch": 1.2034922029431145, + "grad_norm": 2.335247755050659, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7298279404640198, + "num_tokens": 280981145.0, + "step": 10959 + }, + { + "epoch": 1.203602020645728, + "grad_norm": 2.0952320098876953, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.705098032951355, + "num_tokens": 281009346.0, + "step": 10960 + }, + { + "epoch": 1.2037118383483418, + "grad_norm": 2.295532703399658, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7338622808456421, + "num_tokens": 281034205.0, + "step": 10961 + }, + { + "epoch": 1.2038216560509554, + "grad_norm": 2.151017189025879, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7114689350128174, + "num_tokens": 281062217.0, + "step": 10962 + }, + { + "epoch": 1.2039314737535691, + "grad_norm": 2.4416191577911377, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7121986150741577, + "num_tokens": 281083465.0, + "step": 10963 + }, + { + "epoch": 1.2040412914561827, + "grad_norm": 2.3951332569122314, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7186967134475708, + "num_tokens": 281105537.0, + "step": 10964 + }, + { + "epoch": 1.2041511091587964, + "grad_norm": 2.3725249767303467, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7152841091156006, + "num_tokens": 281129225.0, + "step": 10965 + }, + { + "epoch": 1.20426092686141, + "grad_norm": 2.630004405975342, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7279635071754456, + "num_tokens": 281150027.0, + "step": 10966 + }, + { + "epoch": 1.2043707445640237, + "grad_norm": 2.4315414428710938, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.723975658416748, + "num_tokens": 281171975.0, + "step": 10967 + }, + { + "epoch": 1.2044805622666375, + "grad_norm": 2.434678792953491, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7283146381378174, + "num_tokens": 281194502.0, + "step": 10968 + }, + { + "epoch": 1.204590379969251, + "grad_norm": 2.1808245182037354, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7188755869865417, + "num_tokens": 281221441.0, + "step": 10969 + }, + { + "epoch": 1.2047001976718648, + "grad_norm": 2.3074097633361816, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.6998776197433472, + "num_tokens": 281246213.0, + "step": 10970 + }, + { + "epoch": 1.2048100153744783, + "grad_norm": 2.4011764526367188, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7327141761779785, + "num_tokens": 281269990.0, + "step": 10971 + }, + { + "epoch": 1.204919833077092, + "grad_norm": 2.332719564437866, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7249892950057983, + "num_tokens": 281294330.0, + "step": 10972 + }, + { + "epoch": 1.2050296507797058, + "grad_norm": 2.1274025440216064, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7339626550674438, + "num_tokens": 281321846.0, + "step": 10973 + }, + { + "epoch": 1.2051394684823193, + "grad_norm": 2.567976236343384, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7519695162773132, + "num_tokens": 281343193.0, + "step": 10974 + }, + { + "epoch": 1.205249286184933, + "grad_norm": 2.549528121948242, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7356386184692383, + "num_tokens": 281361842.0, + "step": 10975 + }, + { + "epoch": 1.2053591038875466, + "grad_norm": 2.165736675262451, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7202314734458923, + "num_tokens": 281388254.0, + "step": 10976 + }, + { + "epoch": 1.2054689215901604, + "grad_norm": 2.2746994495391846, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7081516981124878, + "num_tokens": 281415549.0, + "step": 10977 + }, + { + "epoch": 1.205578739292774, + "grad_norm": 2.1581685543060303, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7005654573440552, + "num_tokens": 281442523.0, + "step": 10978 + }, + { + "epoch": 1.2056885569953877, + "grad_norm": 2.049630880355835, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7060623168945312, + "num_tokens": 281473388.0, + "step": 10979 + }, + { + "epoch": 1.2057983746980012, + "grad_norm": 2.207437753677368, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7230506539344788, + "num_tokens": 281500291.0, + "step": 10980 + }, + { + "epoch": 1.205908192400615, + "grad_norm": 1.9719579219818115, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7308122515678406, + "num_tokens": 281529448.0, + "step": 10981 + }, + { + "epoch": 1.2060180101032287, + "grad_norm": 2.1702303886413574, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7129741311073303, + "num_tokens": 281556135.0, + "step": 10982 + }, + { + "epoch": 1.2061278278058423, + "grad_norm": 2.2459311485290527, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7170319557189941, + "num_tokens": 281582085.0, + "step": 10983 + }, + { + "epoch": 1.206237645508456, + "grad_norm": 2.081467628479004, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7108186483383179, + "num_tokens": 281609911.0, + "step": 10984 + }, + { + "epoch": 1.2063474632110696, + "grad_norm": 2.383862257003784, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7227614521980286, + "num_tokens": 281633173.0, + "step": 10985 + }, + { + "epoch": 1.2064572809136833, + "grad_norm": 2.260850667953491, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7129359841346741, + "num_tokens": 281657779.0, + "step": 10986 + }, + { + "epoch": 1.2065670986162969, + "grad_norm": 2.1306490898132324, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7285609841346741, + "num_tokens": 281685154.0, + "step": 10987 + }, + { + "epoch": 1.2066769163189106, + "grad_norm": 2.083162546157837, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7348461151123047, + "num_tokens": 281712002.0, + "step": 10988 + }, + { + "epoch": 1.2067867340215244, + "grad_norm": 2.299495220184326, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7370842099189758, + "num_tokens": 281734105.0, + "step": 10989 + }, + { + "epoch": 1.206896551724138, + "grad_norm": 2.483741044998169, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7188436985015869, + "num_tokens": 281756363.0, + "step": 10990 + }, + { + "epoch": 1.2070063694267517, + "grad_norm": 2.4684252738952637, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.717968225479126, + "num_tokens": 281778645.0, + "step": 10991 + }, + { + "epoch": 1.2071161871293652, + "grad_norm": 2.322260856628418, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7043988108634949, + "num_tokens": 281802099.0, + "step": 10992 + }, + { + "epoch": 1.207226004831979, + "grad_norm": 2.39690899848938, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7292736768722534, + "num_tokens": 281824266.0, + "step": 10993 + }, + { + "epoch": 1.2073358225345925, + "grad_norm": 2.222368001937866, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7220654487609863, + "num_tokens": 281850060.0, + "step": 10994 + }, + { + "epoch": 1.2074456402372062, + "grad_norm": 2.4694042205810547, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7306103706359863, + "num_tokens": 281873906.0, + "step": 10995 + }, + { + "epoch": 1.20755545793982, + "grad_norm": 2.5472772121429443, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.739897608757019, + "num_tokens": 281895296.0, + "step": 10996 + }, + { + "epoch": 1.2076652756424335, + "grad_norm": 2.296537399291992, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.7091736197471619, + "num_tokens": 281920534.0, + "step": 10997 + }, + { + "epoch": 1.2077750933450473, + "grad_norm": 2.192429780960083, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7222708463668823, + "num_tokens": 281947015.0, + "step": 10998 + }, + { + "epoch": 1.2078849110476608, + "grad_norm": 2.6536977291107178, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7324673533439636, + "num_tokens": 281966823.0, + "step": 10999 + }, + { + "epoch": 1.2079947287502746, + "grad_norm": 2.08903169631958, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7503434419631958, + "num_tokens": 281992213.0, + "step": 11000 + }, + { + "epoch": 1.2081045464528881, + "grad_norm": 2.2316219806671143, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7251947522163391, + "num_tokens": 282017240.0, + "step": 11001 + }, + { + "epoch": 1.2082143641555019, + "grad_norm": 2.3300235271453857, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7258442640304565, + "num_tokens": 282040682.0, + "step": 11002 + }, + { + "epoch": 1.2083241818581154, + "grad_norm": 2.1658787727355957, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7196009159088135, + "num_tokens": 282068626.0, + "step": 11003 + }, + { + "epoch": 1.2084339995607292, + "grad_norm": 2.3079938888549805, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.693938136100769, + "num_tokens": 282093865.0, + "step": 11004 + }, + { + "epoch": 1.208543817263343, + "grad_norm": 2.1673128604888916, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7081190347671509, + "num_tokens": 282123772.0, + "step": 11005 + }, + { + "epoch": 1.2086536349659565, + "grad_norm": 2.2268688678741455, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7301600575447083, + "num_tokens": 282148839.0, + "step": 11006 + }, + { + "epoch": 1.2087634526685702, + "grad_norm": 1.9408420324325562, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7091354131698608, + "num_tokens": 282179376.0, + "step": 11007 + }, + { + "epoch": 1.2088732703711838, + "grad_norm": 2.1486401557922363, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.739984929561615, + "num_tokens": 282205339.0, + "step": 11008 + }, + { + "epoch": 1.2089830880737975, + "grad_norm": 2.187028169631958, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7151157855987549, + "num_tokens": 282233336.0, + "step": 11009 + }, + { + "epoch": 1.2090929057764113, + "grad_norm": 2.427873373031616, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7305887937545776, + "num_tokens": 282257799.0, + "step": 11010 + }, + { + "epoch": 1.2092027234790248, + "grad_norm": 2.1103129386901855, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7355233430862427, + "num_tokens": 282284071.0, + "step": 11011 + }, + { + "epoch": 1.2093125411816386, + "grad_norm": 2.2345149517059326, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7220926284790039, + "num_tokens": 282308365.0, + "step": 11012 + }, + { + "epoch": 1.209422358884252, + "grad_norm": 2.10457706451416, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7115403413772583, + "num_tokens": 282337269.0, + "step": 11013 + }, + { + "epoch": 1.2095321765868658, + "grad_norm": 2.1125316619873047, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7112220525741577, + "num_tokens": 282365672.0, + "step": 11014 + }, + { + "epoch": 1.2096419942894794, + "grad_norm": 2.2097740173339844, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7025067806243896, + "num_tokens": 282391244.0, + "step": 11015 + }, + { + "epoch": 1.2097518119920931, + "grad_norm": 2.1966700553894043, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7296363115310669, + "num_tokens": 282416665.0, + "step": 11016 + }, + { + "epoch": 1.2098616296947067, + "grad_norm": 2.3226053714752197, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7529950141906738, + "num_tokens": 282439389.0, + "step": 11017 + }, + { + "epoch": 1.2099714473973204, + "grad_norm": 2.3743624687194824, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7120869159698486, + "num_tokens": 282461326.0, + "step": 11018 + }, + { + "epoch": 1.2100812650999342, + "grad_norm": 1.9490429162979126, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7135628461837769, + "num_tokens": 282493150.0, + "step": 11019 + }, + { + "epoch": 1.2101910828025477, + "grad_norm": 2.332669496536255, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7212923765182495, + "num_tokens": 282518695.0, + "step": 11020 + }, + { + "epoch": 1.2103009005051615, + "grad_norm": 2.230398178100586, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6914743781089783, + "num_tokens": 282546486.0, + "step": 11021 + }, + { + "epoch": 1.210410718207775, + "grad_norm": 2.163559675216675, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7450097799301147, + "num_tokens": 282572370.0, + "step": 11022 + }, + { + "epoch": 1.2105205359103888, + "grad_norm": 2.078404426574707, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6972962617874146, + "num_tokens": 282600757.0, + "step": 11023 + }, + { + "epoch": 1.2106303536130025, + "grad_norm": 2.430180311203003, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7218387722969055, + "num_tokens": 282622239.0, + "step": 11024 + }, + { + "epoch": 1.210740171315616, + "grad_norm": 2.256963014602661, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7281362414360046, + "num_tokens": 282646517.0, + "step": 11025 + }, + { + "epoch": 1.2108499890182298, + "grad_norm": 2.3167853355407715, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7166457176208496, + "num_tokens": 282671118.0, + "step": 11026 + }, + { + "epoch": 1.2109598067208434, + "grad_norm": 2.090310573577881, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7086626291275024, + "num_tokens": 282699919.0, + "step": 11027 + }, + { + "epoch": 1.2110696244234571, + "grad_norm": 2.255598783493042, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7285319566726685, + "num_tokens": 282726530.0, + "step": 11028 + }, + { + "epoch": 1.2111794421260706, + "grad_norm": 2.4788331985473633, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7333098649978638, + "num_tokens": 282746379.0, + "step": 11029 + }, + { + "epoch": 1.2112892598286844, + "grad_norm": 2.340596914291382, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7135471701622009, + "num_tokens": 282771167.0, + "step": 11030 + }, + { + "epoch": 1.211399077531298, + "grad_norm": 1.9122623205184937, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6898523569107056, + "num_tokens": 282805985.0, + "step": 11031 + }, + { + "epoch": 1.2115088952339117, + "grad_norm": 2.3417322635650635, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7141532897949219, + "num_tokens": 282830251.0, + "step": 11032 + }, + { + "epoch": 1.2116187129365255, + "grad_norm": 2.246349573135376, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7372195720672607, + "num_tokens": 282854843.0, + "step": 11033 + }, + { + "epoch": 1.211728530639139, + "grad_norm": 2.0043742656707764, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7025009393692017, + "num_tokens": 282888107.0, + "step": 11034 + }, + { + "epoch": 1.2118383483417527, + "grad_norm": 2.160823345184326, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7031945586204529, + "num_tokens": 282915667.0, + "step": 11035 + }, + { + "epoch": 1.2119481660443663, + "grad_norm": 2.3232979774475098, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7178215384483337, + "num_tokens": 282941020.0, + "step": 11036 + }, + { + "epoch": 1.21205798374698, + "grad_norm": 2.1652073860168457, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.715088963508606, + "num_tokens": 282968436.0, + "step": 11037 + }, + { + "epoch": 1.2121678014495938, + "grad_norm": 2.4336090087890625, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.717842161655426, + "num_tokens": 282991255.0, + "step": 11038 + }, + { + "epoch": 1.2122776191522073, + "grad_norm": 2.6588287353515625, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.729040801525116, + "num_tokens": 283011321.0, + "step": 11039 + }, + { + "epoch": 1.212387436854821, + "grad_norm": 2.251248598098755, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7290006279945374, + "num_tokens": 283037118.0, + "step": 11040 + }, + { + "epoch": 1.2124972545574346, + "grad_norm": 2.3755106925964355, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7234139442443848, + "num_tokens": 283060292.0, + "step": 11041 + }, + { + "epoch": 1.2126070722600484, + "grad_norm": 2.2125494480133057, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7066559195518494, + "num_tokens": 283086999.0, + "step": 11042 + }, + { + "epoch": 1.212716889962662, + "grad_norm": 1.9688912630081177, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7297221422195435, + "num_tokens": 283118518.0, + "step": 11043 + }, + { + "epoch": 1.2128267076652757, + "grad_norm": 2.389711380004883, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7086789608001709, + "num_tokens": 283143339.0, + "step": 11044 + }, + { + "epoch": 1.2129365253678892, + "grad_norm": 2.324669361114502, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7341653108596802, + "num_tokens": 283165097.0, + "step": 11045 + }, + { + "epoch": 1.213046343070503, + "grad_norm": 2.343977212905884, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7263315320014954, + "num_tokens": 283186442.0, + "step": 11046 + }, + { + "epoch": 1.2131561607731167, + "grad_norm": 2.014876365661621, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7084834575653076, + "num_tokens": 283215887.0, + "step": 11047 + }, + { + "epoch": 1.2132659784757303, + "grad_norm": 2.184584140777588, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7013056874275208, + "num_tokens": 283242588.0, + "step": 11048 + }, + { + "epoch": 1.213375796178344, + "grad_norm": 2.3067140579223633, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.729327917098999, + "num_tokens": 283267032.0, + "step": 11049 + }, + { + "epoch": 1.2134856138809575, + "grad_norm": 2.837376832962036, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7168328166007996, + "num_tokens": 283284044.0, + "step": 11050 + }, + { + "epoch": 1.2135954315835713, + "grad_norm": 2.419482469558716, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6966598033905029, + "num_tokens": 283307815.0, + "step": 11051 + }, + { + "epoch": 1.2137052492861848, + "grad_norm": 2.317553997039795, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7155150175094604, + "num_tokens": 283333054.0, + "step": 11052 + }, + { + "epoch": 1.2138150669887986, + "grad_norm": 2.189150094985962, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6878734827041626, + "num_tokens": 283361482.0, + "step": 11053 + }, + { + "epoch": 1.2139248846914121, + "grad_norm": 1.8602155447006226, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7186450958251953, + "num_tokens": 283395233.0, + "step": 11054 + }, + { + "epoch": 1.2140347023940259, + "grad_norm": 2.0703272819519043, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7120101451873779, + "num_tokens": 283424700.0, + "step": 11055 + }, + { + "epoch": 1.2141445200966396, + "grad_norm": 2.307657241821289, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7208347320556641, + "num_tokens": 283447840.0, + "step": 11056 + }, + { + "epoch": 1.2142543377992532, + "grad_norm": 2.3345539569854736, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7287560701370239, + "num_tokens": 283472358.0, + "step": 11057 + }, + { + "epoch": 1.214364155501867, + "grad_norm": 2.3711795806884766, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7036943435668945, + "num_tokens": 283495963.0, + "step": 11058 + }, + { + "epoch": 1.2144739732044805, + "grad_norm": 2.373187780380249, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.708573043346405, + "num_tokens": 283519602.0, + "step": 11059 + }, + { + "epoch": 1.2145837909070942, + "grad_norm": 2.0218732357025146, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6911471486091614, + "num_tokens": 283549892.0, + "step": 11060 + }, + { + "epoch": 1.214693608609708, + "grad_norm": 2.3912909030914307, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7232096791267395, + "num_tokens": 283573736.0, + "step": 11061 + }, + { + "epoch": 1.2148034263123215, + "grad_norm": 2.2778327465057373, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.710867166519165, + "num_tokens": 283598825.0, + "step": 11062 + }, + { + "epoch": 1.2149132440149353, + "grad_norm": 2.2859866619110107, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7265271544456482, + "num_tokens": 283622489.0, + "step": 11063 + }, + { + "epoch": 1.2150230617175488, + "grad_norm": 2.509960174560547, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7334157228469849, + "num_tokens": 283644419.0, + "step": 11064 + }, + { + "epoch": 1.2151328794201626, + "grad_norm": 2.1484575271606445, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7223668694496155, + "num_tokens": 283672517.0, + "step": 11065 + }, + { + "epoch": 1.215242697122776, + "grad_norm": 2.4047040939331055, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7092688083648682, + "num_tokens": 283696052.0, + "step": 11066 + }, + { + "epoch": 1.2153525148253899, + "grad_norm": 2.4473161697387695, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7100297212600708, + "num_tokens": 283719079.0, + "step": 11067 + }, + { + "epoch": 1.2154623325280034, + "grad_norm": 2.4627442359924316, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7361173629760742, + "num_tokens": 283741902.0, + "step": 11068 + }, + { + "epoch": 1.2155721502306172, + "grad_norm": 2.169898509979248, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.707782506942749, + "num_tokens": 283770611.0, + "step": 11069 + }, + { + "epoch": 1.215681967933231, + "grad_norm": 2.722181558609009, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7226248979568481, + "num_tokens": 283789194.0, + "step": 11070 + }, + { + "epoch": 1.2157917856358444, + "grad_norm": 2.188778877258301, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7076774835586548, + "num_tokens": 283817192.0, + "step": 11071 + }, + { + "epoch": 1.2159016033384582, + "grad_norm": 2.0589113235473633, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.71501624584198, + "num_tokens": 283845971.0, + "step": 11072 + }, + { + "epoch": 1.2160114210410717, + "grad_norm": 2.3292970657348633, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7267175912857056, + "num_tokens": 283870191.0, + "step": 11073 + }, + { + "epoch": 1.2161212387436855, + "grad_norm": 2.209613800048828, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7159038782119751, + "num_tokens": 283897190.0, + "step": 11074 + }, + { + "epoch": 1.2162310564462993, + "grad_norm": 2.5707662105560303, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7371597290039062, + "num_tokens": 283918889.0, + "step": 11075 + }, + { + "epoch": 1.2163408741489128, + "grad_norm": 2.3570287227630615, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7174050807952881, + "num_tokens": 283942770.0, + "step": 11076 + }, + { + "epoch": 1.2164506918515265, + "grad_norm": 2.5087852478027344, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7404227256774902, + "num_tokens": 283963747.0, + "step": 11077 + }, + { + "epoch": 1.21656050955414, + "grad_norm": 2.1288540363311768, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7340388894081116, + "num_tokens": 283990236.0, + "step": 11078 + }, + { + "epoch": 1.2166703272567538, + "grad_norm": 2.159918785095215, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7355383634567261, + "num_tokens": 284015880.0, + "step": 11079 + }, + { + "epoch": 1.2167801449593674, + "grad_norm": 2.6500473022460938, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7212223410606384, + "num_tokens": 284037102.0, + "step": 11080 + }, + { + "epoch": 1.2168899626619811, + "grad_norm": 2.3774194717407227, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7232979536056519, + "num_tokens": 284062660.0, + "step": 11081 + }, + { + "epoch": 1.2169997803645947, + "grad_norm": 2.552720308303833, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7433747053146362, + "num_tokens": 284082892.0, + "step": 11082 + }, + { + "epoch": 1.2171095980672084, + "grad_norm": 2.427408456802368, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7103269100189209, + "num_tokens": 284107305.0, + "step": 11083 + }, + { + "epoch": 1.2172194157698222, + "grad_norm": 2.211686849594116, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.71068274974823, + "num_tokens": 284133911.0, + "step": 11084 + }, + { + "epoch": 1.2173292334724357, + "grad_norm": 2.3440754413604736, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7086858153343201, + "num_tokens": 284159122.0, + "step": 11085 + }, + { + "epoch": 1.2174390511750495, + "grad_norm": 2.372847318649292, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7194852828979492, + "num_tokens": 284182386.0, + "step": 11086 + }, + { + "epoch": 1.217548868877663, + "grad_norm": 2.4190354347229004, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7170805931091309, + "num_tokens": 284205876.0, + "step": 11087 + }, + { + "epoch": 1.2176586865802768, + "grad_norm": 2.3028504848480225, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7386312484741211, + "num_tokens": 284231165.0, + "step": 11088 + }, + { + "epoch": 1.2177685042828905, + "grad_norm": 2.496901512145996, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7344006299972534, + "num_tokens": 284251487.0, + "step": 11089 + }, + { + "epoch": 1.217878321985504, + "grad_norm": 2.4191534519195557, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7332003712654114, + "num_tokens": 284275695.0, + "step": 11090 + }, + { + "epoch": 1.2179881396881178, + "grad_norm": 2.4933104515075684, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7227975130081177, + "num_tokens": 284296296.0, + "step": 11091 + }, + { + "epoch": 1.2180979573907313, + "grad_norm": 2.3624038696289062, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7324098944664001, + "num_tokens": 284318497.0, + "step": 11092 + }, + { + "epoch": 1.218207775093345, + "grad_norm": 2.200880765914917, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7100948691368103, + "num_tokens": 284344924.0, + "step": 11093 + }, + { + "epoch": 1.2183175927959586, + "grad_norm": 1.8602427244186401, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7030683159828186, + "num_tokens": 284379314.0, + "step": 11094 + }, + { + "epoch": 1.2184274104985724, + "grad_norm": 2.5461087226867676, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7287681102752686, + "num_tokens": 284399538.0, + "step": 11095 + }, + { + "epoch": 1.218537228201186, + "grad_norm": 2.4549403190612793, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.70909583568573, + "num_tokens": 284422907.0, + "step": 11096 + }, + { + "epoch": 1.2186470459037997, + "grad_norm": 2.4326744079589844, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7080658674240112, + "num_tokens": 284446941.0, + "step": 11097 + }, + { + "epoch": 1.2187568636064134, + "grad_norm": 2.3829996585845947, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7068091630935669, + "num_tokens": 284473399.0, + "step": 11098 + }, + { + "epoch": 1.218866681309027, + "grad_norm": 2.501492977142334, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7324709892272949, + "num_tokens": 284493721.0, + "step": 11099 + }, + { + "epoch": 1.2189764990116407, + "grad_norm": 2.5243983268737793, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.732029914855957, + "num_tokens": 284515903.0, + "step": 11100 + }, + { + "epoch": 1.2190863167142543, + "grad_norm": 2.4244909286499023, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7112720012664795, + "num_tokens": 284539205.0, + "step": 11101 + }, + { + "epoch": 1.219196134416868, + "grad_norm": 2.3226611614227295, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7218436002731323, + "num_tokens": 284564451.0, + "step": 11102 + }, + { + "epoch": 1.2193059521194818, + "grad_norm": 2.3419997692108154, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.6994870901107788, + "num_tokens": 284590469.0, + "step": 11103 + }, + { + "epoch": 1.2194157698220953, + "grad_norm": 2.196671962738037, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7191686630249023, + "num_tokens": 284616219.0, + "step": 11104 + }, + { + "epoch": 1.219525587524709, + "grad_norm": 2.2356996536254883, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6923357844352722, + "num_tokens": 284640447.0, + "step": 11105 + }, + { + "epoch": 1.2196354052273226, + "grad_norm": 2.464207887649536, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7024170160293579, + "num_tokens": 284662801.0, + "step": 11106 + }, + { + "epoch": 1.2197452229299364, + "grad_norm": 2.0309598445892334, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7193407416343689, + "num_tokens": 284691884.0, + "step": 11107 + }, + { + "epoch": 1.21985504063255, + "grad_norm": 2.3925888538360596, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7101961374282837, + "num_tokens": 284716405.0, + "step": 11108 + }, + { + "epoch": 1.2199648583351637, + "grad_norm": 2.1307997703552246, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7100917100906372, + "num_tokens": 284746868.0, + "step": 11109 + }, + { + "epoch": 1.2200746760377772, + "grad_norm": 2.621373176574707, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7228267192840576, + "num_tokens": 284766872.0, + "step": 11110 + }, + { + "epoch": 1.220184493740391, + "grad_norm": 2.4059360027313232, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7266923785209656, + "num_tokens": 284789100.0, + "step": 11111 + }, + { + "epoch": 1.2202943114430047, + "grad_norm": 2.6075305938720703, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6959989070892334, + "num_tokens": 284811493.0, + "step": 11112 + }, + { + "epoch": 1.2204041291456182, + "grad_norm": 2.266719341278076, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7345918416976929, + "num_tokens": 284836445.0, + "step": 11113 + }, + { + "epoch": 1.220513946848232, + "grad_norm": 2.209883689880371, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7051530480384827, + "num_tokens": 284864942.0, + "step": 11114 + }, + { + "epoch": 1.2206237645508455, + "grad_norm": 2.177098512649536, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.715255618095398, + "num_tokens": 284891960.0, + "step": 11115 + }, + { + "epoch": 1.2207335822534593, + "grad_norm": 2.5956647396087646, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.719652533531189, + "num_tokens": 284911623.0, + "step": 11116 + }, + { + "epoch": 1.2208433999560728, + "grad_norm": 2.2132985591888428, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7279490232467651, + "num_tokens": 284937711.0, + "step": 11117 + }, + { + "epoch": 1.2209532176586866, + "grad_norm": 2.5571324825286865, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7156201004981995, + "num_tokens": 284959894.0, + "step": 11118 + }, + { + "epoch": 1.2210630353613001, + "grad_norm": 2.3753364086151123, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7342122793197632, + "num_tokens": 284983099.0, + "step": 11119 + }, + { + "epoch": 1.2211728530639139, + "grad_norm": 2.4466142654418945, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7076781988143921, + "num_tokens": 285005907.0, + "step": 11120 + }, + { + "epoch": 1.2212826707665276, + "grad_norm": 2.045605421066284, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7217747569084167, + "num_tokens": 285036540.0, + "step": 11121 + }, + { + "epoch": 1.2213924884691412, + "grad_norm": 2.181063652038574, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7028660774230957, + "num_tokens": 285063719.0, + "step": 11122 + }, + { + "epoch": 1.221502306171755, + "grad_norm": 2.313849925994873, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7260009050369263, + "num_tokens": 285089224.0, + "step": 11123 + }, + { + "epoch": 1.2216121238743685, + "grad_norm": 2.2579967975616455, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7098087072372437, + "num_tokens": 285115409.0, + "step": 11124 + }, + { + "epoch": 1.2217219415769822, + "grad_norm": 2.3310141563415527, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.6998928189277649, + "num_tokens": 285140300.0, + "step": 11125 + }, + { + "epoch": 1.221831759279596, + "grad_norm": 2.090125799179077, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7485175132751465, + "num_tokens": 285168493.0, + "step": 11126 + }, + { + "epoch": 1.2219415769822095, + "grad_norm": 2.140048027038574, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6981087923049927, + "num_tokens": 285198261.0, + "step": 11127 + }, + { + "epoch": 1.2220513946848233, + "grad_norm": 2.341346263885498, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7150160074234009, + "num_tokens": 285222901.0, + "step": 11128 + }, + { + "epoch": 1.2221612123874368, + "grad_norm": 2.1110780239105225, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.704125165939331, + "num_tokens": 285252799.0, + "step": 11129 + }, + { + "epoch": 1.2222710300900506, + "grad_norm": 2.565337657928467, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7135064601898193, + "num_tokens": 285274324.0, + "step": 11130 + }, + { + "epoch": 1.222380847792664, + "grad_norm": 2.48842191696167, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7265918254852295, + "num_tokens": 285296175.0, + "step": 11131 + }, + { + "epoch": 1.2224906654952779, + "grad_norm": 2.2687039375305176, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6892156600952148, + "num_tokens": 285323045.0, + "step": 11132 + }, + { + "epoch": 1.2226004831978914, + "grad_norm": 2.222576141357422, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7349554300308228, + "num_tokens": 285348131.0, + "step": 11133 + }, + { + "epoch": 1.2227103009005051, + "grad_norm": 2.3885326385498047, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7269500494003296, + "num_tokens": 285370551.0, + "step": 11134 + }, + { + "epoch": 1.222820118603119, + "grad_norm": 2.3825571537017822, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7308112382888794, + "num_tokens": 285394042.0, + "step": 11135 + }, + { + "epoch": 1.2229299363057324, + "grad_norm": 2.1585943698883057, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7028873562812805, + "num_tokens": 285422908.0, + "step": 11136 + }, + { + "epoch": 1.2230397540083462, + "grad_norm": 1.860082983970642, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7202773690223694, + "num_tokens": 285456254.0, + "step": 11137 + }, + { + "epoch": 1.2231495717109597, + "grad_norm": 1.9038275480270386, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6886948347091675, + "num_tokens": 285490876.0, + "step": 11138 + }, + { + "epoch": 1.2232593894135735, + "grad_norm": 2.3267464637756348, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7018537521362305, + "num_tokens": 285514110.0, + "step": 11139 + }, + { + "epoch": 1.2233692071161872, + "grad_norm": 2.2039055824279785, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6995460391044617, + "num_tokens": 285540436.0, + "step": 11140 + }, + { + "epoch": 1.2234790248188008, + "grad_norm": 2.3038554191589355, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7230563759803772, + "num_tokens": 285563295.0, + "step": 11141 + }, + { + "epoch": 1.2235888425214145, + "grad_norm": 2.071112871170044, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7031376957893372, + "num_tokens": 285592086.0, + "step": 11142 + }, + { + "epoch": 1.223698660224028, + "grad_norm": 2.2605724334716797, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7061092257499695, + "num_tokens": 285616878.0, + "step": 11143 + }, + { + "epoch": 1.2238084779266418, + "grad_norm": 2.1468708515167236, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7256519794464111, + "num_tokens": 285645028.0, + "step": 11144 + }, + { + "epoch": 1.2239182956292554, + "grad_norm": 2.2877578735351562, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7187063097953796, + "num_tokens": 285670709.0, + "step": 11145 + }, + { + "epoch": 1.2240281133318691, + "grad_norm": 2.5263803005218506, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7251905798912048, + "num_tokens": 285693836.0, + "step": 11146 + }, + { + "epoch": 1.2241379310344827, + "grad_norm": 2.032926082611084, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7381081581115723, + "num_tokens": 285720834.0, + "step": 11147 + }, + { + "epoch": 1.2242477487370964, + "grad_norm": 2.2560486793518066, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.744900107383728, + "num_tokens": 285744594.0, + "step": 11148 + }, + { + "epoch": 1.2243575664397102, + "grad_norm": 2.4357316493988037, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7148808240890503, + "num_tokens": 285769197.0, + "step": 11149 + }, + { + "epoch": 1.2244673841423237, + "grad_norm": 2.231294870376587, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7206231355667114, + "num_tokens": 285794978.0, + "step": 11150 + }, + { + "epoch": 1.2245772018449375, + "grad_norm": 2.2685086727142334, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7095376253128052, + "num_tokens": 285821234.0, + "step": 11151 + }, + { + "epoch": 1.224687019547551, + "grad_norm": 1.9896661043167114, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7203477621078491, + "num_tokens": 285852426.0, + "step": 11152 + }, + { + "epoch": 1.2247968372501647, + "grad_norm": 2.2855958938598633, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7272738218307495, + "num_tokens": 285876353.0, + "step": 11153 + }, + { + "epoch": 1.2249066549527785, + "grad_norm": 2.313735008239746, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.723076581954956, + "num_tokens": 285902735.0, + "step": 11154 + }, + { + "epoch": 1.225016472655392, + "grad_norm": 2.080021619796753, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7171952724456787, + "num_tokens": 285931773.0, + "step": 11155 + }, + { + "epoch": 1.2251262903580058, + "grad_norm": 2.2704198360443115, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6975042819976807, + "num_tokens": 285956983.0, + "step": 11156 + }, + { + "epoch": 1.2252361080606193, + "grad_norm": 2.2832393646240234, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.699383020401001, + "num_tokens": 285982385.0, + "step": 11157 + }, + { + "epoch": 1.225345925763233, + "grad_norm": 2.248011827468872, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7193011045455933, + "num_tokens": 286009214.0, + "step": 11158 + }, + { + "epoch": 1.2254557434658466, + "grad_norm": 2.190723180770874, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7111606597900391, + "num_tokens": 286034694.0, + "step": 11159 + }, + { + "epoch": 1.2255655611684604, + "grad_norm": 2.3374826908111572, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7349228262901306, + "num_tokens": 286057958.0, + "step": 11160 + }, + { + "epoch": 1.225675378871074, + "grad_norm": 2.2964138984680176, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7339135408401489, + "num_tokens": 286080250.0, + "step": 11161 + }, + { + "epoch": 1.2257851965736877, + "grad_norm": 2.1912379264831543, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7166274785995483, + "num_tokens": 286105484.0, + "step": 11162 + }, + { + "epoch": 1.2258950142763014, + "grad_norm": 2.2472949028015137, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7255090475082397, + "num_tokens": 286130391.0, + "step": 11163 + }, + { + "epoch": 1.226004831978915, + "grad_norm": 2.506298542022705, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7366619110107422, + "num_tokens": 286151026.0, + "step": 11164 + }, + { + "epoch": 1.2261146496815287, + "grad_norm": 2.5050547122955322, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7126574516296387, + "num_tokens": 286171828.0, + "step": 11165 + }, + { + "epoch": 1.2262244673841423, + "grad_norm": 2.122847080230713, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7170533537864685, + "num_tokens": 286201528.0, + "step": 11166 + }, + { + "epoch": 1.226334285086756, + "grad_norm": 2.630971670150757, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7231569290161133, + "num_tokens": 286223384.0, + "step": 11167 + }, + { + "epoch": 1.2264441027893696, + "grad_norm": 2.028691053390503, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.722206711769104, + "num_tokens": 286252252.0, + "step": 11168 + }, + { + "epoch": 1.2265539204919833, + "grad_norm": 2.0660252571105957, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.705303430557251, + "num_tokens": 286281341.0, + "step": 11169 + }, + { + "epoch": 1.226663738194597, + "grad_norm": 2.2023680210113525, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7098425626754761, + "num_tokens": 286307498.0, + "step": 11170 + }, + { + "epoch": 1.2267735558972106, + "grad_norm": 2.3387224674224854, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7250017523765564, + "num_tokens": 286331898.0, + "step": 11171 + }, + { + "epoch": 1.2268833735998244, + "grad_norm": 2.217134952545166, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7192911505699158, + "num_tokens": 286359679.0, + "step": 11172 + }, + { + "epoch": 1.226993191302438, + "grad_norm": 2.090646982192993, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6965826749801636, + "num_tokens": 286389772.0, + "step": 11173 + }, + { + "epoch": 1.2271030090050516, + "grad_norm": 2.264943838119507, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7169047594070435, + "num_tokens": 286416074.0, + "step": 11174 + }, + { + "epoch": 1.2272128267076652, + "grad_norm": 2.3752236366271973, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7151011824607849, + "num_tokens": 286441355.0, + "step": 11175 + }, + { + "epoch": 1.227322644410279, + "grad_norm": 1.9714704751968384, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7281505465507507, + "num_tokens": 286471593.0, + "step": 11176 + }, + { + "epoch": 1.2274324621128927, + "grad_norm": 2.2331347465515137, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7043956518173218, + "num_tokens": 286498871.0, + "step": 11177 + }, + { + "epoch": 1.2275422798155062, + "grad_norm": 2.2477834224700928, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.706101655960083, + "num_tokens": 286528107.0, + "step": 11178 + }, + { + "epoch": 1.22765209751812, + "grad_norm": 2.134434461593628, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6986067891120911, + "num_tokens": 286558727.0, + "step": 11179 + }, + { + "epoch": 1.2277619152207335, + "grad_norm": 2.200129270553589, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7242946624755859, + "num_tokens": 286584344.0, + "step": 11180 + }, + { + "epoch": 1.2278717329233473, + "grad_norm": 2.0663270950317383, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7033019661903381, + "num_tokens": 286614441.0, + "step": 11181 + }, + { + "epoch": 1.2279815506259608, + "grad_norm": 2.432326078414917, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.71703040599823, + "num_tokens": 286637376.0, + "step": 11182 + }, + { + "epoch": 1.2280913683285746, + "grad_norm": 2.232907295227051, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7083047032356262, + "num_tokens": 286663588.0, + "step": 11183 + }, + { + "epoch": 1.228201186031188, + "grad_norm": 2.350893259048462, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7051109671592712, + "num_tokens": 286688368.0, + "step": 11184 + }, + { + "epoch": 1.2283110037338019, + "grad_norm": 2.6342458724975586, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7245474457740784, + "num_tokens": 286707343.0, + "step": 11185 + }, + { + "epoch": 1.2284208214364156, + "grad_norm": 2.0516104698181152, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7216037511825562, + "num_tokens": 286737380.0, + "step": 11186 + }, + { + "epoch": 1.2285306391390292, + "grad_norm": 2.308887243270874, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7121424078941345, + "num_tokens": 286761488.0, + "step": 11187 + }, + { + "epoch": 1.228640456841643, + "grad_norm": 2.498887062072754, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7037603855133057, + "num_tokens": 286784748.0, + "step": 11188 + }, + { + "epoch": 1.2287502745442564, + "grad_norm": 2.504422187805176, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7231457829475403, + "num_tokens": 286805796.0, + "step": 11189 + }, + { + "epoch": 1.2288600922468702, + "grad_norm": 2.258409023284912, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7091130018234253, + "num_tokens": 286831643.0, + "step": 11190 + }, + { + "epoch": 1.228969909949484, + "grad_norm": 2.388772964477539, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7104688286781311, + "num_tokens": 286856102.0, + "step": 11191 + }, + { + "epoch": 1.2290797276520975, + "grad_norm": 2.273956060409546, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7172050476074219, + "num_tokens": 286881554.0, + "step": 11192 + }, + { + "epoch": 1.2291895453547113, + "grad_norm": 2.2667574882507324, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6988766193389893, + "num_tokens": 286906792.0, + "step": 11193 + }, + { + "epoch": 1.2292993630573248, + "grad_norm": 2.1651883125305176, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7345114946365356, + "num_tokens": 286932424.0, + "step": 11194 + }, + { + "epoch": 1.2294091807599385, + "grad_norm": 2.052953004837036, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7182083129882812, + "num_tokens": 286959082.0, + "step": 11195 + }, + { + "epoch": 1.229518998462552, + "grad_norm": 2.2716310024261475, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7042394876480103, + "num_tokens": 286984763.0, + "step": 11196 + }, + { + "epoch": 1.2296288161651658, + "grad_norm": 2.2228903770446777, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7258434295654297, + "num_tokens": 287011256.0, + "step": 11197 + }, + { + "epoch": 1.2297386338677794, + "grad_norm": 2.206747531890869, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7095960378646851, + "num_tokens": 287037441.0, + "step": 11198 + }, + { + "epoch": 1.2298484515703931, + "grad_norm": 2.575587034225464, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7174772024154663, + "num_tokens": 287058271.0, + "step": 11199 + }, + { + "epoch": 1.2299582692730069, + "grad_norm": 2.2984256744384766, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7157387733459473, + "num_tokens": 287082272.0, + "step": 11200 + }, + { + "epoch": 1.2300680869756204, + "grad_norm": 2.1636738777160645, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7161272764205933, + "num_tokens": 287111225.0, + "step": 11201 + }, + { + "epoch": 1.2301779046782342, + "grad_norm": 2.3053722381591797, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.704369068145752, + "num_tokens": 287134728.0, + "step": 11202 + }, + { + "epoch": 1.2302877223808477, + "grad_norm": 2.2161977291107178, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.717154324054718, + "num_tokens": 287161099.0, + "step": 11203 + }, + { + "epoch": 1.2303975400834615, + "grad_norm": 2.3286967277526855, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7115365266799927, + "num_tokens": 287185156.0, + "step": 11204 + }, + { + "epoch": 1.2305073577860752, + "grad_norm": 2.0054829120635986, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6984997987747192, + "num_tokens": 287220056.0, + "step": 11205 + }, + { + "epoch": 1.2306171754886888, + "grad_norm": 2.2761013507843018, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7325839400291443, + "num_tokens": 287244818.0, + "step": 11206 + }, + { + "epoch": 1.2307269931913025, + "grad_norm": 2.398604393005371, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7004464268684387, + "num_tokens": 287268880.0, + "step": 11207 + }, + { + "epoch": 1.230836810893916, + "grad_norm": 2.023216485977173, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7075601816177368, + "num_tokens": 287299381.0, + "step": 11208 + }, + { + "epoch": 1.2309466285965298, + "grad_norm": 2.1014797687530518, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.736748218536377, + "num_tokens": 287325645.0, + "step": 11209 + }, + { + "epoch": 1.2310564462991433, + "grad_norm": 2.2610909938812256, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7069172859191895, + "num_tokens": 287352030.0, + "step": 11210 + }, + { + "epoch": 1.231166264001757, + "grad_norm": 2.6045403480529785, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.715475857257843, + "num_tokens": 287371007.0, + "step": 11211 + }, + { + "epoch": 1.2312760817043706, + "grad_norm": 2.3197739124298096, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7399201393127441, + "num_tokens": 287394706.0, + "step": 11212 + }, + { + "epoch": 1.2313858994069844, + "grad_norm": 2.3465158939361572, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7298060655593872, + "num_tokens": 287418932.0, + "step": 11213 + }, + { + "epoch": 1.2314957171095982, + "grad_norm": 2.1189992427825928, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6977738738059998, + "num_tokens": 287446703.0, + "step": 11214 + }, + { + "epoch": 1.2316055348122117, + "grad_norm": 2.1786630153656006, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6980478763580322, + "num_tokens": 287473374.0, + "step": 11215 + }, + { + "epoch": 1.2317153525148254, + "grad_norm": 2.3522589206695557, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7083582282066345, + "num_tokens": 287496990.0, + "step": 11216 + }, + { + "epoch": 1.231825170217439, + "grad_norm": 2.011878252029419, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.719926118850708, + "num_tokens": 287529030.0, + "step": 11217 + }, + { + "epoch": 1.2319349879200527, + "grad_norm": 2.3071377277374268, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7278857231140137, + "num_tokens": 287551630.0, + "step": 11218 + }, + { + "epoch": 1.2320448056226665, + "grad_norm": 2.509409189224243, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7288763523101807, + "num_tokens": 287572927.0, + "step": 11219 + }, + { + "epoch": 1.23215462332528, + "grad_norm": 2.038520097732544, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7301062941551208, + "num_tokens": 287601615.0, + "step": 11220 + }, + { + "epoch": 1.2322644410278938, + "grad_norm": 2.4456400871276855, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7288423180580139, + "num_tokens": 287623364.0, + "step": 11221 + }, + { + "epoch": 1.2323742587305073, + "grad_norm": 2.4408798217773438, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7405074834823608, + "num_tokens": 287647170.0, + "step": 11222 + }, + { + "epoch": 1.232484076433121, + "grad_norm": 2.7623918056488037, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7339082956314087, + "num_tokens": 287665332.0, + "step": 11223 + }, + { + "epoch": 1.2325938941357346, + "grad_norm": 2.3273262977600098, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7085633277893066, + "num_tokens": 287690478.0, + "step": 11224 + }, + { + "epoch": 1.2327037118383484, + "grad_norm": 2.059404134750366, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7181167006492615, + "num_tokens": 287719958.0, + "step": 11225 + }, + { + "epoch": 1.232813529540962, + "grad_norm": 2.1451802253723145, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.712225079536438, + "num_tokens": 287750661.0, + "step": 11226 + }, + { + "epoch": 1.2329233472435757, + "grad_norm": 2.296565294265747, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6938843131065369, + "num_tokens": 287776864.0, + "step": 11227 + }, + { + "epoch": 1.2330331649461894, + "grad_norm": 2.194286346435547, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7198930978775024, + "num_tokens": 287802149.0, + "step": 11228 + }, + { + "epoch": 1.233142982648803, + "grad_norm": 2.208847761154175, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7127560377120972, + "num_tokens": 287828307.0, + "step": 11229 + }, + { + "epoch": 1.2332528003514167, + "grad_norm": 2.6822612285614014, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7376233339309692, + "num_tokens": 287847406.0, + "step": 11230 + }, + { + "epoch": 1.2333626180540302, + "grad_norm": 2.549966812133789, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7036396265029907, + "num_tokens": 287868485.0, + "step": 11231 + }, + { + "epoch": 1.233472435756644, + "grad_norm": 2.2518935203552246, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7186232805252075, + "num_tokens": 287893879.0, + "step": 11232 + }, + { + "epoch": 1.2335822534592575, + "grad_norm": 2.0529439449310303, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7066510915756226, + "num_tokens": 287923350.0, + "step": 11233 + }, + { + "epoch": 1.2336920711618713, + "grad_norm": 2.3198487758636475, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7332356572151184, + "num_tokens": 287949273.0, + "step": 11234 + }, + { + "epoch": 1.2338018888644848, + "grad_norm": 2.200659990310669, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7263067960739136, + "num_tokens": 287975425.0, + "step": 11235 + }, + { + "epoch": 1.2339117065670986, + "grad_norm": 2.52382493019104, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6963934898376465, + "num_tokens": 287999088.0, + "step": 11236 + }, + { + "epoch": 1.2340215242697123, + "grad_norm": 2.399470806121826, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7017743587493896, + "num_tokens": 288024726.0, + "step": 11237 + }, + { + "epoch": 1.2341313419723259, + "grad_norm": 2.0991933345794678, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7111014723777771, + "num_tokens": 288054812.0, + "step": 11238 + }, + { + "epoch": 1.2342411596749396, + "grad_norm": 2.370366334915161, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7153005599975586, + "num_tokens": 288078062.0, + "step": 11239 + }, + { + "epoch": 1.2343509773775532, + "grad_norm": 2.765570640563965, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7234369516372681, + "num_tokens": 288095281.0, + "step": 11240 + }, + { + "epoch": 1.234460795080167, + "grad_norm": 2.3907864093780518, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7057477235794067, + "num_tokens": 288120343.0, + "step": 11241 + }, + { + "epoch": 1.2345706127827807, + "grad_norm": 2.067955255508423, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7242346405982971, + "num_tokens": 288148042.0, + "step": 11242 + }, + { + "epoch": 1.2346804304853942, + "grad_norm": 2.289435386657715, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7075939178466797, + "num_tokens": 288174816.0, + "step": 11243 + }, + { + "epoch": 1.234790248188008, + "grad_norm": 2.347280263900757, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6980671882629395, + "num_tokens": 288198347.0, + "step": 11244 + }, + { + "epoch": 1.2349000658906215, + "grad_norm": 2.316693067550659, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7274808883666992, + "num_tokens": 288220975.0, + "step": 11245 + }, + { + "epoch": 1.2350098835932353, + "grad_norm": 2.231722116470337, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7262714505195618, + "num_tokens": 288246565.0, + "step": 11246 + }, + { + "epoch": 1.2351197012958488, + "grad_norm": 2.041898250579834, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7084863185882568, + "num_tokens": 288277647.0, + "step": 11247 + }, + { + "epoch": 1.2352295189984626, + "grad_norm": 2.116942882537842, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7033963203430176, + "num_tokens": 288304679.0, + "step": 11248 + }, + { + "epoch": 1.235339336701076, + "grad_norm": 2.284743547439575, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7136765718460083, + "num_tokens": 288328506.0, + "step": 11249 + }, + { + "epoch": 1.2354491544036899, + "grad_norm": 2.2678627967834473, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7229769229888916, + "num_tokens": 288353690.0, + "step": 11250 + }, + { + "epoch": 1.2355589721063036, + "grad_norm": 2.0400211811065674, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7296081781387329, + "num_tokens": 288384994.0, + "step": 11251 + }, + { + "epoch": 1.2356687898089171, + "grad_norm": 2.4453630447387695, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6951294541358948, + "num_tokens": 288408237.0, + "step": 11252 + }, + { + "epoch": 1.235778607511531, + "grad_norm": 2.043062686920166, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7064054012298584, + "num_tokens": 288438224.0, + "step": 11253 + }, + { + "epoch": 1.2358884252141444, + "grad_norm": 2.260751485824585, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7016313076019287, + "num_tokens": 288463672.0, + "step": 11254 + }, + { + "epoch": 1.2359982429167582, + "grad_norm": 2.4090826511383057, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.708932638168335, + "num_tokens": 288488480.0, + "step": 11255 + }, + { + "epoch": 1.236108060619372, + "grad_norm": 2.688551425933838, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7300704717636108, + "num_tokens": 288506445.0, + "step": 11256 + }, + { + "epoch": 1.2362178783219855, + "grad_norm": 2.2553675174713135, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7101765871047974, + "num_tokens": 288531027.0, + "step": 11257 + }, + { + "epoch": 1.2363276960245992, + "grad_norm": 2.1211252212524414, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7108003497123718, + "num_tokens": 288559662.0, + "step": 11258 + }, + { + "epoch": 1.2364375137272128, + "grad_norm": 2.2430012226104736, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7142783999443054, + "num_tokens": 288585360.0, + "step": 11259 + }, + { + "epoch": 1.2365473314298265, + "grad_norm": 2.1933791637420654, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.709381103515625, + "num_tokens": 288611373.0, + "step": 11260 + }, + { + "epoch": 1.23665714913244, + "grad_norm": 2.2412052154541016, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7035619020462036, + "num_tokens": 288637970.0, + "step": 11261 + }, + { + "epoch": 1.2367669668350538, + "grad_norm": 2.1940932273864746, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7157654762268066, + "num_tokens": 288667427.0, + "step": 11262 + }, + { + "epoch": 1.2368767845376674, + "grad_norm": 2.29892635345459, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7093614935874939, + "num_tokens": 288696171.0, + "step": 11263 + }, + { + "epoch": 1.2369866022402811, + "grad_norm": 2.3281941413879395, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7340665459632874, + "num_tokens": 288720309.0, + "step": 11264 + }, + { + "epoch": 1.2370964199428949, + "grad_norm": 2.1437795162200928, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6965101957321167, + "num_tokens": 288748596.0, + "step": 11265 + }, + { + "epoch": 1.2372062376455084, + "grad_norm": 2.353026866912842, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7616288661956787, + "num_tokens": 288771225.0, + "step": 11266 + }, + { + "epoch": 1.2373160553481222, + "grad_norm": 2.3687822818756104, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7154399156570435, + "num_tokens": 288794020.0, + "step": 11267 + }, + { + "epoch": 1.2374258730507357, + "grad_norm": 2.344805955886841, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.722620964050293, + "num_tokens": 288817283.0, + "step": 11268 + }, + { + "epoch": 1.2375356907533495, + "grad_norm": 2.282792806625366, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7212156653404236, + "num_tokens": 288843153.0, + "step": 11269 + }, + { + "epoch": 1.2376455084559632, + "grad_norm": 2.2526350021362305, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7197108268737793, + "num_tokens": 288867714.0, + "step": 11270 + }, + { + "epoch": 1.2377553261585768, + "grad_norm": 2.181353807449341, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7093318700790405, + "num_tokens": 288894625.0, + "step": 11271 + }, + { + "epoch": 1.2378651438611905, + "grad_norm": 2.4303672313690186, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7143371105194092, + "num_tokens": 288917834.0, + "step": 11272 + }, + { + "epoch": 1.237974961563804, + "grad_norm": 2.2590253353118896, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.724694013595581, + "num_tokens": 288943382.0, + "step": 11273 + }, + { + "epoch": 1.2380847792664178, + "grad_norm": 2.459526300430298, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7222226858139038, + "num_tokens": 288964727.0, + "step": 11274 + }, + { + "epoch": 1.2381945969690313, + "grad_norm": 2.373499631881714, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7275867462158203, + "num_tokens": 288986482.0, + "step": 11275 + }, + { + "epoch": 1.238304414671645, + "grad_norm": 2.141793727874756, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7112488746643066, + "num_tokens": 289013763.0, + "step": 11276 + }, + { + "epoch": 1.2384142323742586, + "grad_norm": 2.208889961242676, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7186897993087769, + "num_tokens": 289039826.0, + "step": 11277 + }, + { + "epoch": 1.2385240500768724, + "grad_norm": 2.6342976093292236, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7044431567192078, + "num_tokens": 289059885.0, + "step": 11278 + }, + { + "epoch": 1.2386338677794861, + "grad_norm": 2.603539228439331, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7436877489089966, + "num_tokens": 289079468.0, + "step": 11279 + }, + { + "epoch": 1.2387436854820997, + "grad_norm": 2.491603136062622, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7210259437561035, + "num_tokens": 289100823.0, + "step": 11280 + }, + { + "epoch": 1.2388535031847134, + "grad_norm": 2.056443214416504, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7142585515975952, + "num_tokens": 289132319.0, + "step": 11281 + }, + { + "epoch": 1.238963320887327, + "grad_norm": 1.9258642196655273, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7156717777252197, + "num_tokens": 289163259.0, + "step": 11282 + }, + { + "epoch": 1.2390731385899407, + "grad_norm": 2.4797115325927734, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7113704085350037, + "num_tokens": 289184346.0, + "step": 11283 + }, + { + "epoch": 1.2391829562925545, + "grad_norm": 2.219684362411499, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7234957218170166, + "num_tokens": 289209796.0, + "step": 11284 + }, + { + "epoch": 1.239292773995168, + "grad_norm": 2.36133074760437, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7066754102706909, + "num_tokens": 289235427.0, + "step": 11285 + }, + { + "epoch": 1.2394025916977818, + "grad_norm": 2.142751932144165, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6768662333488464, + "num_tokens": 289265898.0, + "step": 11286 + }, + { + "epoch": 1.2395124094003953, + "grad_norm": 2.255388021469116, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7074683904647827, + "num_tokens": 289291131.0, + "step": 11287 + }, + { + "epoch": 1.239622227103009, + "grad_norm": 2.3959646224975586, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7202900648117065, + "num_tokens": 289314240.0, + "step": 11288 + }, + { + "epoch": 1.2397320448056226, + "grad_norm": 2.1865453720092773, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.71429044008255, + "num_tokens": 289341267.0, + "step": 11289 + }, + { + "epoch": 1.2398418625082364, + "grad_norm": 2.2840189933776855, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7250064611434937, + "num_tokens": 289370062.0, + "step": 11290 + }, + { + "epoch": 1.23995168021085, + "grad_norm": 2.029388427734375, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7377256751060486, + "num_tokens": 289398208.0, + "step": 11291 + }, + { + "epoch": 1.2400614979134637, + "grad_norm": 2.3918211460113525, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7267688512802124, + "num_tokens": 289421559.0, + "step": 11292 + }, + { + "epoch": 1.2401713156160774, + "grad_norm": 2.2713420391082764, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7265554666519165, + "num_tokens": 289447242.0, + "step": 11293 + }, + { + "epoch": 1.240281133318691, + "grad_norm": 2.3762245178222656, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7050148248672485, + "num_tokens": 289473096.0, + "step": 11294 + }, + { + "epoch": 1.2403909510213047, + "grad_norm": 2.2246170043945312, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6918282508850098, + "num_tokens": 289501578.0, + "step": 11295 + }, + { + "epoch": 1.2405007687239182, + "grad_norm": 2.4361507892608643, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7274651527404785, + "num_tokens": 289524930.0, + "step": 11296 + }, + { + "epoch": 1.240610586426532, + "grad_norm": 2.398397207260132, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7481821775436401, + "num_tokens": 289549876.0, + "step": 11297 + }, + { + "epoch": 1.2407204041291455, + "grad_norm": 2.0020103454589844, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.706276535987854, + "num_tokens": 289581303.0, + "step": 11298 + }, + { + "epoch": 1.2408302218317593, + "grad_norm": 2.2612671852111816, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7132216691970825, + "num_tokens": 289605845.0, + "step": 11299 + }, + { + "epoch": 1.2409400395343728, + "grad_norm": 2.245872735977173, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7293515205383301, + "num_tokens": 289632157.0, + "step": 11300 + }, + { + "epoch": 1.2410498572369866, + "grad_norm": 2.6572608947753906, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7205843329429626, + "num_tokens": 289652503.0, + "step": 11301 + }, + { + "epoch": 1.2411596749396003, + "grad_norm": 2.328521490097046, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7283278107643127, + "num_tokens": 289674977.0, + "step": 11302 + }, + { + "epoch": 1.2412694926422139, + "grad_norm": 2.238375425338745, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.7012910842895508, + "num_tokens": 289704288.0, + "step": 11303 + }, + { + "epoch": 1.2413793103448276, + "grad_norm": 2.1814584732055664, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.693015456199646, + "num_tokens": 289732528.0, + "step": 11304 + }, + { + "epoch": 1.2414891280474412, + "grad_norm": 2.3672738075256348, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7170436382293701, + "num_tokens": 289755618.0, + "step": 11305 + }, + { + "epoch": 1.241598945750055, + "grad_norm": 2.407317638397217, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7389131188392639, + "num_tokens": 289777426.0, + "step": 11306 + }, + { + "epoch": 1.2417087634526687, + "grad_norm": 2.3987226486206055, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.717099666595459, + "num_tokens": 289799548.0, + "step": 11307 + }, + { + "epoch": 1.2418185811552822, + "grad_norm": 2.3296303749084473, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7328755855560303, + "num_tokens": 289823223.0, + "step": 11308 + }, + { + "epoch": 1.241928398857896, + "grad_norm": 2.1462628841400146, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7201638221740723, + "num_tokens": 289849802.0, + "step": 11309 + }, + { + "epoch": 1.2420382165605095, + "grad_norm": 2.1979315280914307, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7212725877761841, + "num_tokens": 289875695.0, + "step": 11310 + }, + { + "epoch": 1.2421480342631233, + "grad_norm": 2.4979989528656006, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7033559083938599, + "num_tokens": 289898476.0, + "step": 11311 + }, + { + "epoch": 1.2422578519657368, + "grad_norm": 2.3032760620117188, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7195594906806946, + "num_tokens": 289923395.0, + "step": 11312 + }, + { + "epoch": 1.2423676696683505, + "grad_norm": 2.2330877780914307, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7324614524841309, + "num_tokens": 289951428.0, + "step": 11313 + }, + { + "epoch": 1.242477487370964, + "grad_norm": 1.9635096788406372, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.705450177192688, + "num_tokens": 289979954.0, + "step": 11314 + }, + { + "epoch": 1.2425873050735778, + "grad_norm": 2.377166986465454, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7279183864593506, + "num_tokens": 290002714.0, + "step": 11315 + }, + { + "epoch": 1.2426971227761916, + "grad_norm": 2.1703927516937256, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7196686267852783, + "num_tokens": 290030130.0, + "step": 11316 + }, + { + "epoch": 1.2428069404788051, + "grad_norm": 2.2654645442962646, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7421483993530273, + "num_tokens": 290052983.0, + "step": 11317 + }, + { + "epoch": 1.242916758181419, + "grad_norm": 2.345580577850342, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7167044878005981, + "num_tokens": 290075824.0, + "step": 11318 + }, + { + "epoch": 1.2430265758840324, + "grad_norm": 2.2405173778533936, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7418687343597412, + "num_tokens": 290099451.0, + "step": 11319 + }, + { + "epoch": 1.2431363935866462, + "grad_norm": 2.005455493927002, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6976089477539062, + "num_tokens": 290130922.0, + "step": 11320 + }, + { + "epoch": 1.24324621128926, + "grad_norm": 2.1901462078094482, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7236382961273193, + "num_tokens": 290156089.0, + "step": 11321 + }, + { + "epoch": 1.2433560289918735, + "grad_norm": 2.6721365451812744, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.726703405380249, + "num_tokens": 290176033.0, + "step": 11322 + }, + { + "epoch": 1.2434658466944872, + "grad_norm": 2.0628113746643066, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.724433183670044, + "num_tokens": 290206467.0, + "step": 11323 + }, + { + "epoch": 1.2435756643971008, + "grad_norm": 2.0728800296783447, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.6999120712280273, + "num_tokens": 290236326.0, + "step": 11324 + }, + { + "epoch": 1.2436854820997145, + "grad_norm": 2.423760414123535, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7209010720252991, + "num_tokens": 290259451.0, + "step": 11325 + }, + { + "epoch": 1.243795299802328, + "grad_norm": 2.362442970275879, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7322116494178772, + "num_tokens": 290288620.0, + "step": 11326 + }, + { + "epoch": 1.2439051175049418, + "grad_norm": 2.236205577850342, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7174652814865112, + "num_tokens": 290315755.0, + "step": 11327 + }, + { + "epoch": 1.2440149352075554, + "grad_norm": 2.2736692428588867, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7248607873916626, + "num_tokens": 290339601.0, + "step": 11328 + }, + { + "epoch": 1.244124752910169, + "grad_norm": 2.2969512939453125, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7154927849769592, + "num_tokens": 290364325.0, + "step": 11329 + }, + { + "epoch": 1.2442345706127829, + "grad_norm": 2.234541416168213, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7285151481628418, + "num_tokens": 290390897.0, + "step": 11330 + }, + { + "epoch": 1.2443443883153964, + "grad_norm": 2.59379243850708, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7418434619903564, + "num_tokens": 290411904.0, + "step": 11331 + }, + { + "epoch": 1.2444542060180102, + "grad_norm": 2.1656579971313477, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7153376340866089, + "num_tokens": 290439615.0, + "step": 11332 + }, + { + "epoch": 1.2445640237206237, + "grad_norm": 2.2033329010009766, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7246483564376831, + "num_tokens": 290467090.0, + "step": 11333 + }, + { + "epoch": 1.2446738414232374, + "grad_norm": 2.2867043018341064, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7258703708648682, + "num_tokens": 290490544.0, + "step": 11334 + }, + { + "epoch": 1.2447836591258512, + "grad_norm": 2.267988920211792, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7204962968826294, + "num_tokens": 290517588.0, + "step": 11335 + }, + { + "epoch": 1.2448934768284647, + "grad_norm": 2.2782747745513916, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.726957380771637, + "num_tokens": 290543624.0, + "step": 11336 + }, + { + "epoch": 1.2450032945310785, + "grad_norm": 2.1696407794952393, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7276219725608826, + "num_tokens": 290570740.0, + "step": 11337 + }, + { + "epoch": 1.245113112233692, + "grad_norm": 2.274834632873535, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7039347887039185, + "num_tokens": 290596327.0, + "step": 11338 + }, + { + "epoch": 1.2452229299363058, + "grad_norm": 2.332935333251953, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7264297008514404, + "num_tokens": 290617569.0, + "step": 11339 + }, + { + "epoch": 1.2453327476389193, + "grad_norm": 2.364779233932495, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7264488935470581, + "num_tokens": 290641756.0, + "step": 11340 + }, + { + "epoch": 1.245442565341533, + "grad_norm": 2.1391241550445557, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7069782018661499, + "num_tokens": 290668382.0, + "step": 11341 + }, + { + "epoch": 1.2455523830441466, + "grad_norm": 2.0125229358673096, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7005840539932251, + "num_tokens": 290697888.0, + "step": 11342 + }, + { + "epoch": 1.2456622007467604, + "grad_norm": 2.623959541320801, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7484846711158752, + "num_tokens": 290717605.0, + "step": 11343 + }, + { + "epoch": 1.2457720184493741, + "grad_norm": 2.583040714263916, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6993722915649414, + "num_tokens": 290740795.0, + "step": 11344 + }, + { + "epoch": 1.2458818361519877, + "grad_norm": 2.243661642074585, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7138344049453735, + "num_tokens": 290766405.0, + "step": 11345 + }, + { + "epoch": 1.2459916538546014, + "grad_norm": 2.370356798171997, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.713687539100647, + "num_tokens": 290789187.0, + "step": 11346 + }, + { + "epoch": 1.246101471557215, + "grad_norm": 2.3926010131835938, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7380934357643127, + "num_tokens": 290812758.0, + "step": 11347 + }, + { + "epoch": 1.2462112892598287, + "grad_norm": 2.1550190448760986, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7235379219055176, + "num_tokens": 290840109.0, + "step": 11348 + }, + { + "epoch": 1.2463211069624422, + "grad_norm": 2.3174281120300293, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6952371597290039, + "num_tokens": 290866226.0, + "step": 11349 + }, + { + "epoch": 1.246430924665056, + "grad_norm": 2.0619804859161377, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.70206618309021, + "num_tokens": 290896241.0, + "step": 11350 + }, + { + "epoch": 1.2465407423676698, + "grad_norm": 2.146552801132202, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7177778482437134, + "num_tokens": 290923083.0, + "step": 11351 + }, + { + "epoch": 1.2466505600702833, + "grad_norm": 2.183011054992676, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7073512673377991, + "num_tokens": 290949542.0, + "step": 11352 + }, + { + "epoch": 1.246760377772897, + "grad_norm": 2.2414894104003906, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7163843512535095, + "num_tokens": 290975131.0, + "step": 11353 + }, + { + "epoch": 1.2468701954755106, + "grad_norm": 2.331697702407837, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7163658142089844, + "num_tokens": 290999040.0, + "step": 11354 + }, + { + "epoch": 1.2469800131781243, + "grad_norm": 2.5327277183532715, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.732637345790863, + "num_tokens": 291018851.0, + "step": 11355 + }, + { + "epoch": 1.2470898308807379, + "grad_norm": 2.1533150672912598, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7063790559768677, + "num_tokens": 291047849.0, + "step": 11356 + }, + { + "epoch": 1.2471996485833516, + "grad_norm": 1.9811242818832397, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6986732482910156, + "num_tokens": 291081181.0, + "step": 11357 + }, + { + "epoch": 1.2473094662859654, + "grad_norm": 2.2120931148529053, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7223014831542969, + "num_tokens": 291107503.0, + "step": 11358 + }, + { + "epoch": 1.247419283988579, + "grad_norm": 1.975481390953064, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7188500761985779, + "num_tokens": 291137888.0, + "step": 11359 + }, + { + "epoch": 1.2475291016911927, + "grad_norm": 2.0973100662231445, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7028049230575562, + "num_tokens": 291167738.0, + "step": 11360 + }, + { + "epoch": 1.2476389193938062, + "grad_norm": 2.3945465087890625, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7372533082962036, + "num_tokens": 291189701.0, + "step": 11361 + }, + { + "epoch": 1.24774873709642, + "grad_norm": 2.069284200668335, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.711094081401825, + "num_tokens": 291217916.0, + "step": 11362 + }, + { + "epoch": 1.2478585547990335, + "grad_norm": 2.4505887031555176, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7332063317298889, + "num_tokens": 291239281.0, + "step": 11363 + }, + { + "epoch": 1.2479683725016473, + "grad_norm": 2.335287570953369, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7280968427658081, + "num_tokens": 291263232.0, + "step": 11364 + }, + { + "epoch": 1.2480781902042608, + "grad_norm": 2.6558687686920166, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7353671789169312, + "num_tokens": 291281873.0, + "step": 11365 + }, + { + "epoch": 1.2481880079068746, + "grad_norm": 2.204322338104248, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7117989659309387, + "num_tokens": 291306880.0, + "step": 11366 + }, + { + "epoch": 1.2482978256094883, + "grad_norm": 2.1052067279815674, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7087274193763733, + "num_tokens": 291337058.0, + "step": 11367 + }, + { + "epoch": 1.2484076433121019, + "grad_norm": 2.110647439956665, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7320330739021301, + "num_tokens": 291364155.0, + "step": 11368 + }, + { + "epoch": 1.2485174610147156, + "grad_norm": 2.491455316543579, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7297975420951843, + "num_tokens": 291385270.0, + "step": 11369 + }, + { + "epoch": 1.2486272787173291, + "grad_norm": 2.3506546020507812, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7070410847663879, + "num_tokens": 291409422.0, + "step": 11370 + }, + { + "epoch": 1.248737096419943, + "grad_norm": 2.095583438873291, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7086962461471558, + "num_tokens": 291439394.0, + "step": 11371 + }, + { + "epoch": 1.2488469141225567, + "grad_norm": 2.1469924449920654, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7277176976203918, + "num_tokens": 291467671.0, + "step": 11372 + }, + { + "epoch": 1.2489567318251702, + "grad_norm": 2.2016375064849854, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7013684511184692, + "num_tokens": 291493863.0, + "step": 11373 + }, + { + "epoch": 1.249066549527784, + "grad_norm": 2.3396096229553223, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7476997375488281, + "num_tokens": 291517042.0, + "step": 11374 + }, + { + "epoch": 1.2491763672303975, + "grad_norm": 2.401420831680298, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7217795252799988, + "num_tokens": 291539251.0, + "step": 11375 + }, + { + "epoch": 1.2492861849330112, + "grad_norm": 2.2003302574157715, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6894601583480835, + "num_tokens": 291566718.0, + "step": 11376 + }, + { + "epoch": 1.2493960026356248, + "grad_norm": 2.4373841285705566, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7468145489692688, + "num_tokens": 291587774.0, + "step": 11377 + }, + { + "epoch": 1.2495058203382385, + "grad_norm": 2.524946689605713, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7193921804428101, + "num_tokens": 291608490.0, + "step": 11378 + }, + { + "epoch": 1.249615638040852, + "grad_norm": 2.1083500385284424, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.712928295135498, + "num_tokens": 291635726.0, + "step": 11379 + }, + { + "epoch": 1.2497254557434658, + "grad_norm": 2.214362382888794, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.730223536491394, + "num_tokens": 291661134.0, + "step": 11380 + }, + { + "epoch": 1.2498352734460796, + "grad_norm": 2.23222279548645, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7093549370765686, + "num_tokens": 291686447.0, + "step": 11381 + }, + { + "epoch": 1.2499450911486931, + "grad_norm": 2.088674545288086, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7096591591835022, + "num_tokens": 291713082.0, + "step": 11382 + }, + { + "epoch": 1.2500549088513069, + "grad_norm": 2.1707475185394287, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7043039202690125, + "num_tokens": 291741230.0, + "step": 11383 + }, + { + "epoch": 1.2501647265539204, + "grad_norm": 2.116272449493408, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7165394425392151, + "num_tokens": 291766314.0, + "step": 11384 + }, + { + "epoch": 1.2502745442565342, + "grad_norm": 2.2887275218963623, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.718288242816925, + "num_tokens": 291790557.0, + "step": 11385 + }, + { + "epoch": 1.250384361959148, + "grad_norm": 2.6519885063171387, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7185100317001343, + "num_tokens": 291811320.0, + "step": 11386 + }, + { + "epoch": 1.2504941796617615, + "grad_norm": 2.3889541625976562, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7177404165267944, + "num_tokens": 291833177.0, + "step": 11387 + }, + { + "epoch": 1.2506039973643752, + "grad_norm": 2.425734043121338, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7206344604492188, + "num_tokens": 291855153.0, + "step": 11388 + }, + { + "epoch": 1.2507138150669888, + "grad_norm": 2.37087082862854, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7502752542495728, + "num_tokens": 291879508.0, + "step": 11389 + }, + { + "epoch": 1.2508236327696025, + "grad_norm": 1.9950580596923828, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6846712827682495, + "num_tokens": 291911792.0, + "step": 11390 + }, + { + "epoch": 1.250933450472216, + "grad_norm": 2.1502416133880615, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7079389095306396, + "num_tokens": 291942307.0, + "step": 11391 + }, + { + "epoch": 1.2510432681748298, + "grad_norm": 2.267836809158325, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7286021709442139, + "num_tokens": 291965588.0, + "step": 11392 + }, + { + "epoch": 1.2511530858774433, + "grad_norm": 2.5665152072906494, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7429171800613403, + "num_tokens": 291984016.0, + "step": 11393 + }, + { + "epoch": 1.251262903580057, + "grad_norm": 2.2104735374450684, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7359796762466431, + "num_tokens": 292009315.0, + "step": 11394 + }, + { + "epoch": 1.2513727212826709, + "grad_norm": 2.5141429901123047, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7326078414916992, + "num_tokens": 292030209.0, + "step": 11395 + }, + { + "epoch": 1.2514825389852844, + "grad_norm": 2.523188829421997, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7150329947471619, + "num_tokens": 292054040.0, + "step": 11396 + }, + { + "epoch": 1.2515923566878981, + "grad_norm": 2.48724627494812, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7441396117210388, + "num_tokens": 292076055.0, + "step": 11397 + }, + { + "epoch": 1.2517021743905117, + "grad_norm": 2.0654232501983643, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7113080620765686, + "num_tokens": 292106292.0, + "step": 11398 + }, + { + "epoch": 1.2518119920931254, + "grad_norm": 2.2616608142852783, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7289430499076843, + "num_tokens": 292133016.0, + "step": 11399 + }, + { + "epoch": 1.2519218097957392, + "grad_norm": 2.4219555854797363, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.731071412563324, + "num_tokens": 292156383.0, + "step": 11400 + }, + { + "epoch": 1.2520316274983527, + "grad_norm": 2.1780335903167725, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7090122103691101, + "num_tokens": 292183649.0, + "step": 11401 + }, + { + "epoch": 1.2521414452009663, + "grad_norm": 2.0549323558807373, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7184218168258667, + "num_tokens": 292213073.0, + "step": 11402 + }, + { + "epoch": 1.25225126290358, + "grad_norm": 2.357133150100708, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7066696882247925, + "num_tokens": 292237880.0, + "step": 11403 + }, + { + "epoch": 1.2523610806061938, + "grad_norm": 2.2688162326812744, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7266491651535034, + "num_tokens": 292261850.0, + "step": 11404 + }, + { + "epoch": 1.2524708983088073, + "grad_norm": 2.1978917121887207, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7205857038497925, + "num_tokens": 292288399.0, + "step": 11405 + }, + { + "epoch": 1.252580716011421, + "grad_norm": 2.3277809619903564, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7341321706771851, + "num_tokens": 292312720.0, + "step": 11406 + }, + { + "epoch": 1.2526905337140346, + "grad_norm": 2.529247760772705, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7312011122703552, + "num_tokens": 292337006.0, + "step": 11407 + }, + { + "epoch": 1.2528003514166484, + "grad_norm": 2.162468910217285, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.737540602684021, + "num_tokens": 292364693.0, + "step": 11408 + }, + { + "epoch": 1.2529101691192621, + "grad_norm": 2.4766933917999268, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7266005873680115, + "num_tokens": 292385488.0, + "step": 11409 + }, + { + "epoch": 1.2530199868218757, + "grad_norm": 2.340944766998291, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7248830795288086, + "num_tokens": 292409637.0, + "step": 11410 + }, + { + "epoch": 1.2531298045244894, + "grad_norm": 2.154646635055542, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7220711708068848, + "num_tokens": 292435737.0, + "step": 11411 + }, + { + "epoch": 1.253239622227103, + "grad_norm": 2.507803201675415, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.6997020244598389, + "num_tokens": 292457565.0, + "step": 11412 + }, + { + "epoch": 1.2533494399297167, + "grad_norm": 2.2204535007476807, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7333242297172546, + "num_tokens": 292483728.0, + "step": 11413 + }, + { + "epoch": 1.2534592576323305, + "grad_norm": 2.239346981048584, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7183346748352051, + "num_tokens": 292510088.0, + "step": 11414 + }, + { + "epoch": 1.253569075334944, + "grad_norm": 2.1376283168792725, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7198631763458252, + "num_tokens": 292538581.0, + "step": 11415 + }, + { + "epoch": 1.2536788930375575, + "grad_norm": 2.1229805946350098, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7081143856048584, + "num_tokens": 292569189.0, + "step": 11416 + }, + { + "epoch": 1.2537887107401713, + "grad_norm": 2.062542200088501, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7162446975708008, + "num_tokens": 292597785.0, + "step": 11417 + }, + { + "epoch": 1.253898528442785, + "grad_norm": 2.174853801727295, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7285436987876892, + "num_tokens": 292623814.0, + "step": 11418 + }, + { + "epoch": 1.2540083461453986, + "grad_norm": 2.2895350456237793, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7219107151031494, + "num_tokens": 292648918.0, + "step": 11419 + }, + { + "epoch": 1.2541181638480123, + "grad_norm": 2.10365629196167, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7177805304527283, + "num_tokens": 292676498.0, + "step": 11420 + }, + { + "epoch": 1.2542279815506259, + "grad_norm": 2.114123821258545, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7194868326187134, + "num_tokens": 292705485.0, + "step": 11421 + }, + { + "epoch": 1.2543377992532396, + "grad_norm": 2.1238174438476562, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7133160829544067, + "num_tokens": 292735403.0, + "step": 11422 + }, + { + "epoch": 1.2544476169558534, + "grad_norm": 2.158865451812744, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7219213247299194, + "num_tokens": 292762055.0, + "step": 11423 + }, + { + "epoch": 1.254557434658467, + "grad_norm": 2.079341411590576, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7239229679107666, + "num_tokens": 292789483.0, + "step": 11424 + }, + { + "epoch": 1.2546672523610807, + "grad_norm": 2.391934871673584, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.732698380947113, + "num_tokens": 292811593.0, + "step": 11425 + }, + { + "epoch": 1.2547770700636942, + "grad_norm": 2.4707448482513428, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7002909183502197, + "num_tokens": 292834676.0, + "step": 11426 + }, + { + "epoch": 1.254886887766308, + "grad_norm": 2.1193175315856934, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7138004302978516, + "num_tokens": 292862911.0, + "step": 11427 + }, + { + "epoch": 1.2549967054689217, + "grad_norm": 2.5648231506347656, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7283384799957275, + "num_tokens": 292882305.0, + "step": 11428 + }, + { + "epoch": 1.2551065231715353, + "grad_norm": 2.2860279083251953, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7241407632827759, + "num_tokens": 292908219.0, + "step": 11429 + }, + { + "epoch": 1.2552163408741488, + "grad_norm": 2.078622579574585, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6951004266738892, + "num_tokens": 292939643.0, + "step": 11430 + }, + { + "epoch": 1.2553261585767626, + "grad_norm": 2.647897720336914, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7256683111190796, + "num_tokens": 292958366.0, + "step": 11431 + }, + { + "epoch": 1.2554359762793763, + "grad_norm": 2.4333226680755615, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7343023419380188, + "num_tokens": 292981166.0, + "step": 11432 + }, + { + "epoch": 1.2555457939819898, + "grad_norm": 2.1778464317321777, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.723408579826355, + "num_tokens": 293008260.0, + "step": 11433 + }, + { + "epoch": 1.2556556116846036, + "grad_norm": 2.1760101318359375, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7173614501953125, + "num_tokens": 293034074.0, + "step": 11434 + }, + { + "epoch": 1.2557654293872171, + "grad_norm": 2.2501559257507324, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7269201278686523, + "num_tokens": 293058726.0, + "step": 11435 + }, + { + "epoch": 1.255875247089831, + "grad_norm": 1.984302282333374, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6935210227966309, + "num_tokens": 293093363.0, + "step": 11436 + }, + { + "epoch": 1.2559850647924446, + "grad_norm": 2.2389302253723145, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7284829616546631, + "num_tokens": 293118466.0, + "step": 11437 + }, + { + "epoch": 1.2560948824950582, + "grad_norm": 2.1122453212738037, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7087268829345703, + "num_tokens": 293147504.0, + "step": 11438 + }, + { + "epoch": 1.256204700197672, + "grad_norm": 2.151082992553711, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7065977454185486, + "num_tokens": 293177266.0, + "step": 11439 + }, + { + "epoch": 1.2563145179002855, + "grad_norm": 2.5754356384277344, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7070536613464355, + "num_tokens": 293198170.0, + "step": 11440 + }, + { + "epoch": 1.2564243356028992, + "grad_norm": 2.2162160873413086, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7167080640792847, + "num_tokens": 293222522.0, + "step": 11441 + }, + { + "epoch": 1.2565341533055128, + "grad_norm": 2.591867685317993, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7262065410614014, + "num_tokens": 293242998.0, + "step": 11442 + }, + { + "epoch": 1.2566439710081265, + "grad_norm": 2.0506417751312256, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7144070267677307, + "num_tokens": 293272096.0, + "step": 11443 + }, + { + "epoch": 1.25675378871074, + "grad_norm": 2.567272663116455, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7488353252410889, + "num_tokens": 293290704.0, + "step": 11444 + }, + { + "epoch": 1.2568636064133538, + "grad_norm": 2.2485928535461426, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7325606346130371, + "num_tokens": 293314917.0, + "step": 11445 + }, + { + "epoch": 1.2569734241159676, + "grad_norm": 2.041552782058716, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7351765036582947, + "num_tokens": 293343742.0, + "step": 11446 + }, + { + "epoch": 1.257083241818581, + "grad_norm": 2.109133005142212, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7216674089431763, + "num_tokens": 293373101.0, + "step": 11447 + }, + { + "epoch": 1.2571930595211949, + "grad_norm": 2.142909288406372, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.702523946762085, + "num_tokens": 293401186.0, + "step": 11448 + }, + { + "epoch": 1.2573028772238084, + "grad_norm": 2.08561110496521, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7238116264343262, + "num_tokens": 293428995.0, + "step": 11449 + }, + { + "epoch": 1.2574126949264222, + "grad_norm": 2.404757261276245, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7217487096786499, + "num_tokens": 293450273.0, + "step": 11450 + }, + { + "epoch": 1.257522512629036, + "grad_norm": 1.9546931982040405, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7140631675720215, + "num_tokens": 293481856.0, + "step": 11451 + }, + { + "epoch": 1.2576323303316495, + "grad_norm": 2.4384961128234863, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7328577637672424, + "num_tokens": 293503652.0, + "step": 11452 + }, + { + "epoch": 1.257742148034263, + "grad_norm": 2.340911388397217, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7103034257888794, + "num_tokens": 293528479.0, + "step": 11453 + }, + { + "epoch": 1.2578519657368767, + "grad_norm": 2.8511569499969482, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7410882711410522, + "num_tokens": 293546213.0, + "step": 11454 + }, + { + "epoch": 1.2579617834394905, + "grad_norm": 1.985644817352295, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7215551733970642, + "num_tokens": 293576338.0, + "step": 11455 + }, + { + "epoch": 1.258071601142104, + "grad_norm": 2.2220966815948486, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7280735969543457, + "num_tokens": 293603008.0, + "step": 11456 + }, + { + "epoch": 1.2581814188447178, + "grad_norm": 2.3485360145568848, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7211863398551941, + "num_tokens": 293628257.0, + "step": 11457 + }, + { + "epoch": 1.2582912365473313, + "grad_norm": 2.1043124198913574, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.729237973690033, + "num_tokens": 293657028.0, + "step": 11458 + }, + { + "epoch": 1.258401054249945, + "grad_norm": 1.9403035640716553, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7353593111038208, + "num_tokens": 293689014.0, + "step": 11459 + }, + { + "epoch": 1.2585108719525588, + "grad_norm": 1.9988995790481567, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7083603739738464, + "num_tokens": 293721722.0, + "step": 11460 + }, + { + "epoch": 1.2586206896551724, + "grad_norm": 2.0722498893737793, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7355254292488098, + "num_tokens": 293750335.0, + "step": 11461 + }, + { + "epoch": 1.2587305073577861, + "grad_norm": 2.502626657485962, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7073326110839844, + "num_tokens": 293773420.0, + "step": 11462 + }, + { + "epoch": 1.2588403250603997, + "grad_norm": 2.308044910430908, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7203642129898071, + "num_tokens": 293799121.0, + "step": 11463 + }, + { + "epoch": 1.2589501427630134, + "grad_norm": 1.972996473312378, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6927014589309692, + "num_tokens": 293832592.0, + "step": 11464 + }, + { + "epoch": 1.2590599604656272, + "grad_norm": 2.530853271484375, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7384138703346252, + "num_tokens": 293852344.0, + "step": 11465 + }, + { + "epoch": 1.2591697781682407, + "grad_norm": 2.2840826511383057, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7341300249099731, + "num_tokens": 293876504.0, + "step": 11466 + }, + { + "epoch": 1.2592795958708543, + "grad_norm": 1.9966380596160889, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6978263854980469, + "num_tokens": 293906491.0, + "step": 11467 + }, + { + "epoch": 1.259389413573468, + "grad_norm": 2.2737314701080322, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7417792081832886, + "num_tokens": 293927926.0, + "step": 11468 + }, + { + "epoch": 1.2594992312760818, + "grad_norm": 2.1146461963653564, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7180012464523315, + "num_tokens": 293953018.0, + "step": 11469 + }, + { + "epoch": 1.2596090489786953, + "grad_norm": 2.3021278381347656, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6944898366928101, + "num_tokens": 293977648.0, + "step": 11470 + }, + { + "epoch": 1.259718866681309, + "grad_norm": 2.1878528594970703, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7225767374038696, + "num_tokens": 294005400.0, + "step": 11471 + }, + { + "epoch": 1.2598286843839226, + "grad_norm": 2.0771727561950684, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7296015024185181, + "num_tokens": 294034128.0, + "step": 11472 + }, + { + "epoch": 1.2599385020865363, + "grad_norm": 2.098903179168701, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7091460824012756, + "num_tokens": 294060657.0, + "step": 11473 + }, + { + "epoch": 1.26004831978915, + "grad_norm": 2.005643844604492, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7031999230384827, + "num_tokens": 294090187.0, + "step": 11474 + }, + { + "epoch": 1.2601581374917636, + "grad_norm": 2.2013988494873047, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7283097505569458, + "num_tokens": 294116394.0, + "step": 11475 + }, + { + "epoch": 1.2602679551943774, + "grad_norm": 2.1508450508117676, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6969414949417114, + "num_tokens": 294146934.0, + "step": 11476 + }, + { + "epoch": 1.260377772896991, + "grad_norm": 2.430079221725464, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7424169778823853, + "num_tokens": 294169073.0, + "step": 11477 + }, + { + "epoch": 1.2604875905996047, + "grad_norm": 2.097015857696533, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7256131172180176, + "num_tokens": 294197774.0, + "step": 11478 + }, + { + "epoch": 1.2605974083022184, + "grad_norm": 2.4789278507232666, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7124192714691162, + "num_tokens": 294220667.0, + "step": 11479 + }, + { + "epoch": 1.260707226004832, + "grad_norm": 2.331486701965332, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7237384915351868, + "num_tokens": 294244146.0, + "step": 11480 + }, + { + "epoch": 1.2608170437074455, + "grad_norm": 2.6500141620635986, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7153251767158508, + "num_tokens": 294265229.0, + "step": 11481 + }, + { + "epoch": 1.2609268614100593, + "grad_norm": 2.1584832668304443, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7167360782623291, + "num_tokens": 294292771.0, + "step": 11482 + }, + { + "epoch": 1.261036679112673, + "grad_norm": 2.249846935272217, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7008084654808044, + "num_tokens": 294319581.0, + "step": 11483 + }, + { + "epoch": 1.2611464968152866, + "grad_norm": 2.1274383068084717, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7154914140701294, + "num_tokens": 294347675.0, + "step": 11484 + }, + { + "epoch": 1.2612563145179003, + "grad_norm": 2.229631185531616, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7062649726867676, + "num_tokens": 294374074.0, + "step": 11485 + }, + { + "epoch": 1.2613661322205139, + "grad_norm": 2.4213767051696777, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7295495271682739, + "num_tokens": 294397064.0, + "step": 11486 + }, + { + "epoch": 1.2614759499231276, + "grad_norm": 2.272235155105591, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7430864572525024, + "num_tokens": 294421535.0, + "step": 11487 + }, + { + "epoch": 1.2615857676257414, + "grad_norm": 1.972198247909546, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7184466123580933, + "num_tokens": 294453511.0, + "step": 11488 + }, + { + "epoch": 1.261695585328355, + "grad_norm": 2.339298963546753, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7479605674743652, + "num_tokens": 294473376.0, + "step": 11489 + }, + { + "epoch": 1.2618054030309687, + "grad_norm": 2.3580563068389893, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7158287763595581, + "num_tokens": 294496679.0, + "step": 11490 + }, + { + "epoch": 1.2619152207335822, + "grad_norm": 2.532357692718506, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7246235013008118, + "num_tokens": 294516625.0, + "step": 11491 + }, + { + "epoch": 1.262025038436196, + "grad_norm": 2.3781118392944336, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7627700567245483, + "num_tokens": 294537385.0, + "step": 11492 + }, + { + "epoch": 1.2621348561388097, + "grad_norm": 2.2239058017730713, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7112288475036621, + "num_tokens": 294564108.0, + "step": 11493 + }, + { + "epoch": 1.2622446738414232, + "grad_norm": 2.2024641036987305, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7033507823944092, + "num_tokens": 294592814.0, + "step": 11494 + }, + { + "epoch": 1.2623544915440368, + "grad_norm": 2.291304111480713, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7106805443763733, + "num_tokens": 294619962.0, + "step": 11495 + }, + { + "epoch": 1.2624643092466505, + "grad_norm": 2.360297918319702, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7084293365478516, + "num_tokens": 294645210.0, + "step": 11496 + }, + { + "epoch": 1.2625741269492643, + "grad_norm": 2.224799394607544, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6889728307723999, + "num_tokens": 294670784.0, + "step": 11497 + }, + { + "epoch": 1.2626839446518778, + "grad_norm": 2.5467302799224854, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7275737524032593, + "num_tokens": 294691498.0, + "step": 11498 + }, + { + "epoch": 1.2627937623544916, + "grad_norm": 2.354567527770996, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7191380858421326, + "num_tokens": 294714234.0, + "step": 11499 + }, + { + "epoch": 1.2629035800571051, + "grad_norm": 2.3685925006866455, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7067873477935791, + "num_tokens": 294737601.0, + "step": 11500 + }, + { + "epoch": 1.2630133977597189, + "grad_norm": 2.3527512550354004, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7266253232955933, + "num_tokens": 294762288.0, + "step": 11501 + }, + { + "epoch": 1.2631232154623326, + "grad_norm": 2.3229587078094482, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7294724583625793, + "num_tokens": 294787703.0, + "step": 11502 + }, + { + "epoch": 1.2632330331649462, + "grad_norm": 2.513197660446167, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7204638719558716, + "num_tokens": 294809958.0, + "step": 11503 + }, + { + "epoch": 1.26334285086756, + "grad_norm": 2.0264766216278076, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7382886409759521, + "num_tokens": 294839327.0, + "step": 11504 + }, + { + "epoch": 1.2634526685701735, + "grad_norm": 2.2545249462127686, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7166784405708313, + "num_tokens": 294864665.0, + "step": 11505 + }, + { + "epoch": 1.2635624862727872, + "grad_norm": 2.4479973316192627, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7568090558052063, + "num_tokens": 294886860.0, + "step": 11506 + }, + { + "epoch": 1.2636723039754008, + "grad_norm": 2.4093093872070312, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7325229644775391, + "num_tokens": 294908206.0, + "step": 11507 + }, + { + "epoch": 1.2637821216780145, + "grad_norm": 2.163006067276001, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7111679911613464, + "num_tokens": 294935512.0, + "step": 11508 + }, + { + "epoch": 1.263891939380628, + "grad_norm": 2.3889641761779785, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7137029767036438, + "num_tokens": 294957925.0, + "step": 11509 + }, + { + "epoch": 1.2640017570832418, + "grad_norm": 2.1956892013549805, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7052860260009766, + "num_tokens": 294986069.0, + "step": 11510 + }, + { + "epoch": 1.2641115747858556, + "grad_norm": 2.7814583778381348, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7418946623802185, + "num_tokens": 295004921.0, + "step": 11511 + }, + { + "epoch": 1.264221392488469, + "grad_norm": 2.698690176010132, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.738609790802002, + "num_tokens": 295021911.0, + "step": 11512 + }, + { + "epoch": 1.2643312101910829, + "grad_norm": 1.930043339729309, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6913569569587708, + "num_tokens": 295056834.0, + "step": 11513 + }, + { + "epoch": 1.2644410278936964, + "grad_norm": 2.3583426475524902, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7226248979568481, + "num_tokens": 295081540.0, + "step": 11514 + }, + { + "epoch": 1.2645508455963101, + "grad_norm": 2.900010108947754, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7273187041282654, + "num_tokens": 295098922.0, + "step": 11515 + }, + { + "epoch": 1.264660663298924, + "grad_norm": 2.414263963699341, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7086902260780334, + "num_tokens": 295123212.0, + "step": 11516 + }, + { + "epoch": 1.2647704810015374, + "grad_norm": 2.0760409832000732, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7332963347434998, + "num_tokens": 295150207.0, + "step": 11517 + }, + { + "epoch": 1.264880298704151, + "grad_norm": 2.390033483505249, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7013120651245117, + "num_tokens": 295173511.0, + "step": 11518 + }, + { + "epoch": 1.2649901164067647, + "grad_norm": 2.3156824111938477, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7209678292274475, + "num_tokens": 295198317.0, + "step": 11519 + }, + { + "epoch": 1.2650999341093785, + "grad_norm": 2.324341297149658, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6959928274154663, + "num_tokens": 295224040.0, + "step": 11520 + }, + { + "epoch": 1.265209751811992, + "grad_norm": 2.432908058166504, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7331539392471313, + "num_tokens": 295246210.0, + "step": 11521 + }, + { + "epoch": 1.2653195695146058, + "grad_norm": 2.192845344543457, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7092691659927368, + "num_tokens": 295271891.0, + "step": 11522 + }, + { + "epoch": 1.2654293872172193, + "grad_norm": 1.9780985116958618, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7126212120056152, + "num_tokens": 295304399.0, + "step": 11523 + }, + { + "epoch": 1.265539204919833, + "grad_norm": 2.4072444438934326, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7372525930404663, + "num_tokens": 295324346.0, + "step": 11524 + }, + { + "epoch": 1.2656490226224468, + "grad_norm": 2.235881805419922, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7105696201324463, + "num_tokens": 295351498.0, + "step": 11525 + }, + { + "epoch": 1.2657588403250604, + "grad_norm": 2.2735610008239746, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7371329069137573, + "num_tokens": 295374633.0, + "step": 11526 + }, + { + "epoch": 1.2658686580276741, + "grad_norm": 2.1934547424316406, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6925760507583618, + "num_tokens": 295401093.0, + "step": 11527 + }, + { + "epoch": 1.2659784757302877, + "grad_norm": 2.465707778930664, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7396000027656555, + "num_tokens": 295422230.0, + "step": 11528 + }, + { + "epoch": 1.2660882934329014, + "grad_norm": 2.5827884674072266, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7517714500427246, + "num_tokens": 295440601.0, + "step": 11529 + }, + { + "epoch": 1.2661981111355152, + "grad_norm": 2.391573667526245, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7363845109939575, + "num_tokens": 295463499.0, + "step": 11530 + }, + { + "epoch": 1.2663079288381287, + "grad_norm": 2.115396499633789, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7137497067451477, + "num_tokens": 295491722.0, + "step": 11531 + }, + { + "epoch": 1.2664177465407422, + "grad_norm": 2.0124120712280273, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7117030620574951, + "num_tokens": 295521960.0, + "step": 11532 + }, + { + "epoch": 1.266527564243356, + "grad_norm": 2.2590930461883545, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7257227897644043, + "num_tokens": 295548083.0, + "step": 11533 + }, + { + "epoch": 1.2666373819459698, + "grad_norm": 2.22178053855896, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7331787347793579, + "num_tokens": 295573382.0, + "step": 11534 + }, + { + "epoch": 1.2667471996485833, + "grad_norm": 2.135533094406128, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7113406658172607, + "num_tokens": 295601060.0, + "step": 11535 + }, + { + "epoch": 1.266857017351197, + "grad_norm": 2.117755174636841, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7381216883659363, + "num_tokens": 295632047.0, + "step": 11536 + }, + { + "epoch": 1.2669668350538106, + "grad_norm": 2.09662127494812, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7014442086219788, + "num_tokens": 295662089.0, + "step": 11537 + }, + { + "epoch": 1.2670766527564243, + "grad_norm": 1.9960216283798218, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7172994017601013, + "num_tokens": 295694037.0, + "step": 11538 + }, + { + "epoch": 1.267186470459038, + "grad_norm": 2.241942882537842, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7122846245765686, + "num_tokens": 295718580.0, + "step": 11539 + }, + { + "epoch": 1.2672962881616516, + "grad_norm": 2.3002994060516357, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7300783395767212, + "num_tokens": 295741652.0, + "step": 11540 + }, + { + "epoch": 1.2674061058642654, + "grad_norm": 2.3215649127960205, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7238892316818237, + "num_tokens": 295767272.0, + "step": 11541 + }, + { + "epoch": 1.267515923566879, + "grad_norm": 2.5037424564361572, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.737348198890686, + "num_tokens": 295788828.0, + "step": 11542 + }, + { + "epoch": 1.2676257412694927, + "grad_norm": 2.1907076835632324, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7269406318664551, + "num_tokens": 295815193.0, + "step": 11543 + }, + { + "epoch": 1.2677355589721064, + "grad_norm": 2.246907949447632, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7237060070037842, + "num_tokens": 295840656.0, + "step": 11544 + }, + { + "epoch": 1.26784537667472, + "grad_norm": 2.1322829723358154, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7218793034553528, + "num_tokens": 295866697.0, + "step": 11545 + }, + { + "epoch": 1.2679551943773335, + "grad_norm": 2.0716493129730225, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7110865116119385, + "num_tokens": 295896171.0, + "step": 11546 + }, + { + "epoch": 1.2680650120799473, + "grad_norm": 2.125300884246826, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7123242616653442, + "num_tokens": 295924405.0, + "step": 11547 + }, + { + "epoch": 1.268174829782561, + "grad_norm": 2.0578763484954834, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7058885097503662, + "num_tokens": 295958096.0, + "step": 11548 + }, + { + "epoch": 1.2682846474851746, + "grad_norm": 2.226264476776123, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7201141119003296, + "num_tokens": 295983120.0, + "step": 11549 + }, + { + "epoch": 1.2683944651877883, + "grad_norm": 2.3201754093170166, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7057211399078369, + "num_tokens": 296009627.0, + "step": 11550 + }, + { + "epoch": 1.2685042828904018, + "grad_norm": 2.170398712158203, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7303985953330994, + "num_tokens": 296035337.0, + "step": 11551 + }, + { + "epoch": 1.2686141005930156, + "grad_norm": 2.364983081817627, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.757886528968811, + "num_tokens": 296056663.0, + "step": 11552 + }, + { + "epoch": 1.2687239182956294, + "grad_norm": 2.3540351390838623, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7388188242912292, + "num_tokens": 296078686.0, + "step": 11553 + }, + { + "epoch": 1.268833735998243, + "grad_norm": 2.222700357437134, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.728346586227417, + "num_tokens": 296104798.0, + "step": 11554 + }, + { + "epoch": 1.2689435537008567, + "grad_norm": 2.515483856201172, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7116466164588928, + "num_tokens": 296127144.0, + "step": 11555 + }, + { + "epoch": 1.2690533714034702, + "grad_norm": 2.0930118560791016, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7175934314727783, + "num_tokens": 296154318.0, + "step": 11556 + }, + { + "epoch": 1.269163189106084, + "grad_norm": 2.26236629486084, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7091134786605835, + "num_tokens": 296180275.0, + "step": 11557 + }, + { + "epoch": 1.2692730068086975, + "grad_norm": 2.465277910232544, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7279572486877441, + "num_tokens": 296201255.0, + "step": 11558 + }, + { + "epoch": 1.2693828245113112, + "grad_norm": 2.3499722480773926, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.716373085975647, + "num_tokens": 296227803.0, + "step": 11559 + }, + { + "epoch": 1.2694926422139248, + "grad_norm": 2.197705030441284, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7394963502883911, + "num_tokens": 296253762.0, + "step": 11560 + }, + { + "epoch": 1.2696024599165385, + "grad_norm": 2.456516981124878, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7067159414291382, + "num_tokens": 296275573.0, + "step": 11561 + }, + { + "epoch": 1.2697122776191523, + "grad_norm": 2.253816604614258, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7152782082557678, + "num_tokens": 296300699.0, + "step": 11562 + }, + { + "epoch": 1.2698220953217658, + "grad_norm": 2.1787476539611816, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7295316457748413, + "num_tokens": 296328643.0, + "step": 11563 + }, + { + "epoch": 1.2699319130243796, + "grad_norm": 2.161085367202759, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7373324632644653, + "num_tokens": 296354776.0, + "step": 11564 + }, + { + "epoch": 1.2700417307269931, + "grad_norm": 2.3264715671539307, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7220733165740967, + "num_tokens": 296377165.0, + "step": 11565 + }, + { + "epoch": 1.2701515484296069, + "grad_norm": 2.1076102256774902, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7131943702697754, + "num_tokens": 296406824.0, + "step": 11566 + }, + { + "epoch": 1.2702613661322206, + "grad_norm": 2.4881885051727295, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7235050201416016, + "num_tokens": 296430001.0, + "step": 11567 + }, + { + "epoch": 1.2703711838348342, + "grad_norm": 2.522108316421509, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7376991510391235, + "num_tokens": 296449795.0, + "step": 11568 + }, + { + "epoch": 1.270481001537448, + "grad_norm": 2.366786003112793, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7172930240631104, + "num_tokens": 296472140.0, + "step": 11569 + }, + { + "epoch": 1.2705908192400615, + "grad_norm": 2.109621286392212, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7068934440612793, + "num_tokens": 296500039.0, + "step": 11570 + }, + { + "epoch": 1.2707006369426752, + "grad_norm": 2.160858154296875, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7323768734931946, + "num_tokens": 296525925.0, + "step": 11571 + }, + { + "epoch": 1.2708104546452887, + "grad_norm": 2.574047803878784, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7252973318099976, + "num_tokens": 296546309.0, + "step": 11572 + }, + { + "epoch": 1.2709202723479025, + "grad_norm": 2.26430344581604, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7364437580108643, + "num_tokens": 296571479.0, + "step": 11573 + }, + { + "epoch": 1.271030090050516, + "grad_norm": 2.328441619873047, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7227706909179688, + "num_tokens": 296594614.0, + "step": 11574 + }, + { + "epoch": 1.2711399077531298, + "grad_norm": 2.437446355819702, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7132449150085449, + "num_tokens": 296616946.0, + "step": 11575 + }, + { + "epoch": 1.2712497254557436, + "grad_norm": 1.971563696861267, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7005237340927124, + "num_tokens": 296647229.0, + "step": 11576 + }, + { + "epoch": 1.271359543158357, + "grad_norm": 2.0444557666778564, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7151497602462769, + "num_tokens": 296677553.0, + "step": 11577 + }, + { + "epoch": 1.2714693608609708, + "grad_norm": 2.114095449447632, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.742376446723938, + "num_tokens": 296704386.0, + "step": 11578 + }, + { + "epoch": 1.2715791785635844, + "grad_norm": 2.4064781665802, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7200314998626709, + "num_tokens": 296727770.0, + "step": 11579 + }, + { + "epoch": 1.2716889962661981, + "grad_norm": 2.2788422107696533, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7382361888885498, + "num_tokens": 296752582.0, + "step": 11580 + }, + { + "epoch": 1.271798813968812, + "grad_norm": 2.202683687210083, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7405973672866821, + "num_tokens": 296777322.0, + "step": 11581 + }, + { + "epoch": 1.2719086316714254, + "grad_norm": 2.0591018199920654, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7181009650230408, + "num_tokens": 296807027.0, + "step": 11582 + }, + { + "epoch": 1.272018449374039, + "grad_norm": 2.1910922527313232, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.72477126121521, + "num_tokens": 296834642.0, + "step": 11583 + }, + { + "epoch": 1.2721282670766527, + "grad_norm": 2.1939191818237305, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7360101938247681, + "num_tokens": 296861173.0, + "step": 11584 + }, + { + "epoch": 1.2722380847792665, + "grad_norm": 1.9900015592575073, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7190159559249878, + "num_tokens": 296890126.0, + "step": 11585 + }, + { + "epoch": 1.27234790248188, + "grad_norm": 1.9982320070266724, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7170966267585754, + "num_tokens": 296921154.0, + "step": 11586 + }, + { + "epoch": 1.2724577201844938, + "grad_norm": 2.1384825706481934, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7153247594833374, + "num_tokens": 296949263.0, + "step": 11587 + }, + { + "epoch": 1.2725675378871073, + "grad_norm": 2.163182020187378, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6892303228378296, + "num_tokens": 296975616.0, + "step": 11588 + }, + { + "epoch": 1.272677355589721, + "grad_norm": 2.3245186805725098, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7106646299362183, + "num_tokens": 297000785.0, + "step": 11589 + }, + { + "epoch": 1.2727871732923348, + "grad_norm": 2.1067323684692383, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7365678548812866, + "num_tokens": 297027935.0, + "step": 11590 + }, + { + "epoch": 1.2728969909949484, + "grad_norm": 2.238464117050171, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7075240612030029, + "num_tokens": 297056376.0, + "step": 11591 + }, + { + "epoch": 1.273006808697562, + "grad_norm": 2.132143497467041, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7176544666290283, + "num_tokens": 297084657.0, + "step": 11592 + }, + { + "epoch": 1.2731166264001756, + "grad_norm": 2.3984336853027344, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7162695527076721, + "num_tokens": 297108233.0, + "step": 11593 + }, + { + "epoch": 1.2732264441027894, + "grad_norm": 2.5741007328033447, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7256815433502197, + "num_tokens": 297129226.0, + "step": 11594 + }, + { + "epoch": 1.2733362618054032, + "grad_norm": 2.291760206222534, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7134803533554077, + "num_tokens": 297152666.0, + "step": 11595 + }, + { + "epoch": 1.2734460795080167, + "grad_norm": 2.412419319152832, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.712867021560669, + "num_tokens": 297176038.0, + "step": 11596 + }, + { + "epoch": 1.2735558972106302, + "grad_norm": 2.4501214027404785, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7133371829986572, + "num_tokens": 297198352.0, + "step": 11597 + }, + { + "epoch": 1.273665714913244, + "grad_norm": 2.0556492805480957, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7002949714660645, + "num_tokens": 297228975.0, + "step": 11598 + }, + { + "epoch": 1.2737755326158577, + "grad_norm": 2.899474620819092, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7238508462905884, + "num_tokens": 297244863.0, + "step": 11599 + }, + { + "epoch": 1.2738853503184713, + "grad_norm": 2.3851871490478516, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7128360271453857, + "num_tokens": 297267970.0, + "step": 11600 + }, + { + "epoch": 1.273995168021085, + "grad_norm": 2.339679479598999, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7236402630805969, + "num_tokens": 297291542.0, + "step": 11601 + }, + { + "epoch": 1.2741049857236986, + "grad_norm": 2.6048197746276855, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7371999025344849, + "num_tokens": 297310300.0, + "step": 11602 + }, + { + "epoch": 1.2742148034263123, + "grad_norm": 2.293196201324463, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7326961755752563, + "num_tokens": 297335420.0, + "step": 11603 + }, + { + "epoch": 1.274324621128926, + "grad_norm": 2.4499218463897705, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7120846509933472, + "num_tokens": 297358185.0, + "step": 11604 + }, + { + "epoch": 1.2744344388315396, + "grad_norm": 2.1474802494049072, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7139260172843933, + "num_tokens": 297386440.0, + "step": 11605 + }, + { + "epoch": 1.2745442565341534, + "grad_norm": 2.7868990898132324, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7188286781311035, + "num_tokens": 297405141.0, + "step": 11606 + }, + { + "epoch": 1.274654074236767, + "grad_norm": 2.2991127967834473, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7239962220191956, + "num_tokens": 297430219.0, + "step": 11607 + }, + { + "epoch": 1.2747638919393807, + "grad_norm": 2.228515148162842, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.706096351146698, + "num_tokens": 297456750.0, + "step": 11608 + }, + { + "epoch": 1.2748737096419944, + "grad_norm": 2.1799468994140625, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7149885892868042, + "num_tokens": 297482789.0, + "step": 11609 + }, + { + "epoch": 1.274983527344608, + "grad_norm": 2.112670421600342, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7099709510803223, + "num_tokens": 297512259.0, + "step": 11610 + }, + { + "epoch": 1.2750933450472215, + "grad_norm": 2.2391364574432373, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7293437719345093, + "num_tokens": 297537079.0, + "step": 11611 + }, + { + "epoch": 1.2752031627498353, + "grad_norm": 2.0864498615264893, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7175518870353699, + "num_tokens": 297568167.0, + "step": 11612 + }, + { + "epoch": 1.275312980452449, + "grad_norm": 2.1644246578216553, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6924727559089661, + "num_tokens": 297593706.0, + "step": 11613 + }, + { + "epoch": 1.2754227981550625, + "grad_norm": 2.449584484100342, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7044609785079956, + "num_tokens": 297617720.0, + "step": 11614 + }, + { + "epoch": 1.2755326158576763, + "grad_norm": 2.1817715167999268, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7186552286148071, + "num_tokens": 297644901.0, + "step": 11615 + }, + { + "epoch": 1.2756424335602898, + "grad_norm": 2.181602954864502, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7087574601173401, + "num_tokens": 297672823.0, + "step": 11616 + }, + { + "epoch": 1.2757522512629036, + "grad_norm": 2.2569732666015625, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7262052297592163, + "num_tokens": 297697201.0, + "step": 11617 + }, + { + "epoch": 1.2758620689655173, + "grad_norm": 2.3311712741851807, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7109954357147217, + "num_tokens": 297719864.0, + "step": 11618 + }, + { + "epoch": 1.2759718866681309, + "grad_norm": 2.246636390686035, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6947163343429565, + "num_tokens": 297746172.0, + "step": 11619 + }, + { + "epoch": 1.2760817043707446, + "grad_norm": 2.761685371398926, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7540212273597717, + "num_tokens": 297763557.0, + "step": 11620 + }, + { + "epoch": 1.2761915220733582, + "grad_norm": 2.253249406814575, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7551249265670776, + "num_tokens": 297789101.0, + "step": 11621 + }, + { + "epoch": 1.276301339775972, + "grad_norm": 2.3705031871795654, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7342863082885742, + "num_tokens": 297812298.0, + "step": 11622 + }, + { + "epoch": 1.2764111574785855, + "grad_norm": 2.2119693756103516, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7150737047195435, + "num_tokens": 297839369.0, + "step": 11623 + }, + { + "epoch": 1.2765209751811992, + "grad_norm": 2.294689655303955, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7235442996025085, + "num_tokens": 297864875.0, + "step": 11624 + }, + { + "epoch": 1.2766307928838128, + "grad_norm": 2.5686099529266357, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7315473556518555, + "num_tokens": 297885659.0, + "step": 11625 + }, + { + "epoch": 1.2767406105864265, + "grad_norm": 2.172166347503662, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7107536196708679, + "num_tokens": 297914426.0, + "step": 11626 + }, + { + "epoch": 1.2768504282890403, + "grad_norm": 2.3208205699920654, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7002018690109253, + "num_tokens": 297937296.0, + "step": 11627 + }, + { + "epoch": 1.2769602459916538, + "grad_norm": 2.144984245300293, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6897881031036377, + "num_tokens": 297964029.0, + "step": 11628 + }, + { + "epoch": 1.2770700636942676, + "grad_norm": 2.2918949127197266, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7190157771110535, + "num_tokens": 297989825.0, + "step": 11629 + }, + { + "epoch": 1.277179881396881, + "grad_norm": 2.2952497005462646, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7156010270118713, + "num_tokens": 298015018.0, + "step": 11630 + }, + { + "epoch": 1.2772896990994949, + "grad_norm": 2.184481143951416, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7278536558151245, + "num_tokens": 298040192.0, + "step": 11631 + }, + { + "epoch": 1.2773995168021086, + "grad_norm": 2.683011531829834, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7319711446762085, + "num_tokens": 298058377.0, + "step": 11632 + }, + { + "epoch": 1.2775093345047221, + "grad_norm": 2.2160420417785645, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6855241060256958, + "num_tokens": 298086286.0, + "step": 11633 + }, + { + "epoch": 1.2776191522073357, + "grad_norm": 2.853428840637207, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7427542209625244, + "num_tokens": 298102716.0, + "step": 11634 + }, + { + "epoch": 1.2777289699099494, + "grad_norm": 2.1277523040771484, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7230616211891174, + "num_tokens": 298127900.0, + "step": 11635 + }, + { + "epoch": 1.2778387876125632, + "grad_norm": 2.0624401569366455, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7252359390258789, + "num_tokens": 298156685.0, + "step": 11636 + }, + { + "epoch": 1.2779486053151767, + "grad_norm": 2.2634077072143555, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7231647968292236, + "num_tokens": 298180224.0, + "step": 11637 + }, + { + "epoch": 1.2780584230177905, + "grad_norm": 2.1626529693603516, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.700799822807312, + "num_tokens": 298208973.0, + "step": 11638 + }, + { + "epoch": 1.278168240720404, + "grad_norm": 2.1974294185638428, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7195795774459839, + "num_tokens": 298234953.0, + "step": 11639 + }, + { + "epoch": 1.2782780584230178, + "grad_norm": 2.097186326980591, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7145643830299377, + "num_tokens": 298260973.0, + "step": 11640 + }, + { + "epoch": 1.2783878761256315, + "grad_norm": 1.972607135772705, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7227950096130371, + "num_tokens": 298291590.0, + "step": 11641 + }, + { + "epoch": 1.278497693828245, + "grad_norm": 2.2244789600372314, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7076830863952637, + "num_tokens": 298319055.0, + "step": 11642 + }, + { + "epoch": 1.2786075115308588, + "grad_norm": 2.1848220825195312, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7204685211181641, + "num_tokens": 298344571.0, + "step": 11643 + }, + { + "epoch": 1.2787173292334724, + "grad_norm": 2.284508466720581, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7065397500991821, + "num_tokens": 298369878.0, + "step": 11644 + }, + { + "epoch": 1.2788271469360861, + "grad_norm": 2.096320867538452, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.733425498008728, + "num_tokens": 298396131.0, + "step": 11645 + }, + { + "epoch": 1.2789369646386999, + "grad_norm": 2.6021246910095215, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7339349389076233, + "num_tokens": 298416064.0, + "step": 11646 + }, + { + "epoch": 1.2790467823413134, + "grad_norm": 2.144474983215332, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7261120676994324, + "num_tokens": 298442384.0, + "step": 11647 + }, + { + "epoch": 1.279156600043927, + "grad_norm": 2.1693203449249268, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7223901748657227, + "num_tokens": 298469177.0, + "step": 11648 + }, + { + "epoch": 1.2792664177465407, + "grad_norm": 2.5826215744018555, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7217088937759399, + "num_tokens": 298488507.0, + "step": 11649 + }, + { + "epoch": 1.2793762354491545, + "grad_norm": 2.34997820854187, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7169713973999023, + "num_tokens": 298513087.0, + "step": 11650 + }, + { + "epoch": 1.279486053151768, + "grad_norm": 2.325695276260376, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7238783836364746, + "num_tokens": 298536533.0, + "step": 11651 + }, + { + "epoch": 1.2795958708543818, + "grad_norm": 2.3328816890716553, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7305487394332886, + "num_tokens": 298563078.0, + "step": 11652 + }, + { + "epoch": 1.2797056885569953, + "grad_norm": 2.5180559158325195, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.710455060005188, + "num_tokens": 298585429.0, + "step": 11653 + }, + { + "epoch": 1.279815506259609, + "grad_norm": 2.216189384460449, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7208354473114014, + "num_tokens": 298609270.0, + "step": 11654 + }, + { + "epoch": 1.2799253239622228, + "grad_norm": 2.216974973678589, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6962100267410278, + "num_tokens": 298636575.0, + "step": 11655 + }, + { + "epoch": 1.2800351416648363, + "grad_norm": 2.0506160259246826, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7017433643341064, + "num_tokens": 298665148.0, + "step": 11656 + }, + { + "epoch": 1.28014495936745, + "grad_norm": 2.484757423400879, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7430391311645508, + "num_tokens": 298685665.0, + "step": 11657 + }, + { + "epoch": 1.2802547770700636, + "grad_norm": 2.1364853382110596, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7086302638053894, + "num_tokens": 298714005.0, + "step": 11658 + }, + { + "epoch": 1.2803645947726774, + "grad_norm": 2.166738986968994, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7187042832374573, + "num_tokens": 298742598.0, + "step": 11659 + }, + { + "epoch": 1.2804744124752911, + "grad_norm": 2.46885085105896, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7227622270584106, + "num_tokens": 298762860.0, + "step": 11660 + }, + { + "epoch": 1.2805842301779047, + "grad_norm": 2.470245599746704, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7223304510116577, + "num_tokens": 298785814.0, + "step": 11661 + }, + { + "epoch": 1.2806940478805182, + "grad_norm": 2.4710373878479004, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7404358386993408, + "num_tokens": 298806236.0, + "step": 11662 + }, + { + "epoch": 1.280803865583132, + "grad_norm": 2.194157123565674, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7013468742370605, + "num_tokens": 298832749.0, + "step": 11663 + }, + { + "epoch": 1.2809136832857457, + "grad_norm": 2.122319459915161, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.6975753307342529, + "num_tokens": 298861329.0, + "step": 11664 + }, + { + "epoch": 1.2810235009883593, + "grad_norm": 2.3797528743743896, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6906255483627319, + "num_tokens": 298886751.0, + "step": 11665 + }, + { + "epoch": 1.281133318690973, + "grad_norm": 2.0328903198242188, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7244027256965637, + "num_tokens": 298917005.0, + "step": 11666 + }, + { + "epoch": 1.2812431363935866, + "grad_norm": 2.208340883255005, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7205774188041687, + "num_tokens": 298944307.0, + "step": 11667 + }, + { + "epoch": 1.2813529540962003, + "grad_norm": 2.0960464477539062, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7044552564620972, + "num_tokens": 298974696.0, + "step": 11668 + }, + { + "epoch": 1.281462771798814, + "grad_norm": 2.2411820888519287, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7230360507965088, + "num_tokens": 299000683.0, + "step": 11669 + }, + { + "epoch": 1.2815725895014276, + "grad_norm": 2.452343463897705, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7319794297218323, + "num_tokens": 299023466.0, + "step": 11670 + }, + { + "epoch": 1.2816824072040414, + "grad_norm": 2.28181529045105, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7255669832229614, + "num_tokens": 299049048.0, + "step": 11671 + }, + { + "epoch": 1.281792224906655, + "grad_norm": 2.1364314556121826, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.723050594329834, + "num_tokens": 299075856.0, + "step": 11672 + }, + { + "epoch": 1.2819020426092687, + "grad_norm": 2.649944305419922, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7310488224029541, + "num_tokens": 299094684.0, + "step": 11673 + }, + { + "epoch": 1.2820118603118824, + "grad_norm": 2.711951732635498, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7448874115943909, + "num_tokens": 299112888.0, + "step": 11674 + }, + { + "epoch": 1.282121678014496, + "grad_norm": 2.365278720855713, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7407190799713135, + "num_tokens": 299134963.0, + "step": 11675 + }, + { + "epoch": 1.2822314957171095, + "grad_norm": 2.0949838161468506, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7697439193725586, + "num_tokens": 299159396.0, + "step": 11676 + }, + { + "epoch": 1.2823413134197232, + "grad_norm": 2.3037307262420654, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.744853138923645, + "num_tokens": 299182833.0, + "step": 11677 + }, + { + "epoch": 1.282451131122337, + "grad_norm": 2.2115445137023926, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.70699542760849, + "num_tokens": 299213774.0, + "step": 11678 + }, + { + "epoch": 1.2825609488249505, + "grad_norm": 2.3253896236419678, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7193501591682434, + "num_tokens": 299238535.0, + "step": 11679 + }, + { + "epoch": 1.2826707665275643, + "grad_norm": 2.1278445720672607, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7179727554321289, + "num_tokens": 299266470.0, + "step": 11680 + }, + { + "epoch": 1.2827805842301778, + "grad_norm": 2.5476279258728027, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7347071170806885, + "num_tokens": 299287967.0, + "step": 11681 + }, + { + "epoch": 1.2828904019327916, + "grad_norm": 2.1476786136627197, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7202578783035278, + "num_tokens": 299316347.0, + "step": 11682 + }, + { + "epoch": 1.2830002196354053, + "grad_norm": 2.044639825820923, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7055370807647705, + "num_tokens": 299345323.0, + "step": 11683 + }, + { + "epoch": 1.2831100373380189, + "grad_norm": 2.426180601119995, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7262395620346069, + "num_tokens": 299368256.0, + "step": 11684 + }, + { + "epoch": 1.2832198550406326, + "grad_norm": 2.488811492919922, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7465195059776306, + "num_tokens": 299389098.0, + "step": 11685 + }, + { + "epoch": 1.2833296727432462, + "grad_norm": 2.398516893386841, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7147114872932434, + "num_tokens": 299414300.0, + "step": 11686 + }, + { + "epoch": 1.28343949044586, + "grad_norm": 2.3534655570983887, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.71066814661026, + "num_tokens": 299437850.0, + "step": 11687 + }, + { + "epoch": 1.2835493081484735, + "grad_norm": 2.424671173095703, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7153933644294739, + "num_tokens": 299461034.0, + "step": 11688 + }, + { + "epoch": 1.2836591258510872, + "grad_norm": 2.170600175857544, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7083941102027893, + "num_tokens": 299486919.0, + "step": 11689 + }, + { + "epoch": 1.2837689435537007, + "grad_norm": 2.3291215896606445, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7122806906700134, + "num_tokens": 299509439.0, + "step": 11690 + }, + { + "epoch": 1.2838787612563145, + "grad_norm": 2.1018624305725098, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7447175979614258, + "num_tokens": 299537210.0, + "step": 11691 + }, + { + "epoch": 1.2839885789589283, + "grad_norm": 2.4925296306610107, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7278276681900024, + "num_tokens": 299559049.0, + "step": 11692 + }, + { + "epoch": 1.2840983966615418, + "grad_norm": 1.9963129758834839, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7288668751716614, + "num_tokens": 299586507.0, + "step": 11693 + }, + { + "epoch": 1.2842082143641556, + "grad_norm": 2.3051698207855225, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7118105292320251, + "num_tokens": 299611525.0, + "step": 11694 + }, + { + "epoch": 1.284318032066769, + "grad_norm": 2.5464253425598145, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7393651008605957, + "num_tokens": 299630192.0, + "step": 11695 + }, + { + "epoch": 1.2844278497693828, + "grad_norm": 2.3316633701324463, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7286752462387085, + "num_tokens": 299655467.0, + "step": 11696 + }, + { + "epoch": 1.2845376674719966, + "grad_norm": 2.156998872756958, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7117652893066406, + "num_tokens": 299682878.0, + "step": 11697 + }, + { + "epoch": 1.2846474851746101, + "grad_norm": 2.3476877212524414, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7178727388381958, + "num_tokens": 299707943.0, + "step": 11698 + }, + { + "epoch": 1.2847573028772237, + "grad_norm": 2.2243764400482178, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7101380825042725, + "num_tokens": 299733147.0, + "step": 11699 + }, + { + "epoch": 1.2848671205798374, + "grad_norm": 2.26452898979187, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7033114433288574, + "num_tokens": 299764161.0, + "step": 11700 + }, + { + "epoch": 1.2849769382824512, + "grad_norm": 2.4133026599884033, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7266011238098145, + "num_tokens": 299788335.0, + "step": 11701 + }, + { + "epoch": 1.2850867559850647, + "grad_norm": 2.1908862590789795, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7117274403572083, + "num_tokens": 299816392.0, + "step": 11702 + }, + { + "epoch": 1.2851965736876785, + "grad_norm": 2.0621020793914795, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7291154265403748, + "num_tokens": 299846689.0, + "step": 11703 + }, + { + "epoch": 1.285306391390292, + "grad_norm": 2.05922794342041, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7016551494598389, + "num_tokens": 299875127.0, + "step": 11704 + }, + { + "epoch": 1.2854162090929058, + "grad_norm": 2.3024725914001465, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7370809316635132, + "num_tokens": 299897495.0, + "step": 11705 + }, + { + "epoch": 1.2855260267955195, + "grad_norm": 1.8814899921417236, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7319327592849731, + "num_tokens": 299930292.0, + "step": 11706 + }, + { + "epoch": 1.285635844498133, + "grad_norm": 2.4517290592193604, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7343670129776001, + "num_tokens": 299952126.0, + "step": 11707 + }, + { + "epoch": 1.2857456622007468, + "grad_norm": 2.2278940677642822, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7353081703186035, + "num_tokens": 299978165.0, + "step": 11708 + }, + { + "epoch": 1.2858554799033604, + "grad_norm": 2.1277732849121094, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7258405685424805, + "num_tokens": 300005149.0, + "step": 11709 + }, + { + "epoch": 1.2859652976059741, + "grad_norm": 1.8792823553085327, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7177155017852783, + "num_tokens": 300042045.0, + "step": 11710 + }, + { + "epoch": 1.2860751153085879, + "grad_norm": 2.2998452186584473, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7201186418533325, + "num_tokens": 300068156.0, + "step": 11711 + }, + { + "epoch": 1.2861849330112014, + "grad_norm": 2.343330144882202, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7158188223838806, + "num_tokens": 300092772.0, + "step": 11712 + }, + { + "epoch": 1.286294750713815, + "grad_norm": 2.3644464015960693, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7154165506362915, + "num_tokens": 300116553.0, + "step": 11713 + }, + { + "epoch": 1.2864045684164287, + "grad_norm": 2.2371673583984375, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.730975866317749, + "num_tokens": 300141706.0, + "step": 11714 + }, + { + "epoch": 1.2865143861190425, + "grad_norm": 2.133281946182251, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7019833326339722, + "num_tokens": 300170539.0, + "step": 11715 + }, + { + "epoch": 1.286624203821656, + "grad_norm": 2.502845048904419, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7461341619491577, + "num_tokens": 300191481.0, + "step": 11716 + }, + { + "epoch": 1.2867340215242697, + "grad_norm": 2.4083733558654785, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7265980243682861, + "num_tokens": 300213188.0, + "step": 11717 + }, + { + "epoch": 1.2868438392268833, + "grad_norm": 2.267335891723633, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7025699019432068, + "num_tokens": 300238583.0, + "step": 11718 + }, + { + "epoch": 1.286953656929497, + "grad_norm": 2.0302913188934326, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7074440717697144, + "num_tokens": 300268019.0, + "step": 11719 + }, + { + "epoch": 1.2870634746321108, + "grad_norm": 2.2984097003936768, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7394120097160339, + "num_tokens": 300291528.0, + "step": 11720 + }, + { + "epoch": 1.2871732923347243, + "grad_norm": 2.0623466968536377, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7185107469558716, + "num_tokens": 300321016.0, + "step": 11721 + }, + { + "epoch": 1.287283110037338, + "grad_norm": 2.346158027648926, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7133394479751587, + "num_tokens": 300345198.0, + "step": 11722 + }, + { + "epoch": 1.2873929277399516, + "grad_norm": 2.1727821826934814, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7419211268424988, + "num_tokens": 300372097.0, + "step": 11723 + }, + { + "epoch": 1.2875027454425654, + "grad_norm": 2.3634462356567383, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7111934423446655, + "num_tokens": 300395161.0, + "step": 11724 + }, + { + "epoch": 1.2876125631451791, + "grad_norm": 2.0142202377319336, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7078344821929932, + "num_tokens": 300426656.0, + "step": 11725 + }, + { + "epoch": 1.2877223808477927, + "grad_norm": 2.2568771839141846, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7226724028587341, + "num_tokens": 300453473.0, + "step": 11726 + }, + { + "epoch": 1.2878321985504062, + "grad_norm": 2.2424166202545166, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7363753318786621, + "num_tokens": 300478742.0, + "step": 11727 + }, + { + "epoch": 1.28794201625302, + "grad_norm": 2.21952748298645, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7003549337387085, + "num_tokens": 300507360.0, + "step": 11728 + }, + { + "epoch": 1.2880518339556337, + "grad_norm": 2.315627336502075, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7177777886390686, + "num_tokens": 300533085.0, + "step": 11729 + }, + { + "epoch": 1.2881616516582473, + "grad_norm": 2.399397134780884, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7141829133033752, + "num_tokens": 300556714.0, + "step": 11730 + }, + { + "epoch": 1.288271469360861, + "grad_norm": 2.3129496574401855, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.706606388092041, + "num_tokens": 300580138.0, + "step": 11731 + }, + { + "epoch": 1.2883812870634745, + "grad_norm": 2.1294944286346436, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7098597288131714, + "num_tokens": 300608596.0, + "step": 11732 + }, + { + "epoch": 1.2884911047660883, + "grad_norm": 2.2593750953674316, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.6997461318969727, + "num_tokens": 300635417.0, + "step": 11733 + }, + { + "epoch": 1.288600922468702, + "grad_norm": 2.360264301300049, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7208810448646545, + "num_tokens": 300659737.0, + "step": 11734 + }, + { + "epoch": 1.2887107401713156, + "grad_norm": 2.352921724319458, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7340147495269775, + "num_tokens": 300683654.0, + "step": 11735 + }, + { + "epoch": 1.2888205578739294, + "grad_norm": 1.9478627443313599, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6945568919181824, + "num_tokens": 300719561.0, + "step": 11736 + }, + { + "epoch": 1.2889303755765429, + "grad_norm": 1.9742803573608398, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7370074391365051, + "num_tokens": 300749695.0, + "step": 11737 + }, + { + "epoch": 1.2890401932791566, + "grad_norm": 2.4878251552581787, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7411110997200012, + "num_tokens": 300770457.0, + "step": 11738 + }, + { + "epoch": 1.2891500109817704, + "grad_norm": 2.1767640113830566, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7296063303947449, + "num_tokens": 300798063.0, + "step": 11739 + }, + { + "epoch": 1.289259828684384, + "grad_norm": 2.2054200172424316, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6933335661888123, + "num_tokens": 300825836.0, + "step": 11740 + }, + { + "epoch": 1.2893696463869975, + "grad_norm": 2.0456318855285645, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7248700857162476, + "num_tokens": 300857437.0, + "step": 11741 + }, + { + "epoch": 1.2894794640896112, + "grad_norm": 2.239330768585205, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7028440237045288, + "num_tokens": 300884021.0, + "step": 11742 + }, + { + "epoch": 1.289589281792225, + "grad_norm": 2.2518651485443115, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7178442478179932, + "num_tokens": 300914443.0, + "step": 11743 + }, + { + "epoch": 1.2896990994948385, + "grad_norm": 2.2460901737213135, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7423757314682007, + "num_tokens": 300939409.0, + "step": 11744 + }, + { + "epoch": 1.2898089171974523, + "grad_norm": 2.185769557952881, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.731634795665741, + "num_tokens": 300965251.0, + "step": 11745 + }, + { + "epoch": 1.2899187349000658, + "grad_norm": 2.386197090148926, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7094622850418091, + "num_tokens": 300991045.0, + "step": 11746 + }, + { + "epoch": 1.2900285526026796, + "grad_norm": 2.38992977142334, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7265033721923828, + "num_tokens": 301014052.0, + "step": 11747 + }, + { + "epoch": 1.2901383703052933, + "grad_norm": 2.088312864303589, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.713188648223877, + "num_tokens": 301045437.0, + "step": 11748 + }, + { + "epoch": 1.2902481880079069, + "grad_norm": 2.4132144451141357, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7150302529335022, + "num_tokens": 301067565.0, + "step": 11749 + }, + { + "epoch": 1.2903580057105206, + "grad_norm": 2.1584725379943848, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7373209595680237, + "num_tokens": 301093904.0, + "step": 11750 + }, + { + "epoch": 1.2904678234131342, + "grad_norm": 2.259690284729004, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7082434296607971, + "num_tokens": 301121985.0, + "step": 11751 + }, + { + "epoch": 1.290577641115748, + "grad_norm": 2.1414365768432617, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7173342704772949, + "num_tokens": 301148509.0, + "step": 11752 + }, + { + "epoch": 1.2906874588183614, + "grad_norm": 2.3503406047821045, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7099355459213257, + "num_tokens": 301171686.0, + "step": 11753 + }, + { + "epoch": 1.2907972765209752, + "grad_norm": 2.043118715286255, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.702217161655426, + "num_tokens": 301200980.0, + "step": 11754 + }, + { + "epoch": 1.2909070942235887, + "grad_norm": 2.6799542903900146, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7367451786994934, + "num_tokens": 301219985.0, + "step": 11755 + }, + { + "epoch": 1.2910169119262025, + "grad_norm": 2.6213929653167725, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7319218516349792, + "num_tokens": 301241274.0, + "step": 11756 + }, + { + "epoch": 1.2911267296288162, + "grad_norm": 2.2825543880462646, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7132927775382996, + "num_tokens": 301267947.0, + "step": 11757 + }, + { + "epoch": 1.2912365473314298, + "grad_norm": 2.327453374862671, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6973538398742676, + "num_tokens": 301291904.0, + "step": 11758 + }, + { + "epoch": 1.2913463650340435, + "grad_norm": 2.4385523796081543, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7235880494117737, + "num_tokens": 301313046.0, + "step": 11759 + }, + { + "epoch": 1.291456182736657, + "grad_norm": 2.2484757900238037, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7212008833885193, + "num_tokens": 301337609.0, + "step": 11760 + }, + { + "epoch": 1.2915660004392708, + "grad_norm": 2.2399518489837646, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7455079555511475, + "num_tokens": 301362561.0, + "step": 11761 + }, + { + "epoch": 1.2916758181418846, + "grad_norm": 2.541166067123413, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7376212477684021, + "num_tokens": 301382657.0, + "step": 11762 + }, + { + "epoch": 1.2917856358444981, + "grad_norm": 2.3060719966888428, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7252281904220581, + "num_tokens": 301406209.0, + "step": 11763 + }, + { + "epoch": 1.2918954535471117, + "grad_norm": 2.155228853225708, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7271879315376282, + "num_tokens": 301432514.0, + "step": 11764 + }, + { + "epoch": 1.2920052712497254, + "grad_norm": 2.142791509628296, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7069239616394043, + "num_tokens": 301461400.0, + "step": 11765 + }, + { + "epoch": 1.2921150889523392, + "grad_norm": 2.2860910892486572, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.6958998441696167, + "num_tokens": 301487574.0, + "step": 11766 + }, + { + "epoch": 1.2922249066549527, + "grad_norm": 2.031418800354004, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.696853756904602, + "num_tokens": 301519701.0, + "step": 11767 + }, + { + "epoch": 1.2923347243575665, + "grad_norm": 2.4850423336029053, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7229210138320923, + "num_tokens": 301542938.0, + "step": 11768 + }, + { + "epoch": 1.29244454206018, + "grad_norm": 2.427993059158325, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7046446800231934, + "num_tokens": 301567234.0, + "step": 11769 + }, + { + "epoch": 1.2925543597627938, + "grad_norm": 2.4928877353668213, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7303037643432617, + "num_tokens": 301588687.0, + "step": 11770 + }, + { + "epoch": 1.2926641774654075, + "grad_norm": 2.113229990005493, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7219205498695374, + "num_tokens": 301618012.0, + "step": 11771 + }, + { + "epoch": 1.292773995168021, + "grad_norm": 2.112950325012207, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7332800626754761, + "num_tokens": 301644525.0, + "step": 11772 + }, + { + "epoch": 1.2928838128706348, + "grad_norm": 2.346766710281372, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7397762537002563, + "num_tokens": 301667625.0, + "step": 11773 + }, + { + "epoch": 1.2929936305732483, + "grad_norm": 2.2832348346710205, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7005814909934998, + "num_tokens": 301692806.0, + "step": 11774 + }, + { + "epoch": 1.293103448275862, + "grad_norm": 2.4644126892089844, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7047052979469299, + "num_tokens": 301716507.0, + "step": 11775 + }, + { + "epoch": 1.2932132659784759, + "grad_norm": 2.30472731590271, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7180511355400085, + "num_tokens": 301742552.0, + "step": 11776 + }, + { + "epoch": 1.2933230836810894, + "grad_norm": 2.2842226028442383, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7329561710357666, + "num_tokens": 301766642.0, + "step": 11777 + }, + { + "epoch": 1.293432901383703, + "grad_norm": 1.9901856184005737, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7319025993347168, + "num_tokens": 301798393.0, + "step": 11778 + }, + { + "epoch": 1.2935427190863167, + "grad_norm": 2.0549278259277344, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7469875812530518, + "num_tokens": 301825715.0, + "step": 11779 + }, + { + "epoch": 1.2936525367889304, + "grad_norm": 2.150413751602173, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7244378328323364, + "num_tokens": 301850788.0, + "step": 11780 + }, + { + "epoch": 1.293762354491544, + "grad_norm": 2.165459394454956, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7353769540786743, + "num_tokens": 301875324.0, + "step": 11781 + }, + { + "epoch": 1.2938721721941577, + "grad_norm": 2.40081524848938, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7077052593231201, + "num_tokens": 301897301.0, + "step": 11782 + }, + { + "epoch": 1.2939819898967713, + "grad_norm": 2.416285276412964, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7319312691688538, + "num_tokens": 301918897.0, + "step": 11783 + }, + { + "epoch": 1.294091807599385, + "grad_norm": 2.2773146629333496, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7160199880599976, + "num_tokens": 301944430.0, + "step": 11784 + }, + { + "epoch": 1.2942016253019988, + "grad_norm": 2.379880905151367, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7029070258140564, + "num_tokens": 301968055.0, + "step": 11785 + }, + { + "epoch": 1.2943114430046123, + "grad_norm": 2.290485382080078, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7317808866500854, + "num_tokens": 301992540.0, + "step": 11786 + }, + { + "epoch": 1.294421260707226, + "grad_norm": 2.0289855003356934, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7110334634780884, + "num_tokens": 302021921.0, + "step": 11787 + }, + { + "epoch": 1.2945310784098396, + "grad_norm": 2.7564778327941895, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.736985981464386, + "num_tokens": 302040518.0, + "step": 11788 + }, + { + "epoch": 1.2946408961124534, + "grad_norm": 2.470569133758545, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7410247325897217, + "num_tokens": 302062219.0, + "step": 11789 + }, + { + "epoch": 1.2947507138150671, + "grad_norm": 2.271965265274048, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7628404498100281, + "num_tokens": 302085064.0, + "step": 11790 + }, + { + "epoch": 1.2948605315176807, + "grad_norm": 2.2303645610809326, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7260950207710266, + "num_tokens": 302112782.0, + "step": 11791 + }, + { + "epoch": 1.2949703492202942, + "grad_norm": 2.414961099624634, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.742534875869751, + "num_tokens": 302134412.0, + "step": 11792 + }, + { + "epoch": 1.295080166922908, + "grad_norm": 2.164508819580078, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6902670860290527, + "num_tokens": 302162404.0, + "step": 11793 + }, + { + "epoch": 1.2951899846255217, + "grad_norm": 2.3696200847625732, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7459216117858887, + "num_tokens": 302183444.0, + "step": 11794 + }, + { + "epoch": 1.2952998023281352, + "grad_norm": 2.1418118476867676, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7118656039237976, + "num_tokens": 302210988.0, + "step": 11795 + }, + { + "epoch": 1.295409620030749, + "grad_norm": 2.3321781158447266, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7052329778671265, + "num_tokens": 302234881.0, + "step": 11796 + }, + { + "epoch": 1.2955194377333625, + "grad_norm": 2.408167600631714, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7201023101806641, + "num_tokens": 302257094.0, + "step": 11797 + }, + { + "epoch": 1.2956292554359763, + "grad_norm": 2.250572443008423, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7180888056755066, + "num_tokens": 302280778.0, + "step": 11798 + }, + { + "epoch": 1.29573907313859, + "grad_norm": 2.379249334335327, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7416864633560181, + "num_tokens": 302302314.0, + "step": 11799 + }, + { + "epoch": 1.2958488908412036, + "grad_norm": 2.332303762435913, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.741366982460022, + "num_tokens": 302324531.0, + "step": 11800 + }, + { + "epoch": 1.2959587085438173, + "grad_norm": 2.6406943798065186, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7336357831954956, + "num_tokens": 302343475.0, + "step": 11801 + }, + { + "epoch": 1.2960685262464309, + "grad_norm": 2.2036054134368896, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7117277979850769, + "num_tokens": 302370553.0, + "step": 11802 + }, + { + "epoch": 1.2961783439490446, + "grad_norm": 2.2206196784973145, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.687157928943634, + "num_tokens": 302395967.0, + "step": 11803 + }, + { + "epoch": 1.2962881616516582, + "grad_norm": 2.6535964012145996, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7105436325073242, + "num_tokens": 302415890.0, + "step": 11804 + }, + { + "epoch": 1.296397979354272, + "grad_norm": 2.545992374420166, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7283387184143066, + "num_tokens": 302437678.0, + "step": 11805 + }, + { + "epoch": 1.2965077970568855, + "grad_norm": 2.193674087524414, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7112494111061096, + "num_tokens": 302463816.0, + "step": 11806 + }, + { + "epoch": 1.2966176147594992, + "grad_norm": 1.958115577697754, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7125297784805298, + "num_tokens": 302494508.0, + "step": 11807 + }, + { + "epoch": 1.296727432462113, + "grad_norm": 1.980042815208435, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7146549820899963, + "num_tokens": 302526560.0, + "step": 11808 + }, + { + "epoch": 1.2968372501647265, + "grad_norm": 2.319918632507324, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7221543192863464, + "num_tokens": 302549851.0, + "step": 11809 + }, + { + "epoch": 1.2969470678673403, + "grad_norm": 2.503864288330078, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7302176356315613, + "num_tokens": 302569492.0, + "step": 11810 + }, + { + "epoch": 1.2970568855699538, + "grad_norm": 2.281677484512329, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7424941062927246, + "num_tokens": 302592342.0, + "step": 11811 + }, + { + "epoch": 1.2971667032725676, + "grad_norm": 2.3739311695098877, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7299410104751587, + "num_tokens": 302614220.0, + "step": 11812 + }, + { + "epoch": 1.2972765209751813, + "grad_norm": 2.206589937210083, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7234668731689453, + "num_tokens": 302639430.0, + "step": 11813 + }, + { + "epoch": 1.2973863386777948, + "grad_norm": 2.3740944862365723, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7038593292236328, + "num_tokens": 302664280.0, + "step": 11814 + }, + { + "epoch": 1.2974961563804086, + "grad_norm": 2.1137726306915283, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.727888822555542, + "num_tokens": 302692320.0, + "step": 11815 + }, + { + "epoch": 1.2976059740830221, + "grad_norm": 2.2811596393585205, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7083852291107178, + "num_tokens": 302718526.0, + "step": 11816 + }, + { + "epoch": 1.297715791785636, + "grad_norm": 2.43530011177063, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7146949172019958, + "num_tokens": 302741272.0, + "step": 11817 + }, + { + "epoch": 1.2978256094882494, + "grad_norm": 2.403937339782715, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7149456739425659, + "num_tokens": 302763267.0, + "step": 11818 + }, + { + "epoch": 1.2979354271908632, + "grad_norm": 2.2300045490264893, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7096058130264282, + "num_tokens": 302790398.0, + "step": 11819 + }, + { + "epoch": 1.2980452448934767, + "grad_norm": 2.111863851547241, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7135679721832275, + "num_tokens": 302820365.0, + "step": 11820 + }, + { + "epoch": 1.2981550625960905, + "grad_norm": 2.0052781105041504, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7468780279159546, + "num_tokens": 302849268.0, + "step": 11821 + }, + { + "epoch": 1.2982648802987042, + "grad_norm": 2.2155463695526123, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7508981227874756, + "num_tokens": 302872994.0, + "step": 11822 + }, + { + "epoch": 1.2983746980013178, + "grad_norm": 2.4786829948425293, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7242661118507385, + "num_tokens": 302896930.0, + "step": 11823 + }, + { + "epoch": 1.2984845157039315, + "grad_norm": 2.302535057067871, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7408327460289001, + "num_tokens": 302920272.0, + "step": 11824 + }, + { + "epoch": 1.298594333406545, + "grad_norm": 2.2514147758483887, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7079278826713562, + "num_tokens": 302946203.0, + "step": 11825 + }, + { + "epoch": 1.2987041511091588, + "grad_norm": 2.022082805633545, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.710543155670166, + "num_tokens": 302976396.0, + "step": 11826 + }, + { + "epoch": 1.2988139688117726, + "grad_norm": 2.274742603302002, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7296464443206787, + "num_tokens": 303002190.0, + "step": 11827 + }, + { + "epoch": 1.2989237865143861, + "grad_norm": 2.8489320278167725, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7325164079666138, + "num_tokens": 303019433.0, + "step": 11828 + }, + { + "epoch": 1.2990336042169996, + "grad_norm": 2.380034923553467, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7233490943908691, + "num_tokens": 303042341.0, + "step": 11829 + }, + { + "epoch": 1.2991434219196134, + "grad_norm": 2.168381929397583, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7028795480728149, + "num_tokens": 303072368.0, + "step": 11830 + }, + { + "epoch": 1.2992532396222272, + "grad_norm": 2.2225735187530518, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7313711643218994, + "num_tokens": 303098518.0, + "step": 11831 + }, + { + "epoch": 1.2993630573248407, + "grad_norm": 2.444561004638672, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7052988409996033, + "num_tokens": 303120829.0, + "step": 11832 + }, + { + "epoch": 1.2994728750274545, + "grad_norm": 2.108567953109741, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7056121826171875, + "num_tokens": 303151152.0, + "step": 11833 + }, + { + "epoch": 1.299582692730068, + "grad_norm": 2.1917357444763184, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7287975549697876, + "num_tokens": 303177078.0, + "step": 11834 + }, + { + "epoch": 1.2996925104326817, + "grad_norm": 2.3222975730895996, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7017043828964233, + "num_tokens": 303203999.0, + "step": 11835 + }, + { + "epoch": 1.2998023281352955, + "grad_norm": 2.756338596343994, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7416719198226929, + "num_tokens": 303220966.0, + "step": 11836 + }, + { + "epoch": 1.299912145837909, + "grad_norm": 2.4400041103363037, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7146010994911194, + "num_tokens": 303242786.0, + "step": 11837 + }, + { + "epoch": 1.3000219635405228, + "grad_norm": 2.589003562927246, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7274201512336731, + "num_tokens": 303263674.0, + "step": 11838 + }, + { + "epoch": 1.3001317812431363, + "grad_norm": 2.0547173023223877, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.70899498462677, + "num_tokens": 303293173.0, + "step": 11839 + }, + { + "epoch": 1.30024159894575, + "grad_norm": 2.136213779449463, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7066265940666199, + "num_tokens": 303320418.0, + "step": 11840 + }, + { + "epoch": 1.3003514166483638, + "grad_norm": 2.2134666442871094, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7084275484085083, + "num_tokens": 303346897.0, + "step": 11841 + }, + { + "epoch": 1.3004612343509774, + "grad_norm": 2.2207114696502686, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7369745969772339, + "num_tokens": 303373176.0, + "step": 11842 + }, + { + "epoch": 1.300571052053591, + "grad_norm": 2.1849782466888428, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.70644611120224, + "num_tokens": 303400606.0, + "step": 11843 + }, + { + "epoch": 1.3006808697562047, + "grad_norm": 2.2160589694976807, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7215202450752258, + "num_tokens": 303426299.0, + "step": 11844 + }, + { + "epoch": 1.3007906874588184, + "grad_norm": 2.274721622467041, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7310746908187866, + "num_tokens": 303451785.0, + "step": 11845 + }, + { + "epoch": 1.300900505161432, + "grad_norm": 2.576948881149292, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7271066904067993, + "num_tokens": 303472420.0, + "step": 11846 + }, + { + "epoch": 1.3010103228640457, + "grad_norm": 1.9759901762008667, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7067770957946777, + "num_tokens": 303504079.0, + "step": 11847 + }, + { + "epoch": 1.3011201405666593, + "grad_norm": 2.3081486225128174, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7320904731750488, + "num_tokens": 303528088.0, + "step": 11848 + }, + { + "epoch": 1.301229958269273, + "grad_norm": 2.344348430633545, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7142200469970703, + "num_tokens": 303552875.0, + "step": 11849 + }, + { + "epoch": 1.3013397759718868, + "grad_norm": 2.1082942485809326, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7179521322250366, + "num_tokens": 303582001.0, + "step": 11850 + }, + { + "epoch": 1.3014495936745003, + "grad_norm": 2.1992084980010986, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.720176100730896, + "num_tokens": 303607395.0, + "step": 11851 + }, + { + "epoch": 1.301559411377114, + "grad_norm": 2.439390182495117, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7070577144622803, + "num_tokens": 303629319.0, + "step": 11852 + }, + { + "epoch": 1.3016692290797276, + "grad_norm": 2.1558430194854736, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.732846200466156, + "num_tokens": 303657157.0, + "step": 11853 + }, + { + "epoch": 1.3017790467823414, + "grad_norm": 2.3834757804870605, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7098445892333984, + "num_tokens": 303681429.0, + "step": 11854 + }, + { + "epoch": 1.301888864484955, + "grad_norm": 2.313652992248535, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.729465901851654, + "num_tokens": 303704881.0, + "step": 11855 + }, + { + "epoch": 1.3019986821875686, + "grad_norm": 2.548100471496582, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7427005767822266, + "num_tokens": 303724858.0, + "step": 11856 + }, + { + "epoch": 1.3021084998901822, + "grad_norm": 2.19114089012146, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7112377285957336, + "num_tokens": 303752087.0, + "step": 11857 + }, + { + "epoch": 1.302218317592796, + "grad_norm": 2.7659502029418945, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7249041795730591, + "num_tokens": 303770741.0, + "step": 11858 + }, + { + "epoch": 1.3023281352954097, + "grad_norm": 2.218245267868042, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7179753184318542, + "num_tokens": 303795712.0, + "step": 11859 + }, + { + "epoch": 1.3024379529980232, + "grad_norm": 2.460477352142334, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7262865900993347, + "num_tokens": 303817266.0, + "step": 11860 + }, + { + "epoch": 1.302547770700637, + "grad_norm": 2.555318832397461, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7524545788764954, + "num_tokens": 303835760.0, + "step": 11861 + }, + { + "epoch": 1.3026575884032505, + "grad_norm": 2.261138677597046, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6884884238243103, + "num_tokens": 303863964.0, + "step": 11862 + }, + { + "epoch": 1.3027674061058643, + "grad_norm": 2.3547000885009766, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7199309468269348, + "num_tokens": 303888379.0, + "step": 11863 + }, + { + "epoch": 1.302877223808478, + "grad_norm": 2.208270788192749, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7116984128952026, + "num_tokens": 303915637.0, + "step": 11864 + }, + { + "epoch": 1.3029870415110916, + "grad_norm": 2.3373496532440186, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7170519828796387, + "num_tokens": 303939279.0, + "step": 11865 + }, + { + "epoch": 1.3030968592137053, + "grad_norm": 2.022932529449463, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7458180785179138, + "num_tokens": 303967000.0, + "step": 11866 + }, + { + "epoch": 1.3032066769163189, + "grad_norm": 2.120168447494507, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7220925092697144, + "num_tokens": 303996737.0, + "step": 11867 + }, + { + "epoch": 1.3033164946189326, + "grad_norm": 2.2979867458343506, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7315453886985779, + "num_tokens": 304018587.0, + "step": 11868 + }, + { + "epoch": 1.3034263123215462, + "grad_norm": 2.434086799621582, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7378686666488647, + "num_tokens": 304041678.0, + "step": 11869 + }, + { + "epoch": 1.30353613002416, + "grad_norm": 2.275890588760376, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7239930629730225, + "num_tokens": 304065927.0, + "step": 11870 + }, + { + "epoch": 1.3036459477267734, + "grad_norm": 1.9882193803787231, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7208954095840454, + "num_tokens": 304095957.0, + "step": 11871 + }, + { + "epoch": 1.3037557654293872, + "grad_norm": 2.2971913814544678, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7235639095306396, + "num_tokens": 304123452.0, + "step": 11872 + }, + { + "epoch": 1.303865583132001, + "grad_norm": 2.1075263023376465, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7252195477485657, + "num_tokens": 304152845.0, + "step": 11873 + }, + { + "epoch": 1.3039754008346145, + "grad_norm": 2.1745662689208984, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7230459451675415, + "num_tokens": 304179218.0, + "step": 11874 + }, + { + "epoch": 1.3040852185372283, + "grad_norm": 2.1855366230010986, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7341441512107849, + "num_tokens": 304205364.0, + "step": 11875 + }, + { + "epoch": 1.3041950362398418, + "grad_norm": 2.1318016052246094, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.6996059417724609, + "num_tokens": 304233786.0, + "step": 11876 + }, + { + "epoch": 1.3043048539424555, + "grad_norm": 2.230834484100342, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7205061316490173, + "num_tokens": 304260383.0, + "step": 11877 + }, + { + "epoch": 1.3044146716450693, + "grad_norm": 2.422320604324341, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7263878583908081, + "num_tokens": 304283686.0, + "step": 11878 + }, + { + "epoch": 1.3045244893476828, + "grad_norm": 2.581578493118286, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7136732339859009, + "num_tokens": 304305078.0, + "step": 11879 + }, + { + "epoch": 1.3046343070502964, + "grad_norm": 2.1607918739318848, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7021796107292175, + "num_tokens": 304334485.0, + "step": 11880 + }, + { + "epoch": 1.3047441247529101, + "grad_norm": 2.3766493797302246, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.736011803150177, + "num_tokens": 304359044.0, + "step": 11881 + }, + { + "epoch": 1.3048539424555239, + "grad_norm": 2.3537375926971436, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7108321785926819, + "num_tokens": 304381838.0, + "step": 11882 + }, + { + "epoch": 1.3049637601581374, + "grad_norm": 2.4444315433502197, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7390888929367065, + "num_tokens": 304402989.0, + "step": 11883 + }, + { + "epoch": 1.3050735778607512, + "grad_norm": 2.436250925064087, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7213064432144165, + "num_tokens": 304426002.0, + "step": 11884 + }, + { + "epoch": 1.3051833955633647, + "grad_norm": 2.3534412384033203, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7117396593093872, + "num_tokens": 304449336.0, + "step": 11885 + }, + { + "epoch": 1.3052932132659785, + "grad_norm": 2.3787176609039307, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7354037165641785, + "num_tokens": 304472062.0, + "step": 11886 + }, + { + "epoch": 1.3054030309685922, + "grad_norm": 2.1538264751434326, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7017670273780823, + "num_tokens": 304501374.0, + "step": 11887 + }, + { + "epoch": 1.3055128486712058, + "grad_norm": 2.148975133895874, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7253814935684204, + "num_tokens": 304529966.0, + "step": 11888 + }, + { + "epoch": 1.3056226663738195, + "grad_norm": 1.9676129817962646, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7250992059707642, + "num_tokens": 304560517.0, + "step": 11889 + }, + { + "epoch": 1.305732484076433, + "grad_norm": 1.9387646913528442, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7043216228485107, + "num_tokens": 304593717.0, + "step": 11890 + }, + { + "epoch": 1.3058423017790468, + "grad_norm": 2.6979482173919678, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7472162246704102, + "num_tokens": 304612964.0, + "step": 11891 + }, + { + "epoch": 1.3059521194816606, + "grad_norm": 2.1500675678253174, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7451566457748413, + "num_tokens": 304640084.0, + "step": 11892 + }, + { + "epoch": 1.306061937184274, + "grad_norm": 2.1995670795440674, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7188074588775635, + "num_tokens": 304667703.0, + "step": 11893 + }, + { + "epoch": 1.3061717548868876, + "grad_norm": 2.035059928894043, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7299419641494751, + "num_tokens": 304697970.0, + "step": 11894 + }, + { + "epoch": 1.3062815725895014, + "grad_norm": 2.1794838905334473, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7281512022018433, + "num_tokens": 304724762.0, + "step": 11895 + }, + { + "epoch": 1.3063913902921152, + "grad_norm": 2.441673755645752, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7234630584716797, + "num_tokens": 304745082.0, + "step": 11896 + }, + { + "epoch": 1.3065012079947287, + "grad_norm": 2.4601879119873047, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7397069334983826, + "num_tokens": 304766967.0, + "step": 11897 + }, + { + "epoch": 1.3066110256973424, + "grad_norm": 2.363072156906128, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7342947721481323, + "num_tokens": 304793321.0, + "step": 11898 + }, + { + "epoch": 1.306720843399956, + "grad_norm": 2.2528538703918457, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6983351707458496, + "num_tokens": 304817320.0, + "step": 11899 + }, + { + "epoch": 1.3068306611025697, + "grad_norm": 2.2396793365478516, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6945922374725342, + "num_tokens": 304845569.0, + "step": 11900 + }, + { + "epoch": 1.3069404788051835, + "grad_norm": 2.1448471546173096, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7156601548194885, + "num_tokens": 304871061.0, + "step": 11901 + }, + { + "epoch": 1.307050296507797, + "grad_norm": 2.0349514484405518, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7064558863639832, + "num_tokens": 304901197.0, + "step": 11902 + }, + { + "epoch": 1.3071601142104108, + "grad_norm": 2.2909798622131348, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7254059314727783, + "num_tokens": 304926096.0, + "step": 11903 + }, + { + "epoch": 1.3072699319130243, + "grad_norm": 2.353671073913574, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7253129482269287, + "num_tokens": 304949754.0, + "step": 11904 + }, + { + "epoch": 1.307379749615638, + "grad_norm": 2.2260990142822266, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7016756534576416, + "num_tokens": 304979448.0, + "step": 11905 + }, + { + "epoch": 1.3074895673182518, + "grad_norm": 2.41477370262146, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7278366088867188, + "num_tokens": 304999763.0, + "step": 11906 + }, + { + "epoch": 1.3075993850208654, + "grad_norm": 2.064945936203003, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.698906421661377, + "num_tokens": 305030205.0, + "step": 11907 + }, + { + "epoch": 1.307709202723479, + "grad_norm": 1.9962589740753174, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7072571516036987, + "num_tokens": 305060099.0, + "step": 11908 + }, + { + "epoch": 1.3078190204260927, + "grad_norm": 2.1639490127563477, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7079030275344849, + "num_tokens": 305088419.0, + "step": 11909 + }, + { + "epoch": 1.3079288381287064, + "grad_norm": 2.2261228561401367, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7062832117080688, + "num_tokens": 305114674.0, + "step": 11910 + }, + { + "epoch": 1.30803865583132, + "grad_norm": 2.474439859390259, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7293730974197388, + "num_tokens": 305136900.0, + "step": 11911 + }, + { + "epoch": 1.3081484735339337, + "grad_norm": 2.654897928237915, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7388569116592407, + "num_tokens": 305156823.0, + "step": 11912 + }, + { + "epoch": 1.3082582912365472, + "grad_norm": 2.16400408744812, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7076308727264404, + "num_tokens": 305185458.0, + "step": 11913 + }, + { + "epoch": 1.308368108939161, + "grad_norm": 2.5173721313476562, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.750305712223053, + "num_tokens": 305204498.0, + "step": 11914 + }, + { + "epoch": 1.3084779266417748, + "grad_norm": 1.9765193462371826, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7096996903419495, + "num_tokens": 305235699.0, + "step": 11915 + }, + { + "epoch": 1.3085877443443883, + "grad_norm": 2.3188607692718506, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7053221464157104, + "num_tokens": 305260275.0, + "step": 11916 + }, + { + "epoch": 1.308697562047002, + "grad_norm": 2.362778425216675, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7465388774871826, + "num_tokens": 305283647.0, + "step": 11917 + }, + { + "epoch": 1.3088073797496156, + "grad_norm": 1.9708601236343384, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6981146931648254, + "num_tokens": 305316069.0, + "step": 11918 + }, + { + "epoch": 1.3089171974522293, + "grad_norm": 2.2922840118408203, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7231724858283997, + "num_tokens": 305341407.0, + "step": 11919 + }, + { + "epoch": 1.309027015154843, + "grad_norm": 2.249844789505005, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7364595532417297, + "num_tokens": 305367125.0, + "step": 11920 + }, + { + "epoch": 1.3091368328574566, + "grad_norm": 2.1094748973846436, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7075828313827515, + "num_tokens": 305394979.0, + "step": 11921 + }, + { + "epoch": 1.3092466505600702, + "grad_norm": 2.2795357704162598, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7232270240783691, + "num_tokens": 305420991.0, + "step": 11922 + }, + { + "epoch": 1.309356468262684, + "grad_norm": 2.476822853088379, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7283753752708435, + "num_tokens": 305441887.0, + "step": 11923 + }, + { + "epoch": 1.3094662859652977, + "grad_norm": 2.2346551418304443, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7481539249420166, + "num_tokens": 305467630.0, + "step": 11924 + }, + { + "epoch": 1.3095761036679112, + "grad_norm": 2.418163776397705, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7290653586387634, + "num_tokens": 305491331.0, + "step": 11925 + }, + { + "epoch": 1.309685921370525, + "grad_norm": 2.2041900157928467, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7115779519081116, + "num_tokens": 305516245.0, + "step": 11926 + }, + { + "epoch": 1.3097957390731385, + "grad_norm": 2.4535534381866455, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7365351915359497, + "num_tokens": 305538755.0, + "step": 11927 + }, + { + "epoch": 1.3099055567757523, + "grad_norm": 2.150244951248169, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7267842292785645, + "num_tokens": 305566232.0, + "step": 11928 + }, + { + "epoch": 1.310015374478366, + "grad_norm": 1.964901089668274, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.686618447303772, + "num_tokens": 305599148.0, + "step": 11929 + }, + { + "epoch": 1.3101251921809796, + "grad_norm": 2.1484692096710205, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.6986502408981323, + "num_tokens": 305625300.0, + "step": 11930 + }, + { + "epoch": 1.3102350098835933, + "grad_norm": 2.1605169773101807, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7145642042160034, + "num_tokens": 305655938.0, + "step": 11931 + }, + { + "epoch": 1.3103448275862069, + "grad_norm": 1.9878700971603394, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7241477370262146, + "num_tokens": 305686749.0, + "step": 11932 + }, + { + "epoch": 1.3104546452888206, + "grad_norm": 2.1074821949005127, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7092588543891907, + "num_tokens": 305715706.0, + "step": 11933 + }, + { + "epoch": 1.3105644629914341, + "grad_norm": 2.0719680786132812, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7080506086349487, + "num_tokens": 305744184.0, + "step": 11934 + }, + { + "epoch": 1.310674280694048, + "grad_norm": 2.297483444213867, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6887078285217285, + "num_tokens": 305774267.0, + "step": 11935 + }, + { + "epoch": 1.3107840983966614, + "grad_norm": 2.3077738285064697, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7115791440010071, + "num_tokens": 305800699.0, + "step": 11936 + }, + { + "epoch": 1.3108939160992752, + "grad_norm": 2.3553383350372314, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.699470043182373, + "num_tokens": 305824823.0, + "step": 11937 + }, + { + "epoch": 1.311003733801889, + "grad_norm": 2.2253763675689697, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7209555506706238, + "num_tokens": 305850206.0, + "step": 11938 + }, + { + "epoch": 1.3111135515045025, + "grad_norm": 2.402031421661377, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7265965342521667, + "num_tokens": 305873214.0, + "step": 11939 + }, + { + "epoch": 1.3112233692071162, + "grad_norm": 2.2256178855895996, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7377262115478516, + "num_tokens": 305899247.0, + "step": 11940 + }, + { + "epoch": 1.3113331869097298, + "grad_norm": 2.1125943660736084, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6947757601737976, + "num_tokens": 305928446.0, + "step": 11941 + }, + { + "epoch": 1.3114430046123435, + "grad_norm": 2.306063175201416, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7072373628616333, + "num_tokens": 305953971.0, + "step": 11942 + }, + { + "epoch": 1.3115528223149573, + "grad_norm": 2.450432300567627, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7168930768966675, + "num_tokens": 305977360.0, + "step": 11943 + }, + { + "epoch": 1.3116626400175708, + "grad_norm": 2.22056245803833, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6979374885559082, + "num_tokens": 306003420.0, + "step": 11944 + }, + { + "epoch": 1.3117724577201844, + "grad_norm": 2.237327814102173, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.703787088394165, + "num_tokens": 306028567.0, + "step": 11945 + }, + { + "epoch": 1.3118822754227981, + "grad_norm": 2.2595865726470947, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7004503607749939, + "num_tokens": 306054129.0, + "step": 11946 + }, + { + "epoch": 1.3119920931254119, + "grad_norm": 2.082859516143799, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7134011387825012, + "num_tokens": 306080241.0, + "step": 11947 + }, + { + "epoch": 1.3121019108280254, + "grad_norm": 2.0855727195739746, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7053567171096802, + "num_tokens": 306107754.0, + "step": 11948 + }, + { + "epoch": 1.3122117285306392, + "grad_norm": 2.1908798217773438, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6968950033187866, + "num_tokens": 306133318.0, + "step": 11949 + }, + { + "epoch": 1.3123215462332527, + "grad_norm": 2.2111475467681885, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7029799222946167, + "num_tokens": 306160816.0, + "step": 11950 + }, + { + "epoch": 1.3124313639358665, + "grad_norm": 2.251420736312866, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7268205285072327, + "num_tokens": 306184879.0, + "step": 11951 + }, + { + "epoch": 1.3125411816384802, + "grad_norm": 2.2949202060699463, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7112108469009399, + "num_tokens": 306208833.0, + "step": 11952 + }, + { + "epoch": 1.3126509993410937, + "grad_norm": 2.163893461227417, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7084038257598877, + "num_tokens": 306234238.0, + "step": 11953 + }, + { + "epoch": 1.3127608170437075, + "grad_norm": 2.2488491535186768, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6985445618629456, + "num_tokens": 306262176.0, + "step": 11954 + }, + { + "epoch": 1.312870634746321, + "grad_norm": 2.314444065093994, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7188737392425537, + "num_tokens": 306286323.0, + "step": 11955 + }, + { + "epoch": 1.3129804524489348, + "grad_norm": 2.669879674911499, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7278177738189697, + "num_tokens": 306305295.0, + "step": 11956 + }, + { + "epoch": 1.3130902701515486, + "grad_norm": 2.4332361221313477, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7267751097679138, + "num_tokens": 306327137.0, + "step": 11957 + }, + { + "epoch": 1.313200087854162, + "grad_norm": 2.375081777572632, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7268650531768799, + "num_tokens": 306347950.0, + "step": 11958 + }, + { + "epoch": 1.3133099055567756, + "grad_norm": 2.153926372528076, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6970025300979614, + "num_tokens": 306377767.0, + "step": 11959 + }, + { + "epoch": 1.3134197232593894, + "grad_norm": 2.5791709423065186, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.719841718673706, + "num_tokens": 306398015.0, + "step": 11960 + }, + { + "epoch": 1.3135295409620031, + "grad_norm": 2.0695252418518066, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.705717146396637, + "num_tokens": 306427969.0, + "step": 11961 + }, + { + "epoch": 1.3136393586646167, + "grad_norm": 2.115326166152954, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.710609495639801, + "num_tokens": 306459610.0, + "step": 11962 + }, + { + "epoch": 1.3137491763672304, + "grad_norm": 2.357034921646118, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7206381559371948, + "num_tokens": 306484023.0, + "step": 11963 + }, + { + "epoch": 1.313858994069844, + "grad_norm": 2.344461679458618, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7496513724327087, + "num_tokens": 306507075.0, + "step": 11964 + }, + { + "epoch": 1.3139688117724577, + "grad_norm": 2.3131914138793945, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6987152099609375, + "num_tokens": 306532906.0, + "step": 11965 + }, + { + "epoch": 1.3140786294750715, + "grad_norm": 2.0698344707489014, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7049930095672607, + "num_tokens": 306560812.0, + "step": 11966 + }, + { + "epoch": 1.314188447177685, + "grad_norm": 2.2042489051818848, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7329088449478149, + "num_tokens": 306586422.0, + "step": 11967 + }, + { + "epoch": 1.3142982648802988, + "grad_norm": 2.0690832138061523, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6868413090705872, + "num_tokens": 306616254.0, + "step": 11968 + }, + { + "epoch": 1.3144080825829123, + "grad_norm": 2.4617233276367188, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7325592637062073, + "num_tokens": 306637501.0, + "step": 11969 + }, + { + "epoch": 1.314517900285526, + "grad_norm": 2.1140143871307373, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7134963274002075, + "num_tokens": 306666541.0, + "step": 11970 + }, + { + "epoch": 1.3146277179881398, + "grad_norm": 2.422367572784424, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7090790867805481, + "num_tokens": 306689239.0, + "step": 11971 + }, + { + "epoch": 1.3147375356907534, + "grad_norm": 2.4489212036132812, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7270507216453552, + "num_tokens": 306710792.0, + "step": 11972 + }, + { + "epoch": 1.314847353393367, + "grad_norm": 2.1845381259918213, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6953263282775879, + "num_tokens": 306739057.0, + "step": 11973 + }, + { + "epoch": 1.3149571710959806, + "grad_norm": 2.5361053943634033, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7143853306770325, + "num_tokens": 306760180.0, + "step": 11974 + }, + { + "epoch": 1.3150669887985944, + "grad_norm": 2.25646710395813, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7141836285591125, + "num_tokens": 306784493.0, + "step": 11975 + }, + { + "epoch": 1.315176806501208, + "grad_norm": 2.0925071239471436, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7093380689620972, + "num_tokens": 306812951.0, + "step": 11976 + }, + { + "epoch": 1.3152866242038217, + "grad_norm": 2.241763114929199, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7270137071609497, + "num_tokens": 306836690.0, + "step": 11977 + }, + { + "epoch": 1.3153964419064352, + "grad_norm": 2.0320496559143066, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7026647925376892, + "num_tokens": 306868010.0, + "step": 11978 + }, + { + "epoch": 1.315506259609049, + "grad_norm": 2.4351537227630615, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7129456996917725, + "num_tokens": 306890386.0, + "step": 11979 + }, + { + "epoch": 1.3156160773116627, + "grad_norm": 2.1506435871124268, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7128118276596069, + "num_tokens": 306915537.0, + "step": 11980 + }, + { + "epoch": 1.3157258950142763, + "grad_norm": 1.8461288213729858, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7238898277282715, + "num_tokens": 306950477.0, + "step": 11981 + }, + { + "epoch": 1.31583571271689, + "grad_norm": 2.450840711593628, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.736984133720398, + "num_tokens": 306971786.0, + "step": 11982 + }, + { + "epoch": 1.3159455304195036, + "grad_norm": 2.235402822494507, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7259956002235413, + "num_tokens": 306998189.0, + "step": 11983 + }, + { + "epoch": 1.3160553481221173, + "grad_norm": 2.289867401123047, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7296105623245239, + "num_tokens": 307024197.0, + "step": 11984 + }, + { + "epoch": 1.3161651658247309, + "grad_norm": 2.0560550689697266, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7264326214790344, + "num_tokens": 307051990.0, + "step": 11985 + }, + { + "epoch": 1.3162749835273446, + "grad_norm": 2.2204885482788086, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7048957347869873, + "num_tokens": 307080194.0, + "step": 11986 + }, + { + "epoch": 1.3163848012299582, + "grad_norm": 2.4920096397399902, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7307509183883667, + "num_tokens": 307100801.0, + "step": 11987 + }, + { + "epoch": 1.316494618932572, + "grad_norm": 2.383072853088379, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7064321637153625, + "num_tokens": 307123166.0, + "step": 11988 + }, + { + "epoch": 1.3166044366351857, + "grad_norm": 2.1497082710266113, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7426655888557434, + "num_tokens": 307151633.0, + "step": 11989 + }, + { + "epoch": 1.3167142543377992, + "grad_norm": 1.9565556049346924, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.700575590133667, + "num_tokens": 307182758.0, + "step": 11990 + }, + { + "epoch": 1.316824072040413, + "grad_norm": 2.3042314052581787, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7292892932891846, + "num_tokens": 307207705.0, + "step": 11991 + }, + { + "epoch": 1.3169338897430265, + "grad_norm": 2.273696184158325, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7035586833953857, + "num_tokens": 307234066.0, + "step": 11992 + }, + { + "epoch": 1.3170437074456403, + "grad_norm": 2.155250072479248, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7096934914588928, + "num_tokens": 307263059.0, + "step": 11993 + }, + { + "epoch": 1.317153525148254, + "grad_norm": 2.3941450119018555, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.732258141040802, + "num_tokens": 307286627.0, + "step": 11994 + }, + { + "epoch": 1.3172633428508675, + "grad_norm": 2.086103677749634, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7254502773284912, + "num_tokens": 307314665.0, + "step": 11995 + }, + { + "epoch": 1.3173731605534813, + "grad_norm": 2.4214413166046143, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7235016822814941, + "num_tokens": 307337080.0, + "step": 11996 + }, + { + "epoch": 1.3174829782560948, + "grad_norm": 2.173325300216675, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.751720666885376, + "num_tokens": 307363161.0, + "step": 11997 + }, + { + "epoch": 1.3175927959587086, + "grad_norm": 2.168544054031372, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6918696761131287, + "num_tokens": 307391808.0, + "step": 11998 + }, + { + "epoch": 1.3177026136613221, + "grad_norm": 2.1143958568573, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7070952653884888, + "num_tokens": 307420234.0, + "step": 11999 + }, + { + "epoch": 1.3178124313639359, + "grad_norm": 2.224243640899658, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7123855948448181, + "num_tokens": 307447678.0, + "step": 12000 + }, + { + "epoch": 1.3179222490665494, + "grad_norm": 2.597400188446045, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7287521362304688, + "num_tokens": 307466819.0, + "step": 12001 + }, + { + "epoch": 1.3180320667691632, + "grad_norm": 2.3107144832611084, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7115925550460815, + "num_tokens": 307492703.0, + "step": 12002 + }, + { + "epoch": 1.318141884471777, + "grad_norm": 2.199955940246582, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7208971381187439, + "num_tokens": 307517049.0, + "step": 12003 + }, + { + "epoch": 1.3182517021743905, + "grad_norm": 2.2700905799865723, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7021389603614807, + "num_tokens": 307542473.0, + "step": 12004 + }, + { + "epoch": 1.3183615198770042, + "grad_norm": 2.3515853881835938, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7250012159347534, + "num_tokens": 307567446.0, + "step": 12005 + }, + { + "epoch": 1.3184713375796178, + "grad_norm": 2.391502857208252, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7073314785957336, + "num_tokens": 307592437.0, + "step": 12006 + }, + { + "epoch": 1.3185811552822315, + "grad_norm": 2.233513832092285, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7025352716445923, + "num_tokens": 307619791.0, + "step": 12007 + }, + { + "epoch": 1.3186909729848453, + "grad_norm": 2.339078903198242, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7276414036750793, + "num_tokens": 307643351.0, + "step": 12008 + }, + { + "epoch": 1.3188007906874588, + "grad_norm": 2.33889102935791, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7291192412376404, + "num_tokens": 307667379.0, + "step": 12009 + }, + { + "epoch": 1.3189106083900723, + "grad_norm": 2.3739259243011475, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7255827784538269, + "num_tokens": 307691506.0, + "step": 12010 + }, + { + "epoch": 1.319020426092686, + "grad_norm": 2.523035764694214, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7399492263793945, + "num_tokens": 307711490.0, + "step": 12011 + }, + { + "epoch": 1.3191302437952999, + "grad_norm": 2.4924118518829346, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.721621036529541, + "num_tokens": 307734691.0, + "step": 12012 + }, + { + "epoch": 1.3192400614979134, + "grad_norm": 2.101750135421753, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7169593572616577, + "num_tokens": 307761356.0, + "step": 12013 + }, + { + "epoch": 1.3193498792005272, + "grad_norm": 2.2403564453125, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7207628488540649, + "num_tokens": 307787634.0, + "step": 12014 + }, + { + "epoch": 1.3194596969031407, + "grad_norm": 2.5421621799468994, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7366665601730347, + "num_tokens": 307809343.0, + "step": 12015 + }, + { + "epoch": 1.3195695146057544, + "grad_norm": 2.517820358276367, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7366313934326172, + "num_tokens": 307831171.0, + "step": 12016 + }, + { + "epoch": 1.3196793323083682, + "grad_norm": 2.3928186893463135, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7353482246398926, + "num_tokens": 307853873.0, + "step": 12017 + }, + { + "epoch": 1.3197891500109817, + "grad_norm": 2.3945529460906982, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7258232831954956, + "num_tokens": 307876931.0, + "step": 12018 + }, + { + "epoch": 1.3198989677135955, + "grad_norm": 2.6196017265319824, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7361194491386414, + "num_tokens": 307897031.0, + "step": 12019 + }, + { + "epoch": 1.320008785416209, + "grad_norm": 2.082068920135498, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7407258152961731, + "num_tokens": 307925014.0, + "step": 12020 + }, + { + "epoch": 1.3201186031188228, + "grad_norm": 2.2487714290618896, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7229663133621216, + "num_tokens": 307950349.0, + "step": 12021 + }, + { + "epoch": 1.3202284208214365, + "grad_norm": 2.1894497871398926, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7341681122779846, + "num_tokens": 307976893.0, + "step": 12022 + }, + { + "epoch": 1.32033823852405, + "grad_norm": 2.420978546142578, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7328070998191833, + "num_tokens": 307998473.0, + "step": 12023 + }, + { + "epoch": 1.3204480562266636, + "grad_norm": 2.242497444152832, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.704776406288147, + "num_tokens": 308025208.0, + "step": 12024 + }, + { + "epoch": 1.3205578739292774, + "grad_norm": 2.192828893661499, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7031116485595703, + "num_tokens": 308052885.0, + "step": 12025 + }, + { + "epoch": 1.3206676916318911, + "grad_norm": 2.1825063228607178, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7265143394470215, + "num_tokens": 308079500.0, + "step": 12026 + }, + { + "epoch": 1.3207775093345047, + "grad_norm": 2.121908187866211, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7049773931503296, + "num_tokens": 308108144.0, + "step": 12027 + }, + { + "epoch": 1.3208873270371184, + "grad_norm": 2.179574728012085, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7303131818771362, + "num_tokens": 308134432.0, + "step": 12028 + }, + { + "epoch": 1.320997144739732, + "grad_norm": 2.3814799785614014, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7384150624275208, + "num_tokens": 308156625.0, + "step": 12029 + }, + { + "epoch": 1.3211069624423457, + "grad_norm": 2.4681143760681152, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7086195945739746, + "num_tokens": 308179002.0, + "step": 12030 + }, + { + "epoch": 1.3212167801449595, + "grad_norm": 2.1976823806762695, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.720119833946228, + "num_tokens": 308205185.0, + "step": 12031 + }, + { + "epoch": 1.321326597847573, + "grad_norm": 2.5554332733154297, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.723656415939331, + "num_tokens": 308225796.0, + "step": 12032 + }, + { + "epoch": 1.3214364155501868, + "grad_norm": 2.217636823654175, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7325235605239868, + "num_tokens": 308250614.0, + "step": 12033 + }, + { + "epoch": 1.3215462332528003, + "grad_norm": 2.130763053894043, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7145727872848511, + "num_tokens": 308277974.0, + "step": 12034 + }, + { + "epoch": 1.321656050955414, + "grad_norm": 2.374724864959717, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7355262041091919, + "num_tokens": 308302281.0, + "step": 12035 + }, + { + "epoch": 1.3217658686580278, + "grad_norm": 2.0704047679901123, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7185754776000977, + "num_tokens": 308331570.0, + "step": 12036 + }, + { + "epoch": 1.3218756863606413, + "grad_norm": 2.3011364936828613, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7163898348808289, + "num_tokens": 308355639.0, + "step": 12037 + }, + { + "epoch": 1.3219855040632549, + "grad_norm": 2.2563202381134033, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7374037504196167, + "num_tokens": 308380852.0, + "step": 12038 + }, + { + "epoch": 1.3220953217658686, + "grad_norm": 2.1276915073394775, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7061754465103149, + "num_tokens": 308410047.0, + "step": 12039 + }, + { + "epoch": 1.3222051394684824, + "grad_norm": 2.2652764320373535, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.708747148513794, + "num_tokens": 308434331.0, + "step": 12040 + }, + { + "epoch": 1.322314957171096, + "grad_norm": 2.090996026992798, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7081046104431152, + "num_tokens": 308464157.0, + "step": 12041 + }, + { + "epoch": 1.3224247748737097, + "grad_norm": 2.0542807579040527, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.70340895652771, + "num_tokens": 308493860.0, + "step": 12042 + }, + { + "epoch": 1.3225345925763232, + "grad_norm": 2.1870641708374023, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7142007946968079, + "num_tokens": 308520561.0, + "step": 12043 + }, + { + "epoch": 1.322644410278937, + "grad_norm": 2.419095993041992, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7330652475357056, + "num_tokens": 308543174.0, + "step": 12044 + }, + { + "epoch": 1.3227542279815507, + "grad_norm": 2.4065327644348145, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7264304757118225, + "num_tokens": 308565082.0, + "step": 12045 + }, + { + "epoch": 1.3228640456841643, + "grad_norm": 2.1997532844543457, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7153565883636475, + "num_tokens": 308593088.0, + "step": 12046 + }, + { + "epoch": 1.322973863386778, + "grad_norm": 2.289015054702759, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.697298526763916, + "num_tokens": 308620857.0, + "step": 12047 + }, + { + "epoch": 1.3230836810893916, + "grad_norm": 2.244438409805298, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7342570424079895, + "num_tokens": 308645293.0, + "step": 12048 + }, + { + "epoch": 1.3231934987920053, + "grad_norm": 2.178663969039917, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7195369005203247, + "num_tokens": 308672503.0, + "step": 12049 + }, + { + "epoch": 1.3233033164946189, + "grad_norm": 1.9689421653747559, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7367318272590637, + "num_tokens": 308702397.0, + "step": 12050 + }, + { + "epoch": 1.3234131341972326, + "grad_norm": 2.2405755519866943, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7235575318336487, + "num_tokens": 308728953.0, + "step": 12051 + }, + { + "epoch": 1.3235229518998461, + "grad_norm": 2.234609365463257, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7307798862457275, + "num_tokens": 308753718.0, + "step": 12052 + }, + { + "epoch": 1.32363276960246, + "grad_norm": 2.0820152759552, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7226549386978149, + "num_tokens": 308783581.0, + "step": 12053 + }, + { + "epoch": 1.3237425873050737, + "grad_norm": 2.004617929458618, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7099643349647522, + "num_tokens": 308817421.0, + "step": 12054 + }, + { + "epoch": 1.3238524050076872, + "grad_norm": 1.9957222938537598, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.71950364112854, + "num_tokens": 308846451.0, + "step": 12055 + }, + { + "epoch": 1.323962222710301, + "grad_norm": 2.3395869731903076, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7290260195732117, + "num_tokens": 308868622.0, + "step": 12056 + }, + { + "epoch": 1.3240720404129145, + "grad_norm": 2.1976912021636963, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7325940132141113, + "num_tokens": 308893705.0, + "step": 12057 + }, + { + "epoch": 1.3241818581155282, + "grad_norm": 2.463006019592285, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7325831651687622, + "num_tokens": 308915459.0, + "step": 12058 + }, + { + "epoch": 1.324291675818142, + "grad_norm": 2.525292158126831, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7258585691452026, + "num_tokens": 308936957.0, + "step": 12059 + }, + { + "epoch": 1.3244014935207555, + "grad_norm": 2.1327013969421387, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7250024080276489, + "num_tokens": 308963831.0, + "step": 12060 + }, + { + "epoch": 1.324511311223369, + "grad_norm": 2.3305699825286865, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7191656231880188, + "num_tokens": 308989088.0, + "step": 12061 + }, + { + "epoch": 1.3246211289259828, + "grad_norm": 2.5805044174194336, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7312295436859131, + "num_tokens": 309009977.0, + "step": 12062 + }, + { + "epoch": 1.3247309466285966, + "grad_norm": 2.214844226837158, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6979530453681946, + "num_tokens": 309040063.0, + "step": 12063 + }, + { + "epoch": 1.3248407643312101, + "grad_norm": 2.4900436401367188, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7065783739089966, + "num_tokens": 309062905.0, + "step": 12064 + }, + { + "epoch": 1.3249505820338239, + "grad_norm": 1.8816295862197876, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7023134827613831, + "num_tokens": 309099336.0, + "step": 12065 + }, + { + "epoch": 1.3250603997364374, + "grad_norm": 2.417726755142212, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7185986638069153, + "num_tokens": 309121144.0, + "step": 12066 + }, + { + "epoch": 1.3251702174390512, + "grad_norm": 2.5172038078308105, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7192394137382507, + "num_tokens": 309142151.0, + "step": 12067 + }, + { + "epoch": 1.325280035141665, + "grad_norm": 2.3045597076416016, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7209614515304565, + "num_tokens": 309167302.0, + "step": 12068 + }, + { + "epoch": 1.3253898528442785, + "grad_norm": 2.3256030082702637, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7030925750732422, + "num_tokens": 309192864.0, + "step": 12069 + }, + { + "epoch": 1.3254996705468922, + "grad_norm": 2.0473110675811768, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7119449973106384, + "num_tokens": 309220967.0, + "step": 12070 + }, + { + "epoch": 1.3256094882495058, + "grad_norm": 2.3613650798797607, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7188015580177307, + "num_tokens": 309243727.0, + "step": 12071 + }, + { + "epoch": 1.3257193059521195, + "grad_norm": 2.1448731422424316, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7160806655883789, + "num_tokens": 309271660.0, + "step": 12072 + }, + { + "epoch": 1.3258291236547333, + "grad_norm": 2.063080072402954, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7125124335289001, + "num_tokens": 309301969.0, + "step": 12073 + }, + { + "epoch": 1.3259389413573468, + "grad_norm": 2.320643663406372, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7466107606887817, + "num_tokens": 309324063.0, + "step": 12074 + }, + { + "epoch": 1.3260487590599603, + "grad_norm": 2.4333105087280273, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7101114988327026, + "num_tokens": 309346824.0, + "step": 12075 + }, + { + "epoch": 1.326158576762574, + "grad_norm": 2.0607268810272217, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7389935255050659, + "num_tokens": 309375765.0, + "step": 12076 + }, + { + "epoch": 1.3262683944651878, + "grad_norm": 2.1491053104400635, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7051472663879395, + "num_tokens": 309405386.0, + "step": 12077 + }, + { + "epoch": 1.3263782121678014, + "grad_norm": 2.0674266815185547, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7279430627822876, + "num_tokens": 309434379.0, + "step": 12078 + }, + { + "epoch": 1.3264880298704151, + "grad_norm": 2.18316912651062, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7073274850845337, + "num_tokens": 309462318.0, + "step": 12079 + }, + { + "epoch": 1.3265978475730287, + "grad_norm": 2.442185163497925, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7120615243911743, + "num_tokens": 309484580.0, + "step": 12080 + }, + { + "epoch": 1.3267076652756424, + "grad_norm": 2.327794313430786, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7301514744758606, + "num_tokens": 309509438.0, + "step": 12081 + }, + { + "epoch": 1.3268174829782562, + "grad_norm": 2.0820839405059814, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.741858959197998, + "num_tokens": 309536810.0, + "step": 12082 + }, + { + "epoch": 1.3269273006808697, + "grad_norm": 2.392604112625122, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7281903624534607, + "num_tokens": 309559673.0, + "step": 12083 + }, + { + "epoch": 1.3270371183834835, + "grad_norm": 2.4347474575042725, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7091327905654907, + "num_tokens": 309583020.0, + "step": 12084 + }, + { + "epoch": 1.327146936086097, + "grad_norm": 2.1042404174804688, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7218811511993408, + "num_tokens": 309611990.0, + "step": 12085 + }, + { + "epoch": 1.3272567537887108, + "grad_norm": 2.248443365097046, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7292499542236328, + "num_tokens": 309636170.0, + "step": 12086 + }, + { + "epoch": 1.3273665714913245, + "grad_norm": 2.3445863723754883, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7427181005477905, + "num_tokens": 309657512.0, + "step": 12087 + }, + { + "epoch": 1.327476389193938, + "grad_norm": 2.3127403259277344, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.721186637878418, + "num_tokens": 309683821.0, + "step": 12088 + }, + { + "epoch": 1.3275862068965516, + "grad_norm": 2.1406219005584717, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7214912176132202, + "num_tokens": 309711038.0, + "step": 12089 + }, + { + "epoch": 1.3276960245991654, + "grad_norm": 2.242717981338501, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7331901788711548, + "num_tokens": 309736073.0, + "step": 12090 + }, + { + "epoch": 1.3278058423017791, + "grad_norm": 2.1627159118652344, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7223756909370422, + "num_tokens": 309762829.0, + "step": 12091 + }, + { + "epoch": 1.3279156600043927, + "grad_norm": 2.2726166248321533, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7166580557823181, + "num_tokens": 309788474.0, + "step": 12092 + }, + { + "epoch": 1.3280254777070064, + "grad_norm": 2.0843846797943115, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7244686484336853, + "num_tokens": 309818749.0, + "step": 12093 + }, + { + "epoch": 1.32813529540962, + "grad_norm": 2.1282413005828857, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7211594581604004, + "num_tokens": 309845003.0, + "step": 12094 + }, + { + "epoch": 1.3282451131122337, + "grad_norm": 2.312469482421875, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7074328660964966, + "num_tokens": 309872344.0, + "step": 12095 + }, + { + "epoch": 1.3283549308148475, + "grad_norm": 2.210310220718384, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7105628848075867, + "num_tokens": 309900450.0, + "step": 12096 + }, + { + "epoch": 1.328464748517461, + "grad_norm": 2.117870569229126, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6896883249282837, + "num_tokens": 309930363.0, + "step": 12097 + }, + { + "epoch": 1.3285745662200747, + "grad_norm": 2.418779134750366, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7229194641113281, + "num_tokens": 309953106.0, + "step": 12098 + }, + { + "epoch": 1.3286843839226883, + "grad_norm": 2.277409076690674, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7375708818435669, + "num_tokens": 309976927.0, + "step": 12099 + }, + { + "epoch": 1.328794201625302, + "grad_norm": 2.5619518756866455, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7186528444290161, + "num_tokens": 309996535.0, + "step": 12100 + }, + { + "epoch": 1.3289040193279158, + "grad_norm": 2.2754805088043213, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7122681140899658, + "num_tokens": 310021577.0, + "step": 12101 + }, + { + "epoch": 1.3290138370305293, + "grad_norm": 2.2220470905303955, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7147469520568848, + "num_tokens": 310046560.0, + "step": 12102 + }, + { + "epoch": 1.3291236547331429, + "grad_norm": 2.0225934982299805, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7112964391708374, + "num_tokens": 310076926.0, + "step": 12103 + }, + { + "epoch": 1.3292334724357566, + "grad_norm": 2.506772518157959, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7075717449188232, + "num_tokens": 310099524.0, + "step": 12104 + }, + { + "epoch": 1.3293432901383704, + "grad_norm": 2.200134754180908, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7252580523490906, + "num_tokens": 310126753.0, + "step": 12105 + }, + { + "epoch": 1.329453107840984, + "grad_norm": 2.3165283203125, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7178529500961304, + "num_tokens": 310151072.0, + "step": 12106 + }, + { + "epoch": 1.3295629255435977, + "grad_norm": 2.050218105316162, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7177531719207764, + "num_tokens": 310180589.0, + "step": 12107 + }, + { + "epoch": 1.3296727432462112, + "grad_norm": 2.3973171710968018, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7184906601905823, + "num_tokens": 310204501.0, + "step": 12108 + }, + { + "epoch": 1.329782560948825, + "grad_norm": 2.157406806945801, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7144870758056641, + "num_tokens": 310232045.0, + "step": 12109 + }, + { + "epoch": 1.3298923786514387, + "grad_norm": 2.6157031059265137, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7285645008087158, + "num_tokens": 310251158.0, + "step": 12110 + }, + { + "epoch": 1.3300021963540523, + "grad_norm": 2.520418405532837, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7229306101799011, + "num_tokens": 310271940.0, + "step": 12111 + }, + { + "epoch": 1.330112014056666, + "grad_norm": 2.350325584411621, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7256568074226379, + "num_tokens": 310295220.0, + "step": 12112 + }, + { + "epoch": 1.3302218317592795, + "grad_norm": 2.3856091499328613, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7125129699707031, + "num_tokens": 310319875.0, + "step": 12113 + }, + { + "epoch": 1.3303316494618933, + "grad_norm": 2.549635648727417, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7014890909194946, + "num_tokens": 310342949.0, + "step": 12114 + }, + { + "epoch": 1.3304414671645068, + "grad_norm": 2.358694076538086, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7251837849617004, + "num_tokens": 310367004.0, + "step": 12115 + }, + { + "epoch": 1.3305512848671206, + "grad_norm": 2.374851703643799, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7057805061340332, + "num_tokens": 310392758.0, + "step": 12116 + }, + { + "epoch": 1.3306611025697341, + "grad_norm": 2.257699489593506, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7028229832649231, + "num_tokens": 310417807.0, + "step": 12117 + }, + { + "epoch": 1.330770920272348, + "grad_norm": 2.1775293350219727, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7157626748085022, + "num_tokens": 310445267.0, + "step": 12118 + }, + { + "epoch": 1.3308807379749616, + "grad_norm": 2.565032482147217, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7353057265281677, + "num_tokens": 310465200.0, + "step": 12119 + }, + { + "epoch": 1.3309905556775752, + "grad_norm": 2.3644626140594482, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6963592171669006, + "num_tokens": 310490762.0, + "step": 12120 + }, + { + "epoch": 1.331100373380189, + "grad_norm": 2.0387916564941406, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7084258198738098, + "num_tokens": 310520901.0, + "step": 12121 + }, + { + "epoch": 1.3312101910828025, + "grad_norm": 2.363852024078369, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7197427749633789, + "num_tokens": 310543711.0, + "step": 12122 + }, + { + "epoch": 1.3313200087854162, + "grad_norm": 2.1154353618621826, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7339856028556824, + "num_tokens": 310571161.0, + "step": 12123 + }, + { + "epoch": 1.33142982648803, + "grad_norm": 2.0602827072143555, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7164334058761597, + "num_tokens": 310599278.0, + "step": 12124 + }, + { + "epoch": 1.3315396441906435, + "grad_norm": 2.2438247203826904, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7013421058654785, + "num_tokens": 310628311.0, + "step": 12125 + }, + { + "epoch": 1.331649461893257, + "grad_norm": 2.4983105659484863, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7217850685119629, + "num_tokens": 310649825.0, + "step": 12126 + }, + { + "epoch": 1.3317592795958708, + "grad_norm": 1.8400224447250366, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7061644792556763, + "num_tokens": 310682914.0, + "step": 12127 + }, + { + "epoch": 1.3318690972984846, + "grad_norm": 1.9724414348602295, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7216540575027466, + "num_tokens": 310714361.0, + "step": 12128 + }, + { + "epoch": 1.331978915001098, + "grad_norm": 2.1414053440093994, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7133094668388367, + "num_tokens": 310741756.0, + "step": 12129 + }, + { + "epoch": 1.3320887327037119, + "grad_norm": 2.241107702255249, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7149150371551514, + "num_tokens": 310768927.0, + "step": 12130 + }, + { + "epoch": 1.3321985504063254, + "grad_norm": 2.1506543159484863, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7156413197517395, + "num_tokens": 310797973.0, + "step": 12131 + }, + { + "epoch": 1.3323083681089392, + "grad_norm": 1.9649845361709595, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7107862830162048, + "num_tokens": 310831749.0, + "step": 12132 + }, + { + "epoch": 1.332418185811553, + "grad_norm": 2.153212785720825, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7080637812614441, + "num_tokens": 310860122.0, + "step": 12133 + }, + { + "epoch": 1.3325280035141664, + "grad_norm": 2.051710367202759, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7048215866088867, + "num_tokens": 310891094.0, + "step": 12134 + }, + { + "epoch": 1.3326378212167802, + "grad_norm": 2.3534674644470215, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6924401521682739, + "num_tokens": 310916345.0, + "step": 12135 + }, + { + "epoch": 1.3327476389193937, + "grad_norm": 2.0983059406280518, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7079900503158569, + "num_tokens": 310946830.0, + "step": 12136 + }, + { + "epoch": 1.3328574566220075, + "grad_norm": 2.159247636795044, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7135360240936279, + "num_tokens": 310975464.0, + "step": 12137 + }, + { + "epoch": 1.3329672743246213, + "grad_norm": 2.3930459022521973, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7328722476959229, + "num_tokens": 310998027.0, + "step": 12138 + }, + { + "epoch": 1.3330770920272348, + "grad_norm": 2.4506402015686035, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.722716748714447, + "num_tokens": 311020408.0, + "step": 12139 + }, + { + "epoch": 1.3331869097298483, + "grad_norm": 1.994588017463684, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7381975650787354, + "num_tokens": 311051134.0, + "step": 12140 + }, + { + "epoch": 1.333296727432462, + "grad_norm": 2.0119447708129883, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7108117341995239, + "num_tokens": 311080737.0, + "step": 12141 + }, + { + "epoch": 1.3334065451350758, + "grad_norm": 2.1075987815856934, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7261164784431458, + "num_tokens": 311108515.0, + "step": 12142 + }, + { + "epoch": 1.3335163628376894, + "grad_norm": 2.4280338287353516, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7037439346313477, + "num_tokens": 311130975.0, + "step": 12143 + }, + { + "epoch": 1.3336261805403031, + "grad_norm": 2.184535026550293, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7462371587753296, + "num_tokens": 311155665.0, + "step": 12144 + }, + { + "epoch": 1.3337359982429167, + "grad_norm": 2.1619582176208496, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7220394015312195, + "num_tokens": 311183191.0, + "step": 12145 + }, + { + "epoch": 1.3338458159455304, + "grad_norm": 2.22790265083313, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7042077779769897, + "num_tokens": 311210487.0, + "step": 12146 + }, + { + "epoch": 1.3339556336481442, + "grad_norm": 2.1377155780792236, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7117249965667725, + "num_tokens": 311237682.0, + "step": 12147 + }, + { + "epoch": 1.3340654513507577, + "grad_norm": 2.4419913291931152, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.736607551574707, + "num_tokens": 311260161.0, + "step": 12148 + }, + { + "epoch": 1.3341752690533715, + "grad_norm": 2.0805795192718506, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6899142861366272, + "num_tokens": 311288272.0, + "step": 12149 + }, + { + "epoch": 1.334285086755985, + "grad_norm": 2.10274600982666, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6921037435531616, + "num_tokens": 311317046.0, + "step": 12150 + }, + { + "epoch": 1.3343949044585988, + "grad_norm": 2.2525720596313477, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7261518239974976, + "num_tokens": 311341920.0, + "step": 12151 + }, + { + "epoch": 1.3345047221612125, + "grad_norm": 2.071399688720703, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.7072184085845947, + "num_tokens": 311369697.0, + "step": 12152 + }, + { + "epoch": 1.334614539863826, + "grad_norm": 2.544584035873413, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7412317395210266, + "num_tokens": 311389922.0, + "step": 12153 + }, + { + "epoch": 1.3347243575664396, + "grad_norm": 2.3877065181732178, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7173482179641724, + "num_tokens": 311414953.0, + "step": 12154 + }, + { + "epoch": 1.3348341752690533, + "grad_norm": 2.4710824489593506, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.726981520652771, + "num_tokens": 311436808.0, + "step": 12155 + }, + { + "epoch": 1.334943992971667, + "grad_norm": 2.5875351428985596, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7119906544685364, + "num_tokens": 311457531.0, + "step": 12156 + }, + { + "epoch": 1.3350538106742806, + "grad_norm": 2.167314291000366, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7364218235015869, + "num_tokens": 311483198.0, + "step": 12157 + }, + { + "epoch": 1.3351636283768944, + "grad_norm": 2.3552308082580566, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7302059531211853, + "num_tokens": 311504963.0, + "step": 12158 + }, + { + "epoch": 1.335273446079508, + "grad_norm": 2.2709059715270996, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6920618414878845, + "num_tokens": 311530007.0, + "step": 12159 + }, + { + "epoch": 1.3353832637821217, + "grad_norm": 2.4789154529571533, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7050751447677612, + "num_tokens": 311551857.0, + "step": 12160 + }, + { + "epoch": 1.3354930814847354, + "grad_norm": 2.2903544902801514, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7195373773574829, + "num_tokens": 311577125.0, + "step": 12161 + }, + { + "epoch": 1.335602899187349, + "grad_norm": 2.1510868072509766, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7093050479888916, + "num_tokens": 311606898.0, + "step": 12162 + }, + { + "epoch": 1.3357127168899627, + "grad_norm": 2.24924898147583, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7177540063858032, + "num_tokens": 311633126.0, + "step": 12163 + }, + { + "epoch": 1.3358225345925763, + "grad_norm": 2.2359588146209717, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7124772667884827, + "num_tokens": 311659523.0, + "step": 12164 + }, + { + "epoch": 1.33593235229519, + "grad_norm": 2.242636203765869, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7168351411819458, + "num_tokens": 311685318.0, + "step": 12165 + }, + { + "epoch": 1.3360421699978036, + "grad_norm": 1.9191758632659912, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7171177864074707, + "num_tokens": 311719241.0, + "step": 12166 + }, + { + "epoch": 1.3361519877004173, + "grad_norm": 1.9499540328979492, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.696295440196991, + "num_tokens": 311751009.0, + "step": 12167 + }, + { + "epoch": 1.3362618054030309, + "grad_norm": 2.025935649871826, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7126432061195374, + "num_tokens": 311782153.0, + "step": 12168 + }, + { + "epoch": 1.3363716231056446, + "grad_norm": 2.580739736557007, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7279084324836731, + "num_tokens": 311803525.0, + "step": 12169 + }, + { + "epoch": 1.3364814408082584, + "grad_norm": 2.0437660217285156, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6912748217582703, + "num_tokens": 311835839.0, + "step": 12170 + }, + { + "epoch": 1.336591258510872, + "grad_norm": 1.9929447174072266, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6971474289894104, + "num_tokens": 311867831.0, + "step": 12171 + }, + { + "epoch": 1.3367010762134857, + "grad_norm": 2.6711316108703613, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7274467945098877, + "num_tokens": 311888220.0, + "step": 12172 + }, + { + "epoch": 1.3368108939160992, + "grad_norm": 1.9642187356948853, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6984713077545166, + "num_tokens": 311920997.0, + "step": 12173 + }, + { + "epoch": 1.336920711618713, + "grad_norm": 2.080673933029175, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7125394344329834, + "num_tokens": 311950166.0, + "step": 12174 + }, + { + "epoch": 1.3370305293213267, + "grad_norm": 2.6255786418914795, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7326030731201172, + "num_tokens": 311968885.0, + "step": 12175 + }, + { + "epoch": 1.3371403470239402, + "grad_norm": 2.2657957077026367, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7250704765319824, + "num_tokens": 311992771.0, + "step": 12176 + }, + { + "epoch": 1.337250164726554, + "grad_norm": 2.0093414783477783, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7368631362915039, + "num_tokens": 312018706.0, + "step": 12177 + }, + { + "epoch": 1.3373599824291675, + "grad_norm": 2.3319172859191895, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7286087274551392, + "num_tokens": 312043277.0, + "step": 12178 + }, + { + "epoch": 1.3374698001317813, + "grad_norm": 2.2342896461486816, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7131881713867188, + "num_tokens": 312067370.0, + "step": 12179 + }, + { + "epoch": 1.3375796178343948, + "grad_norm": 1.8935514688491821, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7062168121337891, + "num_tokens": 312101209.0, + "step": 12180 + }, + { + "epoch": 1.3376894355370086, + "grad_norm": 2.075021266937256, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7337157726287842, + "num_tokens": 312128806.0, + "step": 12181 + }, + { + "epoch": 1.3377992532396221, + "grad_norm": 2.3088538646698, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7075099945068359, + "num_tokens": 312152915.0, + "step": 12182 + }, + { + "epoch": 1.3379090709422359, + "grad_norm": 2.4457485675811768, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7332133054733276, + "num_tokens": 312175156.0, + "step": 12183 + }, + { + "epoch": 1.3380188886448496, + "grad_norm": 2.2543041706085205, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7149578332901001, + "num_tokens": 312198286.0, + "step": 12184 + }, + { + "epoch": 1.3381287063474632, + "grad_norm": 2.130362033843994, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7078749537467957, + "num_tokens": 312226295.0, + "step": 12185 + }, + { + "epoch": 1.338238524050077, + "grad_norm": 2.520615816116333, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7198030948638916, + "num_tokens": 312247186.0, + "step": 12186 + }, + { + "epoch": 1.3383483417526905, + "grad_norm": 2.1753621101379395, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7250690460205078, + "num_tokens": 312274543.0, + "step": 12187 + }, + { + "epoch": 1.3384581594553042, + "grad_norm": 2.045010566711426, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7221989631652832, + "num_tokens": 312301554.0, + "step": 12188 + }, + { + "epoch": 1.338567977157918, + "grad_norm": 2.218616485595703, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7547710537910461, + "num_tokens": 312326636.0, + "step": 12189 + }, + { + "epoch": 1.3386777948605315, + "grad_norm": 2.23622989654541, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.718721330165863, + "num_tokens": 312352382.0, + "step": 12190 + }, + { + "epoch": 1.338787612563145, + "grad_norm": 2.4939093589782715, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7214270234107971, + "num_tokens": 312373111.0, + "step": 12191 + }, + { + "epoch": 1.3388974302657588, + "grad_norm": 2.1921226978302, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7211395502090454, + "num_tokens": 312399736.0, + "step": 12192 + }, + { + "epoch": 1.3390072479683726, + "grad_norm": 1.9417576789855957, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7268064022064209, + "num_tokens": 312434381.0, + "step": 12193 + }, + { + "epoch": 1.339117065670986, + "grad_norm": 2.3012874126434326, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7282192707061768, + "num_tokens": 312458461.0, + "step": 12194 + }, + { + "epoch": 1.3392268833735999, + "grad_norm": 2.239147663116455, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.72294020652771, + "num_tokens": 312484266.0, + "step": 12195 + }, + { + "epoch": 1.3393367010762134, + "grad_norm": 2.3811707496643066, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.70626300573349, + "num_tokens": 312507964.0, + "step": 12196 + }, + { + "epoch": 1.3394465187788271, + "grad_norm": 2.3095364570617676, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.732323169708252, + "num_tokens": 312533284.0, + "step": 12197 + }, + { + "epoch": 1.339556336481441, + "grad_norm": 2.5489606857299805, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.731423556804657, + "num_tokens": 312553148.0, + "step": 12198 + }, + { + "epoch": 1.3396661541840544, + "grad_norm": 1.9444594383239746, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6869894862174988, + "num_tokens": 312586212.0, + "step": 12199 + }, + { + "epoch": 1.3397759718866682, + "grad_norm": 2.210601329803467, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7155089378356934, + "num_tokens": 312612637.0, + "step": 12200 + }, + { + "epoch": 1.3398857895892817, + "grad_norm": 2.1182873249053955, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7154194712638855, + "num_tokens": 312641017.0, + "step": 12201 + }, + { + "epoch": 1.3399956072918955, + "grad_norm": 2.0138602256774902, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7079528570175171, + "num_tokens": 312673177.0, + "step": 12202 + }, + { + "epoch": 1.3401054249945092, + "grad_norm": 2.6017487049102783, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7538224458694458, + "num_tokens": 312692911.0, + "step": 12203 + }, + { + "epoch": 1.3402152426971228, + "grad_norm": 2.2907207012176514, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7183055877685547, + "num_tokens": 312718271.0, + "step": 12204 + }, + { + "epoch": 1.3403250603997363, + "grad_norm": 2.285191059112549, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.714202880859375, + "num_tokens": 312743189.0, + "step": 12205 + }, + { + "epoch": 1.34043487810235, + "grad_norm": 2.334712266921997, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.719119668006897, + "num_tokens": 312766558.0, + "step": 12206 + }, + { + "epoch": 1.3405446958049638, + "grad_norm": 2.252734661102295, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7072286605834961, + "num_tokens": 312792884.0, + "step": 12207 + }, + { + "epoch": 1.3406545135075774, + "grad_norm": 2.244889974594116, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7427518367767334, + "num_tokens": 312819925.0, + "step": 12208 + }, + { + "epoch": 1.3407643312101911, + "grad_norm": 2.4399633407592773, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7494763135910034, + "num_tokens": 312842506.0, + "step": 12209 + }, + { + "epoch": 1.3408741489128047, + "grad_norm": 2.436446189880371, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7235384583473206, + "num_tokens": 312864912.0, + "step": 12210 + }, + { + "epoch": 1.3409839666154184, + "grad_norm": 2.0510740280151367, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7164615988731384, + "num_tokens": 312892442.0, + "step": 12211 + }, + { + "epoch": 1.3410937843180322, + "grad_norm": 2.398601531982422, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7311630249023438, + "num_tokens": 312915644.0, + "step": 12212 + }, + { + "epoch": 1.3412036020206457, + "grad_norm": 2.332462787628174, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7358975410461426, + "num_tokens": 312938651.0, + "step": 12213 + }, + { + "epoch": 1.3413134197232595, + "grad_norm": 2.2982873916625977, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7461932897567749, + "num_tokens": 312962843.0, + "step": 12214 + }, + { + "epoch": 1.341423237425873, + "grad_norm": 1.9724693298339844, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6973755359649658, + "num_tokens": 312993115.0, + "step": 12215 + }, + { + "epoch": 1.3415330551284868, + "grad_norm": 2.236743211746216, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7171694040298462, + "num_tokens": 313017910.0, + "step": 12216 + }, + { + "epoch": 1.3416428728311005, + "grad_norm": 2.103436231613159, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7046440839767456, + "num_tokens": 313046855.0, + "step": 12217 + }, + { + "epoch": 1.341752690533714, + "grad_norm": 2.8016443252563477, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7217294573783875, + "num_tokens": 313064210.0, + "step": 12218 + }, + { + "epoch": 1.3418625082363276, + "grad_norm": 2.1198365688323975, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7232837677001953, + "num_tokens": 313091761.0, + "step": 12219 + }, + { + "epoch": 1.3419723259389413, + "grad_norm": 2.383831262588501, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.708766758441925, + "num_tokens": 313114913.0, + "step": 12220 + }, + { + "epoch": 1.342082143641555, + "grad_norm": 2.3915774822235107, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7167278528213501, + "num_tokens": 313138341.0, + "step": 12221 + }, + { + "epoch": 1.3421919613441686, + "grad_norm": 1.9494292736053467, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.691679060459137, + "num_tokens": 313171151.0, + "step": 12222 + }, + { + "epoch": 1.3423017790467824, + "grad_norm": 2.254601001739502, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.742737352848053, + "num_tokens": 313195408.0, + "step": 12223 + }, + { + "epoch": 1.342411596749396, + "grad_norm": 2.1333820819854736, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7305204272270203, + "num_tokens": 313220746.0, + "step": 12224 + }, + { + "epoch": 1.3425214144520097, + "grad_norm": 2.200413227081299, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.739958643913269, + "num_tokens": 313246324.0, + "step": 12225 + }, + { + "epoch": 1.3426312321546234, + "grad_norm": 2.4277093410491943, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7198886871337891, + "num_tokens": 313271001.0, + "step": 12226 + }, + { + "epoch": 1.342741049857237, + "grad_norm": 2.29414963722229, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7175464630126953, + "num_tokens": 313296098.0, + "step": 12227 + }, + { + "epoch": 1.3428508675598507, + "grad_norm": 2.5104286670684814, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7105557918548584, + "num_tokens": 313317539.0, + "step": 12228 + }, + { + "epoch": 1.3429606852624643, + "grad_norm": 2.9913220405578613, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7224199175834656, + "num_tokens": 313346779.0, + "step": 12229 + }, + { + "epoch": 1.343070502965078, + "grad_norm": 2.3207192420959473, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7235580682754517, + "num_tokens": 313369957.0, + "step": 12230 + }, + { + "epoch": 1.3431803206676916, + "grad_norm": 2.051642417907715, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7224515676498413, + "num_tokens": 313399568.0, + "step": 12231 + }, + { + "epoch": 1.3432901383703053, + "grad_norm": 2.416604995727539, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7319057583808899, + "num_tokens": 313421795.0, + "step": 12232 + }, + { + "epoch": 1.3433999560729188, + "grad_norm": 2.1151411533355713, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7109743356704712, + "num_tokens": 313451823.0, + "step": 12233 + }, + { + "epoch": 1.3435097737755326, + "grad_norm": 2.084812879562378, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7006890773773193, + "num_tokens": 313480125.0, + "step": 12234 + }, + { + "epoch": 1.3436195914781464, + "grad_norm": 1.95863938331604, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7128251194953918, + "num_tokens": 313510564.0, + "step": 12235 + }, + { + "epoch": 1.34372940918076, + "grad_norm": 2.2121896743774414, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7119941711425781, + "num_tokens": 313536905.0, + "step": 12236 + }, + { + "epoch": 1.3438392268833736, + "grad_norm": 2.3557472229003906, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7290151119232178, + "num_tokens": 313560198.0, + "step": 12237 + }, + { + "epoch": 1.3439490445859872, + "grad_norm": 2.2665536403656006, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6972370147705078, + "num_tokens": 313587856.0, + "step": 12238 + }, + { + "epoch": 1.344058862288601, + "grad_norm": 1.9470341205596924, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7107680439949036, + "num_tokens": 313621747.0, + "step": 12239 + }, + { + "epoch": 1.3441686799912147, + "grad_norm": 2.1903085708618164, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.717883825302124, + "num_tokens": 313649587.0, + "step": 12240 + }, + { + "epoch": 1.3442784976938282, + "grad_norm": 2.1628737449645996, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.691706120967865, + "num_tokens": 313677310.0, + "step": 12241 + }, + { + "epoch": 1.3443883153964418, + "grad_norm": 2.262007474899292, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.710193395614624, + "num_tokens": 313701901.0, + "step": 12242 + }, + { + "epoch": 1.3444981330990555, + "grad_norm": 2.5445621013641357, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7515028715133667, + "num_tokens": 313721800.0, + "step": 12243 + }, + { + "epoch": 1.3446079508016693, + "grad_norm": 1.974241018295288, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7012329697608948, + "num_tokens": 313754771.0, + "step": 12244 + }, + { + "epoch": 1.3447177685042828, + "grad_norm": 2.219550132751465, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7156376838684082, + "num_tokens": 313781346.0, + "step": 12245 + }, + { + "epoch": 1.3448275862068966, + "grad_norm": 2.2869200706481934, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7349095940589905, + "num_tokens": 313803846.0, + "step": 12246 + }, + { + "epoch": 1.34493740390951, + "grad_norm": 2.249910831451416, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7111769318580627, + "num_tokens": 313829990.0, + "step": 12247 + }, + { + "epoch": 1.3450472216121239, + "grad_norm": 2.360585927963257, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.6995413899421692, + "num_tokens": 313855715.0, + "step": 12248 + }, + { + "epoch": 1.3451570393147376, + "grad_norm": 2.1180648803710938, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6874966025352478, + "num_tokens": 313886381.0, + "step": 12249 + }, + { + "epoch": 1.3452668570173512, + "grad_norm": 2.1667895317077637, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7038334608078003, + "num_tokens": 313917336.0, + "step": 12250 + }, + { + "epoch": 1.345376674719965, + "grad_norm": 2.615166664123535, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7376127243041992, + "num_tokens": 313938105.0, + "step": 12251 + }, + { + "epoch": 1.3454864924225785, + "grad_norm": 2.372042179107666, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.71857088804245, + "num_tokens": 313962926.0, + "step": 12252 + }, + { + "epoch": 1.3455963101251922, + "grad_norm": 2.176271677017212, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.723118782043457, + "num_tokens": 313989594.0, + "step": 12253 + }, + { + "epoch": 1.345706127827806, + "grad_norm": 2.3879153728485107, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7133184671401978, + "num_tokens": 314011772.0, + "step": 12254 + }, + { + "epoch": 1.3458159455304195, + "grad_norm": 2.4353466033935547, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7438790798187256, + "num_tokens": 314032710.0, + "step": 12255 + }, + { + "epoch": 1.345925763233033, + "grad_norm": 2.3619425296783447, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.730815589427948, + "num_tokens": 314055959.0, + "step": 12256 + }, + { + "epoch": 1.3460355809356468, + "grad_norm": 2.323598623275757, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7164368629455566, + "num_tokens": 314079472.0, + "step": 12257 + }, + { + "epoch": 1.3461453986382605, + "grad_norm": 2.2147796154022217, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7151405811309814, + "num_tokens": 314108124.0, + "step": 12258 + }, + { + "epoch": 1.346255216340874, + "grad_norm": 2.1729300022125244, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7251507043838501, + "num_tokens": 314134284.0, + "step": 12259 + }, + { + "epoch": 1.3463650340434878, + "grad_norm": 2.3409295082092285, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7375626564025879, + "num_tokens": 314158092.0, + "step": 12260 + }, + { + "epoch": 1.3464748517461014, + "grad_norm": 2.5953359603881836, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7272695302963257, + "num_tokens": 314178896.0, + "step": 12261 + }, + { + "epoch": 1.3465846694487151, + "grad_norm": 2.2582762241363525, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7215445041656494, + "num_tokens": 314204620.0, + "step": 12262 + }, + { + "epoch": 1.346694487151329, + "grad_norm": 2.0544846057891846, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7071475982666016, + "num_tokens": 314234660.0, + "step": 12263 + }, + { + "epoch": 1.3468043048539424, + "grad_norm": 2.253150701522827, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7025479078292847, + "num_tokens": 314262785.0, + "step": 12264 + }, + { + "epoch": 1.3469141225565562, + "grad_norm": 2.1296768188476562, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7262060642242432, + "num_tokens": 314291181.0, + "step": 12265 + }, + { + "epoch": 1.3470239402591697, + "grad_norm": 2.3149216175079346, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7174026370048523, + "num_tokens": 314314472.0, + "step": 12266 + }, + { + "epoch": 1.3471337579617835, + "grad_norm": 2.0511789321899414, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6941652297973633, + "num_tokens": 314344086.0, + "step": 12267 + }, + { + "epoch": 1.3472435756643972, + "grad_norm": 2.131345748901367, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6874916553497314, + "num_tokens": 314373327.0, + "step": 12268 + }, + { + "epoch": 1.3473533933670108, + "grad_norm": 2.525449514389038, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6970745921134949, + "num_tokens": 314393931.0, + "step": 12269 + }, + { + "epoch": 1.3474632110696243, + "grad_norm": 2.2155447006225586, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7016068696975708, + "num_tokens": 314420149.0, + "step": 12270 + }, + { + "epoch": 1.347573028772238, + "grad_norm": 2.0350852012634277, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7351223230361938, + "num_tokens": 314452108.0, + "step": 12271 + }, + { + "epoch": 1.3476828464748518, + "grad_norm": 2.330321788787842, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.719914436340332, + "num_tokens": 314477721.0, + "step": 12272 + }, + { + "epoch": 1.3477926641774653, + "grad_norm": 2.2518157958984375, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.6991525888442993, + "num_tokens": 314503493.0, + "step": 12273 + }, + { + "epoch": 1.347902481880079, + "grad_norm": 2.342255115509033, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7032264471054077, + "num_tokens": 314527536.0, + "step": 12274 + }, + { + "epoch": 1.3480122995826926, + "grad_norm": 2.083890676498413, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7094283103942871, + "num_tokens": 314557805.0, + "step": 12275 + }, + { + "epoch": 1.3481221172853064, + "grad_norm": 2.001075506210327, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7285219430923462, + "num_tokens": 314586749.0, + "step": 12276 + }, + { + "epoch": 1.3482319349879202, + "grad_norm": 2.2820773124694824, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7211087942123413, + "num_tokens": 314611292.0, + "step": 12277 + }, + { + "epoch": 1.3483417526905337, + "grad_norm": 2.1629199981689453, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7026101350784302, + "num_tokens": 314641205.0, + "step": 12278 + }, + { + "epoch": 1.3484515703931474, + "grad_norm": 1.9435433149337769, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6924961805343628, + "num_tokens": 314677440.0, + "step": 12279 + }, + { + "epoch": 1.348561388095761, + "grad_norm": 2.059699058532715, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6978485584259033, + "num_tokens": 314708278.0, + "step": 12280 + }, + { + "epoch": 1.3486712057983747, + "grad_norm": 2.4867970943450928, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7279397249221802, + "num_tokens": 314728941.0, + "step": 12281 + }, + { + "epoch": 1.3487810235009885, + "grad_norm": 2.1839194297790527, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7192437648773193, + "num_tokens": 314754378.0, + "step": 12282 + }, + { + "epoch": 1.348890841203602, + "grad_norm": 2.045870065689087, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7369040250778198, + "num_tokens": 314783316.0, + "step": 12283 + }, + { + "epoch": 1.3490006589062156, + "grad_norm": 2.3000173568725586, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7214858531951904, + "num_tokens": 314806529.0, + "step": 12284 + }, + { + "epoch": 1.3491104766088293, + "grad_norm": 2.096633195877075, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7034104466438293, + "num_tokens": 314834802.0, + "step": 12285 + }, + { + "epoch": 1.349220294311443, + "grad_norm": 2.3427510261535645, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.704820990562439, + "num_tokens": 314859882.0, + "step": 12286 + }, + { + "epoch": 1.3493301120140566, + "grad_norm": 2.388866424560547, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7368300557136536, + "num_tokens": 314882019.0, + "step": 12287 + }, + { + "epoch": 1.3494399297166704, + "grad_norm": 2.4644994735717773, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7398950457572937, + "num_tokens": 314902976.0, + "step": 12288 + }, + { + "epoch": 1.349549747419284, + "grad_norm": 2.139033555984497, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7371370792388916, + "num_tokens": 314928013.0, + "step": 12289 + }, + { + "epoch": 1.3496595651218977, + "grad_norm": 2.184143304824829, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7085432410240173, + "num_tokens": 314953877.0, + "step": 12290 + }, + { + "epoch": 1.3497693828245114, + "grad_norm": 2.327955722808838, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7216236591339111, + "num_tokens": 314977647.0, + "step": 12291 + }, + { + "epoch": 1.349879200527125, + "grad_norm": 2.0810391902923584, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7227431535720825, + "num_tokens": 315005721.0, + "step": 12292 + }, + { + "epoch": 1.3499890182297387, + "grad_norm": 2.394355058670044, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7110939025878906, + "num_tokens": 315029414.0, + "step": 12293 + }, + { + "epoch": 1.3500988359323522, + "grad_norm": 2.2984278202056885, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7177256941795349, + "num_tokens": 315053509.0, + "step": 12294 + }, + { + "epoch": 1.350208653634966, + "grad_norm": 2.3902993202209473, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.718482255935669, + "num_tokens": 315075481.0, + "step": 12295 + }, + { + "epoch": 1.3503184713375795, + "grad_norm": 2.2756497859954834, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7311108112335205, + "num_tokens": 315102131.0, + "step": 12296 + }, + { + "epoch": 1.3504282890401933, + "grad_norm": 2.2523889541625977, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7195752859115601, + "num_tokens": 315130172.0, + "step": 12297 + }, + { + "epoch": 1.3505381067428068, + "grad_norm": 1.9670113325119019, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7224304676055908, + "num_tokens": 315162758.0, + "step": 12298 + }, + { + "epoch": 1.3506479244454206, + "grad_norm": 2.111910104751587, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7050290107727051, + "num_tokens": 315189640.0, + "step": 12299 + }, + { + "epoch": 1.3507577421480343, + "grad_norm": 2.308816432952881, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.733221173286438, + "num_tokens": 315213729.0, + "step": 12300 + }, + { + "epoch": 1.3508675598506479, + "grad_norm": 2.124372720718384, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6919432878494263, + "num_tokens": 315242319.0, + "step": 12301 + }, + { + "epoch": 1.3509773775532616, + "grad_norm": 2.5463626384735107, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7238807678222656, + "num_tokens": 315261910.0, + "step": 12302 + }, + { + "epoch": 1.3510871952558752, + "grad_norm": 2.17305850982666, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7182326316833496, + "num_tokens": 315289109.0, + "step": 12303 + }, + { + "epoch": 1.351197012958489, + "grad_norm": 2.330909013748169, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7093848586082458, + "num_tokens": 315314026.0, + "step": 12304 + }, + { + "epoch": 1.3513068306611027, + "grad_norm": 2.37164044380188, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.6995320320129395, + "num_tokens": 315337790.0, + "step": 12305 + }, + { + "epoch": 1.3514166483637162, + "grad_norm": 2.8049509525299072, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7521098852157593, + "num_tokens": 315354699.0, + "step": 12306 + }, + { + "epoch": 1.3515264660663298, + "grad_norm": 2.15301513671875, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6879130601882935, + "num_tokens": 315381448.0, + "step": 12307 + }, + { + "epoch": 1.3516362837689435, + "grad_norm": 2.5190012454986572, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7331247329711914, + "num_tokens": 315401621.0, + "step": 12308 + }, + { + "epoch": 1.3517461014715573, + "grad_norm": 2.1928794384002686, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.713566780090332, + "num_tokens": 315429055.0, + "step": 12309 + }, + { + "epoch": 1.3518559191741708, + "grad_norm": 2.438777446746826, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7097591161727905, + "num_tokens": 315451833.0, + "step": 12310 + }, + { + "epoch": 1.3519657368767846, + "grad_norm": 2.322610378265381, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7247411608695984, + "num_tokens": 315476478.0, + "step": 12311 + }, + { + "epoch": 1.352075554579398, + "grad_norm": 2.2930667400360107, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7236610054969788, + "num_tokens": 315500234.0, + "step": 12312 + }, + { + "epoch": 1.3521853722820119, + "grad_norm": 2.2204668521881104, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7176826000213623, + "num_tokens": 315525535.0, + "step": 12313 + }, + { + "epoch": 1.3522951899846256, + "grad_norm": 2.0599400997161865, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7290362119674683, + "num_tokens": 315553932.0, + "step": 12314 + }, + { + "epoch": 1.3524050076872391, + "grad_norm": 2.393486738204956, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6967482566833496, + "num_tokens": 315580595.0, + "step": 12315 + }, + { + "epoch": 1.352514825389853, + "grad_norm": 2.268890142440796, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7359826564788818, + "num_tokens": 315608201.0, + "step": 12316 + }, + { + "epoch": 1.3526246430924664, + "grad_norm": 2.22503924369812, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7217658162117004, + "num_tokens": 315634243.0, + "step": 12317 + }, + { + "epoch": 1.3527344607950802, + "grad_norm": 2.217150926589966, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7137420177459717, + "num_tokens": 315662038.0, + "step": 12318 + }, + { + "epoch": 1.352844278497694, + "grad_norm": 2.1937668323516846, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7124263048171997, + "num_tokens": 315687534.0, + "step": 12319 + }, + { + "epoch": 1.3529540962003075, + "grad_norm": 2.1754226684570312, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7063043713569641, + "num_tokens": 315715308.0, + "step": 12320 + }, + { + "epoch": 1.353063913902921, + "grad_norm": 2.3035354614257812, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7238826751708984, + "num_tokens": 315739233.0, + "step": 12321 + }, + { + "epoch": 1.3531737316055348, + "grad_norm": 2.0685808658599854, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7221430540084839, + "num_tokens": 315769342.0, + "step": 12322 + }, + { + "epoch": 1.3532835493081485, + "grad_norm": 2.3736178874969482, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7382689714431763, + "num_tokens": 315792127.0, + "step": 12323 + }, + { + "epoch": 1.353393367010762, + "grad_norm": 2.3235456943511963, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7315866947174072, + "num_tokens": 315815081.0, + "step": 12324 + }, + { + "epoch": 1.3535031847133758, + "grad_norm": 2.199737548828125, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7105039358139038, + "num_tokens": 315840188.0, + "step": 12325 + }, + { + "epoch": 1.3536130024159894, + "grad_norm": 2.497220993041992, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7229201793670654, + "num_tokens": 315863260.0, + "step": 12326 + }, + { + "epoch": 1.3537228201186031, + "grad_norm": 2.160006046295166, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7236818075180054, + "num_tokens": 315889361.0, + "step": 12327 + }, + { + "epoch": 1.3538326378212169, + "grad_norm": 2.298491954803467, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7154903411865234, + "num_tokens": 315913586.0, + "step": 12328 + }, + { + "epoch": 1.3539424555238304, + "grad_norm": 2.2801477909088135, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7210782170295715, + "num_tokens": 315939725.0, + "step": 12329 + }, + { + "epoch": 1.3540522732264442, + "grad_norm": 2.1960690021514893, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7283912301063538, + "num_tokens": 315965250.0, + "step": 12330 + }, + { + "epoch": 1.3541620909290577, + "grad_norm": 2.162731409072876, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7143049836158752, + "num_tokens": 315993355.0, + "step": 12331 + }, + { + "epoch": 1.3542719086316715, + "grad_norm": 2.3884997367858887, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7181006669998169, + "num_tokens": 316017372.0, + "step": 12332 + }, + { + "epoch": 1.3543817263342852, + "grad_norm": 2.460155963897705, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.741264283657074, + "num_tokens": 316040115.0, + "step": 12333 + }, + { + "epoch": 1.3544915440368988, + "grad_norm": 2.202505111694336, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7183667421340942, + "num_tokens": 316068753.0, + "step": 12334 + }, + { + "epoch": 1.3546013617395123, + "grad_norm": 2.0488784313201904, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7002949118614197, + "num_tokens": 316099513.0, + "step": 12335 + }, + { + "epoch": 1.354711179442126, + "grad_norm": 2.7507638931274414, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7359998226165771, + "num_tokens": 316117393.0, + "step": 12336 + }, + { + "epoch": 1.3548209971447398, + "grad_norm": 2.309046506881714, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7091822028160095, + "num_tokens": 316143822.0, + "step": 12337 + }, + { + "epoch": 1.3549308148473533, + "grad_norm": 2.5283992290496826, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7195101976394653, + "num_tokens": 316164263.0, + "step": 12338 + }, + { + "epoch": 1.355040632549967, + "grad_norm": 2.09047269821167, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7263759970664978, + "num_tokens": 316193601.0, + "step": 12339 + }, + { + "epoch": 1.3551504502525806, + "grad_norm": 2.3787083625793457, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7157981991767883, + "num_tokens": 316219350.0, + "step": 12340 + }, + { + "epoch": 1.3552602679551944, + "grad_norm": 2.144939422607422, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7018945217132568, + "num_tokens": 316246009.0, + "step": 12341 + }, + { + "epoch": 1.3553700856578081, + "grad_norm": 2.544909715652466, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7356922030448914, + "num_tokens": 316265407.0, + "step": 12342 + }, + { + "epoch": 1.3554799033604217, + "grad_norm": 2.206552505493164, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7315948009490967, + "num_tokens": 316291495.0, + "step": 12343 + }, + { + "epoch": 1.3555897210630354, + "grad_norm": 2.0713675022125244, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7155003547668457, + "num_tokens": 316321559.0, + "step": 12344 + }, + { + "epoch": 1.355699538765649, + "grad_norm": 2.4145166873931885, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7318111658096313, + "num_tokens": 316343637.0, + "step": 12345 + }, + { + "epoch": 1.3558093564682627, + "grad_norm": 2.4120240211486816, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7238014340400696, + "num_tokens": 316366065.0, + "step": 12346 + }, + { + "epoch": 1.3559191741708763, + "grad_norm": 2.5942957401275635, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7408154010772705, + "num_tokens": 316384745.0, + "step": 12347 + }, + { + "epoch": 1.35602899187349, + "grad_norm": 2.316676616668701, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7254883050918579, + "num_tokens": 316408570.0, + "step": 12348 + }, + { + "epoch": 1.3561388095761036, + "grad_norm": 2.2452492713928223, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7159290909767151, + "num_tokens": 316433766.0, + "step": 12349 + }, + { + "epoch": 1.3562486272787173, + "grad_norm": 2.246685028076172, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7349948287010193, + "num_tokens": 316460300.0, + "step": 12350 + }, + { + "epoch": 1.356358444981331, + "grad_norm": 2.142430543899536, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7132615447044373, + "num_tokens": 316487627.0, + "step": 12351 + }, + { + "epoch": 1.3564682626839446, + "grad_norm": 2.3861629962921143, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7683172225952148, + "num_tokens": 316508133.0, + "step": 12352 + }, + { + "epoch": 1.3565780803865584, + "grad_norm": 2.0739595890045166, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.6958953142166138, + "num_tokens": 316537416.0, + "step": 12353 + }, + { + "epoch": 1.356687898089172, + "grad_norm": 2.080716848373413, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7171356081962585, + "num_tokens": 316566155.0, + "step": 12354 + }, + { + "epoch": 1.3567977157917857, + "grad_norm": 2.1836085319519043, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7317110896110535, + "num_tokens": 316592758.0, + "step": 12355 + }, + { + "epoch": 1.3569075334943994, + "grad_norm": 2.441708564758301, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7050943374633789, + "num_tokens": 316616942.0, + "step": 12356 + }, + { + "epoch": 1.357017351197013, + "grad_norm": 2.429936170578003, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7226463556289673, + "num_tokens": 316640785.0, + "step": 12357 + }, + { + "epoch": 1.3571271688996267, + "grad_norm": 2.143463373184204, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.728075385093689, + "num_tokens": 316668849.0, + "step": 12358 + }, + { + "epoch": 1.3572369866022402, + "grad_norm": 2.5355494022369385, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7442386150360107, + "num_tokens": 316692110.0, + "step": 12359 + }, + { + "epoch": 1.357346804304854, + "grad_norm": 2.0736896991729736, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6930627226829529, + "num_tokens": 316720062.0, + "step": 12360 + }, + { + "epoch": 1.3574566220074675, + "grad_norm": 2.33270525932312, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7169452905654907, + "num_tokens": 316743757.0, + "step": 12361 + }, + { + "epoch": 1.3575664397100813, + "grad_norm": 2.2166075706481934, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.697871744632721, + "num_tokens": 316773125.0, + "step": 12362 + }, + { + "epoch": 1.3576762574126948, + "grad_norm": 2.385831117630005, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7476577758789062, + "num_tokens": 316795759.0, + "step": 12363 + }, + { + "epoch": 1.3577860751153086, + "grad_norm": 2.3739030361175537, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.722139298915863, + "num_tokens": 316818499.0, + "step": 12364 + }, + { + "epoch": 1.3578958928179223, + "grad_norm": 2.2424585819244385, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7531378269195557, + "num_tokens": 316843615.0, + "step": 12365 + }, + { + "epoch": 1.3580057105205359, + "grad_norm": 2.5650599002838135, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7130638360977173, + "num_tokens": 316865270.0, + "step": 12366 + }, + { + "epoch": 1.3581155282231496, + "grad_norm": 2.2591683864593506, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7209588289260864, + "num_tokens": 316892228.0, + "step": 12367 + }, + { + "epoch": 1.3582253459257632, + "grad_norm": 2.4561538696289062, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7431392669677734, + "num_tokens": 316912642.0, + "step": 12368 + }, + { + "epoch": 1.358335163628377, + "grad_norm": 2.574104070663452, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7067314386367798, + "num_tokens": 316934497.0, + "step": 12369 + }, + { + "epoch": 1.3584449813309907, + "grad_norm": 1.9324265718460083, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6921451687812805, + "num_tokens": 316968582.0, + "step": 12370 + }, + { + "epoch": 1.3585547990336042, + "grad_norm": 2.1024091243743896, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.71452796459198, + "num_tokens": 316997418.0, + "step": 12371 + }, + { + "epoch": 1.3586646167362177, + "grad_norm": 2.794706106185913, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7164909839630127, + "num_tokens": 317017556.0, + "step": 12372 + }, + { + "epoch": 1.3587744344388315, + "grad_norm": 2.251582622528076, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6964597105979919, + "num_tokens": 317043799.0, + "step": 12373 + }, + { + "epoch": 1.3588842521414453, + "grad_norm": 2.125896453857422, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6998537182807922, + "num_tokens": 317069839.0, + "step": 12374 + }, + { + "epoch": 1.3589940698440588, + "grad_norm": 2.3195507526397705, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7350407242774963, + "num_tokens": 317093039.0, + "step": 12375 + }, + { + "epoch": 1.3591038875466726, + "grad_norm": 2.232233762741089, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7280214428901672, + "num_tokens": 317119335.0, + "step": 12376 + }, + { + "epoch": 1.359213705249286, + "grad_norm": 2.244685649871826, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7280052304267883, + "num_tokens": 317142813.0, + "step": 12377 + }, + { + "epoch": 1.3593235229518998, + "grad_norm": 2.388354539871216, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7284612059593201, + "num_tokens": 317166328.0, + "step": 12378 + }, + { + "epoch": 1.3594333406545136, + "grad_norm": 2.3088812828063965, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.74130779504776, + "num_tokens": 317189090.0, + "step": 12379 + }, + { + "epoch": 1.3595431583571271, + "grad_norm": 2.5390467643737793, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7153286337852478, + "num_tokens": 317210445.0, + "step": 12380 + }, + { + "epoch": 1.359652976059741, + "grad_norm": 2.0754597187042236, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7269276976585388, + "num_tokens": 317239155.0, + "step": 12381 + }, + { + "epoch": 1.3597627937623544, + "grad_norm": 2.603303909301758, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7154644131660461, + "num_tokens": 317258767.0, + "step": 12382 + }, + { + "epoch": 1.3598726114649682, + "grad_norm": 2.5281295776367188, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7007209062576294, + "num_tokens": 317282499.0, + "step": 12383 + }, + { + "epoch": 1.359982429167582, + "grad_norm": 2.455505847930908, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7098527550697327, + "num_tokens": 317306004.0, + "step": 12384 + }, + { + "epoch": 1.3600922468701955, + "grad_norm": 2.3848602771759033, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.706108033657074, + "num_tokens": 317330977.0, + "step": 12385 + }, + { + "epoch": 1.360202064572809, + "grad_norm": 2.14935564994812, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6931408643722534, + "num_tokens": 317361441.0, + "step": 12386 + }, + { + "epoch": 1.3603118822754228, + "grad_norm": 2.3745388984680176, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.730430543422699, + "num_tokens": 317385321.0, + "step": 12387 + }, + { + "epoch": 1.3604216999780365, + "grad_norm": 2.101226568222046, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7111771106719971, + "num_tokens": 317414928.0, + "step": 12388 + }, + { + "epoch": 1.36053151768065, + "grad_norm": 2.342721700668335, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7378637790679932, + "num_tokens": 317437865.0, + "step": 12389 + }, + { + "epoch": 1.3606413353832638, + "grad_norm": 1.9894686937332153, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7453798651695251, + "num_tokens": 317469089.0, + "step": 12390 + }, + { + "epoch": 1.3607511530858774, + "grad_norm": 2.346182107925415, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7168170213699341, + "num_tokens": 317491497.0, + "step": 12391 + }, + { + "epoch": 1.360860970788491, + "grad_norm": 2.183722972869873, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6952069997787476, + "num_tokens": 317517891.0, + "step": 12392 + }, + { + "epoch": 1.3609707884911049, + "grad_norm": 2.2140207290649414, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7033039331436157, + "num_tokens": 317544852.0, + "step": 12393 + }, + { + "epoch": 1.3610806061937184, + "grad_norm": 2.464942216873169, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7286083698272705, + "num_tokens": 317566895.0, + "step": 12394 + }, + { + "epoch": 1.3611904238963322, + "grad_norm": 2.2189786434173584, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7405460476875305, + "num_tokens": 317591961.0, + "step": 12395 + }, + { + "epoch": 1.3613002415989457, + "grad_norm": 2.3872673511505127, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.722266435623169, + "num_tokens": 317615895.0, + "step": 12396 + }, + { + "epoch": 1.3614100593015594, + "grad_norm": 2.1403982639312744, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6964313983917236, + "num_tokens": 317644688.0, + "step": 12397 + }, + { + "epoch": 1.3615198770041732, + "grad_norm": 2.048884630203247, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7067670822143555, + "num_tokens": 317675310.0, + "step": 12398 + }, + { + "epoch": 1.3616296947067867, + "grad_norm": 2.17830228805542, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6943773031234741, + "num_tokens": 317704873.0, + "step": 12399 + }, + { + "epoch": 1.3617395124094003, + "grad_norm": 2.1922428607940674, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7138329148292542, + "num_tokens": 317732442.0, + "step": 12400 + }, + { + "epoch": 1.361849330112014, + "grad_norm": 2.4741368293762207, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7070953845977783, + "num_tokens": 317754038.0, + "step": 12401 + }, + { + "epoch": 1.3619591478146278, + "grad_norm": 2.266482353210449, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7257840633392334, + "num_tokens": 317777347.0, + "step": 12402 + }, + { + "epoch": 1.3620689655172413, + "grad_norm": 2.458244800567627, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7051630020141602, + "num_tokens": 317800014.0, + "step": 12403 + }, + { + "epoch": 1.362178783219855, + "grad_norm": 2.2362375259399414, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7210767865180969, + "num_tokens": 317827032.0, + "step": 12404 + }, + { + "epoch": 1.3622886009224686, + "grad_norm": 2.1907246112823486, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.695176362991333, + "num_tokens": 317855680.0, + "step": 12405 + }, + { + "epoch": 1.3623984186250824, + "grad_norm": 2.236865282058716, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7418826818466187, + "num_tokens": 317880201.0, + "step": 12406 + }, + { + "epoch": 1.3625082363276961, + "grad_norm": 2.4715452194213867, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7170478105545044, + "num_tokens": 317902511.0, + "step": 12407 + }, + { + "epoch": 1.3626180540303097, + "grad_norm": 2.4895496368408203, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7193045616149902, + "num_tokens": 317925287.0, + "step": 12408 + }, + { + "epoch": 1.3627278717329234, + "grad_norm": 2.12640380859375, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7253681421279907, + "num_tokens": 317953684.0, + "step": 12409 + }, + { + "epoch": 1.362837689435537, + "grad_norm": 2.3056156635284424, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7235131859779358, + "num_tokens": 317981657.0, + "step": 12410 + }, + { + "epoch": 1.3629475071381507, + "grad_norm": 2.299551486968994, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7320646047592163, + "num_tokens": 318006609.0, + "step": 12411 + }, + { + "epoch": 1.3630573248407643, + "grad_norm": 2.1132140159606934, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.70843106508255, + "num_tokens": 318033545.0, + "step": 12412 + }, + { + "epoch": 1.363167142543378, + "grad_norm": 2.239161491394043, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7286036610603333, + "num_tokens": 318057285.0, + "step": 12413 + }, + { + "epoch": 1.3632769602459915, + "grad_norm": 2.3544328212738037, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7340339422225952, + "num_tokens": 318079418.0, + "step": 12414 + }, + { + "epoch": 1.3633867779486053, + "grad_norm": 2.2606899738311768, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7097344994544983, + "num_tokens": 318104465.0, + "step": 12415 + }, + { + "epoch": 1.363496595651219, + "grad_norm": 2.248769998550415, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7008171081542969, + "num_tokens": 318132250.0, + "step": 12416 + }, + { + "epoch": 1.3636064133538326, + "grad_norm": 2.1716160774230957, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7143489122390747, + "num_tokens": 318158500.0, + "step": 12417 + }, + { + "epoch": 1.3637162310564463, + "grad_norm": 1.9618594646453857, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.681909441947937, + "num_tokens": 318192479.0, + "step": 12418 + }, + { + "epoch": 1.3638260487590599, + "grad_norm": 2.0356838703155518, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7116569876670837, + "num_tokens": 318223313.0, + "step": 12419 + }, + { + "epoch": 1.3639358664616736, + "grad_norm": 2.236900806427002, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7280046939849854, + "num_tokens": 318249678.0, + "step": 12420 + }, + { + "epoch": 1.3640456841642874, + "grad_norm": 2.1773910522460938, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7226815223693848, + "num_tokens": 318277009.0, + "step": 12421 + }, + { + "epoch": 1.364155501866901, + "grad_norm": 2.020618200302124, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7022887468338013, + "num_tokens": 318308592.0, + "step": 12422 + }, + { + "epoch": 1.3642653195695145, + "grad_norm": 2.3918979167938232, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7277827262878418, + "num_tokens": 318331388.0, + "step": 12423 + }, + { + "epoch": 1.3643751372721282, + "grad_norm": 2.14566707611084, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7148740291595459, + "num_tokens": 318358451.0, + "step": 12424 + }, + { + "epoch": 1.364484954974742, + "grad_norm": 2.7322816848754883, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7237919569015503, + "num_tokens": 318376762.0, + "step": 12425 + }, + { + "epoch": 1.3645947726773555, + "grad_norm": 2.321981430053711, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.6982921957969666, + "num_tokens": 318403800.0, + "step": 12426 + }, + { + "epoch": 1.3647045903799693, + "grad_norm": 2.5311074256896973, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.709903359413147, + "num_tokens": 318426983.0, + "step": 12427 + }, + { + "epoch": 1.3648144080825828, + "grad_norm": 2.182985782623291, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6956214308738708, + "num_tokens": 318457162.0, + "step": 12428 + }, + { + "epoch": 1.3649242257851966, + "grad_norm": 2.4837517738342285, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7280890941619873, + "num_tokens": 318479251.0, + "step": 12429 + }, + { + "epoch": 1.3650340434878103, + "grad_norm": 2.0221354961395264, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7447447180747986, + "num_tokens": 318506619.0, + "step": 12430 + }, + { + "epoch": 1.3651438611904239, + "grad_norm": 2.012042760848999, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7077908515930176, + "num_tokens": 318537042.0, + "step": 12431 + }, + { + "epoch": 1.3652536788930376, + "grad_norm": 2.1933977603912354, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7142850756645203, + "num_tokens": 318561808.0, + "step": 12432 + }, + { + "epoch": 1.3653634965956511, + "grad_norm": 2.2256407737731934, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7278493642807007, + "num_tokens": 318585992.0, + "step": 12433 + }, + { + "epoch": 1.365473314298265, + "grad_norm": 2.275575876235962, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7172907590866089, + "num_tokens": 318609950.0, + "step": 12434 + }, + { + "epoch": 1.3655831320008787, + "grad_norm": 2.2649941444396973, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7059012651443481, + "num_tokens": 318635429.0, + "step": 12435 + }, + { + "epoch": 1.3656929497034922, + "grad_norm": 2.247263193130493, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7167587280273438, + "num_tokens": 318659591.0, + "step": 12436 + }, + { + "epoch": 1.3658027674061057, + "grad_norm": 2.2308897972106934, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6895384788513184, + "num_tokens": 318686897.0, + "step": 12437 + }, + { + "epoch": 1.3659125851087195, + "grad_norm": 2.396315574645996, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7249096632003784, + "num_tokens": 318709575.0, + "step": 12438 + }, + { + "epoch": 1.3660224028113332, + "grad_norm": 2.5127205848693848, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7221181392669678, + "num_tokens": 318731170.0, + "step": 12439 + }, + { + "epoch": 1.3661322205139468, + "grad_norm": 2.3530452251434326, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.709084153175354, + "num_tokens": 318756533.0, + "step": 12440 + }, + { + "epoch": 1.3662420382165605, + "grad_norm": 2.3933863639831543, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7154716849327087, + "num_tokens": 318780798.0, + "step": 12441 + }, + { + "epoch": 1.366351855919174, + "grad_norm": 2.133375883102417, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6984473466873169, + "num_tokens": 318809897.0, + "step": 12442 + }, + { + "epoch": 1.3664616736217878, + "grad_norm": 2.3138163089752197, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7361907958984375, + "num_tokens": 318832602.0, + "step": 12443 + }, + { + "epoch": 1.3665714913244016, + "grad_norm": 2.555776357650757, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7214133739471436, + "num_tokens": 318852522.0, + "step": 12444 + }, + { + "epoch": 1.3666813090270151, + "grad_norm": 2.3542840480804443, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7366701364517212, + "num_tokens": 318874556.0, + "step": 12445 + }, + { + "epoch": 1.3667911267296289, + "grad_norm": 2.125866413116455, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7180483937263489, + "num_tokens": 318902424.0, + "step": 12446 + }, + { + "epoch": 1.3669009444322424, + "grad_norm": 2.6814870834350586, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7102982401847839, + "num_tokens": 318921897.0, + "step": 12447 + }, + { + "epoch": 1.3670107621348562, + "grad_norm": 2.5761187076568604, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7462646961212158, + "num_tokens": 318941089.0, + "step": 12448 + }, + { + "epoch": 1.36712057983747, + "grad_norm": 2.2844743728637695, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7310261726379395, + "num_tokens": 318964212.0, + "step": 12449 + }, + { + "epoch": 1.3672303975400835, + "grad_norm": 2.331303596496582, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7310582399368286, + "num_tokens": 318985624.0, + "step": 12450 + }, + { + "epoch": 1.367340215242697, + "grad_norm": 2.2535314559936523, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7262625098228455, + "num_tokens": 319010590.0, + "step": 12451 + }, + { + "epoch": 1.3674500329453108, + "grad_norm": 2.389207124710083, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.719741702079773, + "num_tokens": 319033567.0, + "step": 12452 + }, + { + "epoch": 1.3675598506479245, + "grad_norm": 2.1867053508758545, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7062125205993652, + "num_tokens": 319061177.0, + "step": 12453 + }, + { + "epoch": 1.367669668350538, + "grad_norm": 2.050220489501953, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7244167327880859, + "num_tokens": 319090298.0, + "step": 12454 + }, + { + "epoch": 1.3677794860531518, + "grad_norm": 2.373490333557129, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7146214246749878, + "num_tokens": 319112964.0, + "step": 12455 + }, + { + "epoch": 1.3678893037557653, + "grad_norm": 2.050597667694092, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.6967000961303711, + "num_tokens": 319145546.0, + "step": 12456 + }, + { + "epoch": 1.367999121458379, + "grad_norm": 2.0667595863342285, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7138370275497437, + "num_tokens": 319175776.0, + "step": 12457 + }, + { + "epoch": 1.3681089391609929, + "grad_norm": 1.8602516651153564, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7177197933197021, + "num_tokens": 319210665.0, + "step": 12458 + }, + { + "epoch": 1.3682187568636064, + "grad_norm": 2.1444575786590576, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.698074460029602, + "num_tokens": 319237644.0, + "step": 12459 + }, + { + "epoch": 1.3683285745662201, + "grad_norm": 2.3101184368133545, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7363876700401306, + "num_tokens": 319262475.0, + "step": 12460 + }, + { + "epoch": 1.3684383922688337, + "grad_norm": 2.1744961738586426, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6805922985076904, + "num_tokens": 319290402.0, + "step": 12461 + }, + { + "epoch": 1.3685482099714474, + "grad_norm": 2.382416009902954, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7150763273239136, + "num_tokens": 319313723.0, + "step": 12462 + }, + { + "epoch": 1.3686580276740612, + "grad_norm": 2.115727186203003, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7233227491378784, + "num_tokens": 319340383.0, + "step": 12463 + }, + { + "epoch": 1.3687678453766747, + "grad_norm": 2.166072130203247, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7204796671867371, + "num_tokens": 319367335.0, + "step": 12464 + }, + { + "epoch": 1.3688776630792883, + "grad_norm": 2.279099464416504, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7106718420982361, + "num_tokens": 319393932.0, + "step": 12465 + }, + { + "epoch": 1.368987480781902, + "grad_norm": 2.6850767135620117, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7290777564048767, + "num_tokens": 319412649.0, + "step": 12466 + }, + { + "epoch": 1.3690972984845158, + "grad_norm": 2.254673480987549, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.727144718170166, + "num_tokens": 319435839.0, + "step": 12467 + }, + { + "epoch": 1.3692071161871293, + "grad_norm": 2.3185062408447266, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6953086256980896, + "num_tokens": 319461698.0, + "step": 12468 + }, + { + "epoch": 1.369316933889743, + "grad_norm": 2.371279001235962, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7346874475479126, + "num_tokens": 319485170.0, + "step": 12469 + }, + { + "epoch": 1.3694267515923566, + "grad_norm": 2.0273640155792236, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.714478611946106, + "num_tokens": 319519235.0, + "step": 12470 + }, + { + "epoch": 1.3695365692949704, + "grad_norm": 2.1723403930664062, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7152371406555176, + "num_tokens": 319546874.0, + "step": 12471 + }, + { + "epoch": 1.3696463869975841, + "grad_norm": 2.1711180210113525, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.719812273979187, + "num_tokens": 319572081.0, + "step": 12472 + }, + { + "epoch": 1.3697562047001977, + "grad_norm": 2.2846508026123047, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7227150201797485, + "num_tokens": 319599058.0, + "step": 12473 + }, + { + "epoch": 1.3698660224028114, + "grad_norm": 2.3650686740875244, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7032098770141602, + "num_tokens": 319623314.0, + "step": 12474 + }, + { + "epoch": 1.369975840105425, + "grad_norm": 2.0842947959899902, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7165395021438599, + "num_tokens": 319650149.0, + "step": 12475 + }, + { + "epoch": 1.3700856578080387, + "grad_norm": 2.3749897480010986, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7299761176109314, + "num_tokens": 319673179.0, + "step": 12476 + }, + { + "epoch": 1.3701954755106522, + "grad_norm": 2.6289515495300293, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7173047065734863, + "num_tokens": 319697890.0, + "step": 12477 + }, + { + "epoch": 1.370305293213266, + "grad_norm": 1.9965561628341675, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7318137288093567, + "num_tokens": 319728119.0, + "step": 12478 + }, + { + "epoch": 1.3704151109158795, + "grad_norm": 2.28653883934021, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7165735363960266, + "num_tokens": 319752376.0, + "step": 12479 + }, + { + "epoch": 1.3705249286184933, + "grad_norm": 2.2880234718322754, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.736332356929779, + "num_tokens": 319775513.0, + "step": 12480 + }, + { + "epoch": 1.370634746321107, + "grad_norm": 2.3178160190582275, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7228723764419556, + "num_tokens": 319798269.0, + "step": 12481 + }, + { + "epoch": 1.3707445640237206, + "grad_norm": 2.1797382831573486, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7135643362998962, + "num_tokens": 319825532.0, + "step": 12482 + }, + { + "epoch": 1.3708543817263343, + "grad_norm": 2.24465274810791, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7177233099937439, + "num_tokens": 319852786.0, + "step": 12483 + }, + { + "epoch": 1.3709641994289479, + "grad_norm": 2.0864458084106445, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7325562834739685, + "num_tokens": 319881785.0, + "step": 12484 + }, + { + "epoch": 1.3710740171315616, + "grad_norm": 2.3122565746307373, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7094799280166626, + "num_tokens": 319905890.0, + "step": 12485 + }, + { + "epoch": 1.3711838348341754, + "grad_norm": 2.220660448074341, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7422848343849182, + "num_tokens": 319931519.0, + "step": 12486 + }, + { + "epoch": 1.371293652536789, + "grad_norm": 2.135356903076172, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7274153828620911, + "num_tokens": 319958376.0, + "step": 12487 + }, + { + "epoch": 1.3714034702394025, + "grad_norm": 2.225374460220337, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7133774757385254, + "num_tokens": 319985515.0, + "step": 12488 + }, + { + "epoch": 1.3715132879420162, + "grad_norm": 2.2784006595611572, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7343000769615173, + "num_tokens": 320009769.0, + "step": 12489 + }, + { + "epoch": 1.37162310564463, + "grad_norm": 2.4495036602020264, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.718809187412262, + "num_tokens": 320032590.0, + "step": 12490 + }, + { + "epoch": 1.3717329233472435, + "grad_norm": 2.334702968597412, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7259360551834106, + "num_tokens": 320056487.0, + "step": 12491 + }, + { + "epoch": 1.3718427410498573, + "grad_norm": 2.2028443813323975, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7403002381324768, + "num_tokens": 320081494.0, + "step": 12492 + }, + { + "epoch": 1.3719525587524708, + "grad_norm": 2.463895797729492, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7272757291793823, + "num_tokens": 320104636.0, + "step": 12493 + }, + { + "epoch": 1.3720623764550846, + "grad_norm": 2.431544542312622, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7062875628471375, + "num_tokens": 320128036.0, + "step": 12494 + }, + { + "epoch": 1.3721721941576983, + "grad_norm": 2.38260817527771, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7404013872146606, + "num_tokens": 320150597.0, + "step": 12495 + }, + { + "epoch": 1.3722820118603118, + "grad_norm": 2.2886452674865723, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7194418907165527, + "num_tokens": 320174333.0, + "step": 12496 + }, + { + "epoch": 1.3723918295629256, + "grad_norm": 2.093111038208008, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7141421437263489, + "num_tokens": 320204109.0, + "step": 12497 + }, + { + "epoch": 1.3725016472655391, + "grad_norm": 2.1096391677856445, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7264179587364197, + "num_tokens": 320229885.0, + "step": 12498 + }, + { + "epoch": 1.372611464968153, + "grad_norm": 2.22204852104187, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7336159944534302, + "num_tokens": 320254362.0, + "step": 12499 + }, + { + "epoch": 1.3727212826707667, + "grad_norm": 2.181591033935547, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7171130776405334, + "num_tokens": 320284109.0, + "step": 12500 + }, + { + "epoch": 1.3728311003733802, + "grad_norm": 2.7242655754089355, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7277172803878784, + "num_tokens": 320302876.0, + "step": 12501 + }, + { + "epoch": 1.3729409180759937, + "grad_norm": 2.4345786571502686, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7179836630821228, + "num_tokens": 320323885.0, + "step": 12502 + }, + { + "epoch": 1.3730507357786075, + "grad_norm": 2.2852983474731445, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7107013463973999, + "num_tokens": 320348703.0, + "step": 12503 + }, + { + "epoch": 1.3731605534812212, + "grad_norm": 2.080636978149414, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7122960090637207, + "num_tokens": 320378943.0, + "step": 12504 + }, + { + "epoch": 1.3732703711838348, + "grad_norm": 2.2400665283203125, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7081847190856934, + "num_tokens": 320406358.0, + "step": 12505 + }, + { + "epoch": 1.3733801888864485, + "grad_norm": 1.982692837715149, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7234582304954529, + "num_tokens": 320437935.0, + "step": 12506 + }, + { + "epoch": 1.373490006589062, + "grad_norm": 2.430781126022339, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7143413424491882, + "num_tokens": 320460374.0, + "step": 12507 + }, + { + "epoch": 1.3735998242916758, + "grad_norm": 2.112097978591919, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7140798568725586, + "num_tokens": 320487235.0, + "step": 12508 + }, + { + "epoch": 1.3737096419942896, + "grad_norm": 2.201447010040283, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7417435646057129, + "num_tokens": 320513415.0, + "step": 12509 + }, + { + "epoch": 1.3738194596969031, + "grad_norm": 2.33251690864563, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7119659185409546, + "num_tokens": 320537677.0, + "step": 12510 + }, + { + "epoch": 1.3739292773995169, + "grad_norm": 2.4489312171936035, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7177194356918335, + "num_tokens": 320559429.0, + "step": 12511 + }, + { + "epoch": 1.3740390951021304, + "grad_norm": 2.0840935707092285, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.7037347555160522, + "num_tokens": 320589181.0, + "step": 12512 + }, + { + "epoch": 1.3741489128047442, + "grad_norm": 2.2442243099212646, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6839401721954346, + "num_tokens": 320616049.0, + "step": 12513 + }, + { + "epoch": 1.374258730507358, + "grad_norm": 2.053541898727417, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6839072704315186, + "num_tokens": 320647466.0, + "step": 12514 + }, + { + "epoch": 1.3743685482099715, + "grad_norm": 2.3206846714019775, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7218645811080933, + "num_tokens": 320670017.0, + "step": 12515 + }, + { + "epoch": 1.374478365912585, + "grad_norm": 2.3954708576202393, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7195661067962646, + "num_tokens": 320694738.0, + "step": 12516 + }, + { + "epoch": 1.3745881836151987, + "grad_norm": 2.032586097717285, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6944345831871033, + "num_tokens": 320725753.0, + "step": 12517 + }, + { + "epoch": 1.3746980013178125, + "grad_norm": 2.4234161376953125, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7172847986221313, + "num_tokens": 320748060.0, + "step": 12518 + }, + { + "epoch": 1.374807819020426, + "grad_norm": 2.1726605892181396, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7244783043861389, + "num_tokens": 320777298.0, + "step": 12519 + }, + { + "epoch": 1.3749176367230398, + "grad_norm": 2.018047571182251, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7188239097595215, + "num_tokens": 320807611.0, + "step": 12520 + }, + { + "epoch": 1.3750274544256533, + "grad_norm": 2.320005416870117, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7335771322250366, + "num_tokens": 320830362.0, + "step": 12521 + }, + { + "epoch": 1.375137272128267, + "grad_norm": 2.315154552459717, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.726755678653717, + "num_tokens": 320852939.0, + "step": 12522 + }, + { + "epoch": 1.3752470898308808, + "grad_norm": 2.1709914207458496, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.703157901763916, + "num_tokens": 320879499.0, + "step": 12523 + }, + { + "epoch": 1.3753569075334944, + "grad_norm": 2.4572243690490723, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7222065925598145, + "num_tokens": 320902375.0, + "step": 12524 + }, + { + "epoch": 1.3754667252361081, + "grad_norm": 2.2089381217956543, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7446367144584656, + "num_tokens": 320927611.0, + "step": 12525 + }, + { + "epoch": 1.3755765429387217, + "grad_norm": 2.2413244247436523, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7040894031524658, + "num_tokens": 320953078.0, + "step": 12526 + }, + { + "epoch": 1.3756863606413354, + "grad_norm": 2.001065731048584, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7538883686065674, + "num_tokens": 320982020.0, + "step": 12527 + }, + { + "epoch": 1.3757961783439492, + "grad_norm": 2.100046157836914, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7093814611434937, + "num_tokens": 321010428.0, + "step": 12528 + }, + { + "epoch": 1.3759059960465627, + "grad_norm": 2.3513479232788086, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7202861309051514, + "num_tokens": 321035345.0, + "step": 12529 + }, + { + "epoch": 1.3760158137491763, + "grad_norm": 2.356550931930542, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7367192506790161, + "num_tokens": 321057831.0, + "step": 12530 + }, + { + "epoch": 1.37612563145179, + "grad_norm": 2.2830326557159424, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7293610572814941, + "num_tokens": 321083466.0, + "step": 12531 + }, + { + "epoch": 1.3762354491544038, + "grad_norm": 2.3075404167175293, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7392855882644653, + "num_tokens": 321107019.0, + "step": 12532 + }, + { + "epoch": 1.3763452668570173, + "grad_norm": 2.0598654747009277, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7021857500076294, + "num_tokens": 321138187.0, + "step": 12533 + }, + { + "epoch": 1.376455084559631, + "grad_norm": 2.0820813179016113, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7275604009628296, + "num_tokens": 321166334.0, + "step": 12534 + }, + { + "epoch": 1.3765649022622446, + "grad_norm": 2.281554698944092, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7347447276115417, + "num_tokens": 321189151.0, + "step": 12535 + }, + { + "epoch": 1.3766747199648584, + "grad_norm": 2.260037660598755, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7209425568580627, + "num_tokens": 321213668.0, + "step": 12536 + }, + { + "epoch": 1.376784537667472, + "grad_norm": 2.250800848007202, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.710720419883728, + "num_tokens": 321239866.0, + "step": 12537 + }, + { + "epoch": 1.3768943553700856, + "grad_norm": 2.0532562732696533, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7089554071426392, + "num_tokens": 321268147.0, + "step": 12538 + }, + { + "epoch": 1.3770041730726994, + "grad_norm": 2.230586290359497, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7048516273498535, + "num_tokens": 321295184.0, + "step": 12539 + }, + { + "epoch": 1.377113990775313, + "grad_norm": 2.2040014266967773, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7042725086212158, + "num_tokens": 321323467.0, + "step": 12540 + }, + { + "epoch": 1.3772238084779267, + "grad_norm": 2.3286468982696533, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7424589395523071, + "num_tokens": 321347553.0, + "step": 12541 + }, + { + "epoch": 1.3773336261805402, + "grad_norm": 2.1173553466796875, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7681240439414978, + "num_tokens": 321372486.0, + "step": 12542 + }, + { + "epoch": 1.377443443883154, + "grad_norm": 2.188244342803955, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.718826413154602, + "num_tokens": 321400522.0, + "step": 12543 + }, + { + "epoch": 1.3775532615857675, + "grad_norm": 2.2963240146636963, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.740446150302887, + "num_tokens": 321423022.0, + "step": 12544 + }, + { + "epoch": 1.3776630792883813, + "grad_norm": 2.4428927898406982, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7069594860076904, + "num_tokens": 321445166.0, + "step": 12545 + }, + { + "epoch": 1.377772896990995, + "grad_norm": 2.141774892807007, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7262165546417236, + "num_tokens": 321472170.0, + "step": 12546 + }, + { + "epoch": 1.3778827146936086, + "grad_norm": 2.04758620262146, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6983882784843445, + "num_tokens": 321503775.0, + "step": 12547 + }, + { + "epoch": 1.3779925323962223, + "grad_norm": 2.202054262161255, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7531778216362, + "num_tokens": 321528481.0, + "step": 12548 + }, + { + "epoch": 1.3781023500988359, + "grad_norm": 2.4156265258789062, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7231096625328064, + "num_tokens": 321548922.0, + "step": 12549 + }, + { + "epoch": 1.3782121678014496, + "grad_norm": 2.2672131061553955, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7316514849662781, + "num_tokens": 321572623.0, + "step": 12550 + }, + { + "epoch": 1.3783219855040634, + "grad_norm": 2.412947654724121, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7079266905784607, + "num_tokens": 321597977.0, + "step": 12551 + }, + { + "epoch": 1.378431803206677, + "grad_norm": 2.3930346965789795, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7536090612411499, + "num_tokens": 321621974.0, + "step": 12552 + }, + { + "epoch": 1.3785416209092904, + "grad_norm": 2.4844729900360107, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7092752456665039, + "num_tokens": 321646307.0, + "step": 12553 + }, + { + "epoch": 1.3786514386119042, + "grad_norm": 2.1776654720306396, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7513209581375122, + "num_tokens": 321671432.0, + "step": 12554 + }, + { + "epoch": 1.378761256314518, + "grad_norm": 2.9027271270751953, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.738461971282959, + "num_tokens": 321689153.0, + "step": 12555 + }, + { + "epoch": 1.3788710740171315, + "grad_norm": 2.458921194076538, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7154600620269775, + "num_tokens": 321712169.0, + "step": 12556 + }, + { + "epoch": 1.3789808917197452, + "grad_norm": 2.2844362258911133, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7327768802642822, + "num_tokens": 321736947.0, + "step": 12557 + }, + { + "epoch": 1.3790907094223588, + "grad_norm": 2.4078938961029053, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7104986310005188, + "num_tokens": 321761196.0, + "step": 12558 + }, + { + "epoch": 1.3792005271249725, + "grad_norm": 2.1880016326904297, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7374853491783142, + "num_tokens": 321787285.0, + "step": 12559 + }, + { + "epoch": 1.3793103448275863, + "grad_norm": 2.34712815284729, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7236849069595337, + "num_tokens": 321812088.0, + "step": 12560 + }, + { + "epoch": 1.3794201625301998, + "grad_norm": 2.450904130935669, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7329123020172119, + "num_tokens": 321833311.0, + "step": 12561 + }, + { + "epoch": 1.3795299802328136, + "grad_norm": 2.288560628890991, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7241953611373901, + "num_tokens": 321857745.0, + "step": 12562 + }, + { + "epoch": 1.3796397979354271, + "grad_norm": 2.182468891143799, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7000343203544617, + "num_tokens": 321887345.0, + "step": 12563 + }, + { + "epoch": 1.3797496156380409, + "grad_norm": 2.097865104675293, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7008010745048523, + "num_tokens": 321917064.0, + "step": 12564 + }, + { + "epoch": 1.3798594333406546, + "grad_norm": 2.0114474296569824, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7512418031692505, + "num_tokens": 321946524.0, + "step": 12565 + }, + { + "epoch": 1.3799692510432682, + "grad_norm": 2.117713689804077, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7105161547660828, + "num_tokens": 321975444.0, + "step": 12566 + }, + { + "epoch": 1.3800790687458817, + "grad_norm": 2.5232739448547363, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7312843799591064, + "num_tokens": 321996319.0, + "step": 12567 + }, + { + "epoch": 1.3801888864484955, + "grad_norm": 2.142277479171753, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6979014873504639, + "num_tokens": 322025557.0, + "step": 12568 + }, + { + "epoch": 1.3802987041511092, + "grad_norm": 2.2222750186920166, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7059130668640137, + "num_tokens": 322050805.0, + "step": 12569 + }, + { + "epoch": 1.3804085218537228, + "grad_norm": 1.9378669261932373, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7395585775375366, + "num_tokens": 322083509.0, + "step": 12570 + }, + { + "epoch": 1.3805183395563365, + "grad_norm": 2.276848316192627, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7338893413543701, + "num_tokens": 322107107.0, + "step": 12571 + }, + { + "epoch": 1.38062815725895, + "grad_norm": 2.093029499053955, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7267947793006897, + "num_tokens": 322135628.0, + "step": 12572 + }, + { + "epoch": 1.3807379749615638, + "grad_norm": 2.173336982727051, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7178448438644409, + "num_tokens": 322163721.0, + "step": 12573 + }, + { + "epoch": 1.3808477926641776, + "grad_norm": 2.2771739959716797, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7178859710693359, + "num_tokens": 322189722.0, + "step": 12574 + }, + { + "epoch": 1.380957610366791, + "grad_norm": 2.2843194007873535, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7233806252479553, + "num_tokens": 322213411.0, + "step": 12575 + }, + { + "epoch": 1.3810674280694049, + "grad_norm": 2.328839063644409, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.723623514175415, + "num_tokens": 322239456.0, + "step": 12576 + }, + { + "epoch": 1.3811772457720184, + "grad_norm": 2.13968563079834, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7322984933853149, + "num_tokens": 322267453.0, + "step": 12577 + }, + { + "epoch": 1.3812870634746321, + "grad_norm": 2.2779834270477295, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7175336480140686, + "num_tokens": 322293005.0, + "step": 12578 + }, + { + "epoch": 1.381396881177246, + "grad_norm": 2.275789976119995, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7053772211074829, + "num_tokens": 322318982.0, + "step": 12579 + }, + { + "epoch": 1.3815066988798594, + "grad_norm": 2.165010929107666, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7057712078094482, + "num_tokens": 322348293.0, + "step": 12580 + }, + { + "epoch": 1.381616516582473, + "grad_norm": 2.5054783821105957, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7390309572219849, + "num_tokens": 322370115.0, + "step": 12581 + }, + { + "epoch": 1.3817263342850867, + "grad_norm": 2.1411006450653076, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7011505365371704, + "num_tokens": 322400987.0, + "step": 12582 + }, + { + "epoch": 1.3818361519877005, + "grad_norm": 2.1755783557891846, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7357544898986816, + "num_tokens": 322429832.0, + "step": 12583 + }, + { + "epoch": 1.381945969690314, + "grad_norm": 2.1114227771759033, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7172546982765198, + "num_tokens": 322457831.0, + "step": 12584 + }, + { + "epoch": 1.3820557873929278, + "grad_norm": 2.388127326965332, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7181550860404968, + "num_tokens": 322480869.0, + "step": 12585 + }, + { + "epoch": 1.3821656050955413, + "grad_norm": 2.217944383621216, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7247869968414307, + "num_tokens": 322506382.0, + "step": 12586 + }, + { + "epoch": 1.382275422798155, + "grad_norm": 2.180753707885742, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7075265645980835, + "num_tokens": 322534664.0, + "step": 12587 + }, + { + "epoch": 1.3823852405007688, + "grad_norm": 2.479546308517456, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7128156423568726, + "num_tokens": 322558747.0, + "step": 12588 + }, + { + "epoch": 1.3824950582033824, + "grad_norm": 2.374959707260132, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7308583855628967, + "num_tokens": 322581582.0, + "step": 12589 + }, + { + "epoch": 1.3826048759059961, + "grad_norm": 2.0813663005828857, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7192554473876953, + "num_tokens": 322609484.0, + "step": 12590 + }, + { + "epoch": 1.3827146936086097, + "grad_norm": 2.3267908096313477, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7065368294715881, + "num_tokens": 322634092.0, + "step": 12591 + }, + { + "epoch": 1.3828245113112234, + "grad_norm": 1.892406940460205, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7208120822906494, + "num_tokens": 322663517.0, + "step": 12592 + }, + { + "epoch": 1.382934329013837, + "grad_norm": 2.24971604347229, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7439201474189758, + "num_tokens": 322687104.0, + "step": 12593 + }, + { + "epoch": 1.3830441467164507, + "grad_norm": 2.6663880348205566, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7172815203666687, + "num_tokens": 322706969.0, + "step": 12594 + }, + { + "epoch": 1.3831539644190642, + "grad_norm": 2.6372904777526855, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7242085337638855, + "num_tokens": 322727515.0, + "step": 12595 + }, + { + "epoch": 1.383263782121678, + "grad_norm": 2.6054000854492188, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7353676557540894, + "num_tokens": 322746373.0, + "step": 12596 + }, + { + "epoch": 1.3833735998242918, + "grad_norm": 2.182161569595337, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7022207975387573, + "num_tokens": 322775833.0, + "step": 12597 + }, + { + "epoch": 1.3834834175269053, + "grad_norm": 2.2097041606903076, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7252442836761475, + "num_tokens": 322800897.0, + "step": 12598 + }, + { + "epoch": 1.383593235229519, + "grad_norm": 2.2304623126983643, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7077547907829285, + "num_tokens": 322825903.0, + "step": 12599 + }, + { + "epoch": 1.3837030529321326, + "grad_norm": 2.3633079528808594, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7183026671409607, + "num_tokens": 322849423.0, + "step": 12600 + }, + { + "epoch": 1.3838128706347463, + "grad_norm": 2.009413957595825, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7351514101028442, + "num_tokens": 322879907.0, + "step": 12601 + }, + { + "epoch": 1.38392268833736, + "grad_norm": 2.413191080093384, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6975226998329163, + "num_tokens": 322902732.0, + "step": 12602 + }, + { + "epoch": 1.3840325060399736, + "grad_norm": 2.084763288497925, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.705298125743866, + "num_tokens": 322931502.0, + "step": 12603 + }, + { + "epoch": 1.3841423237425874, + "grad_norm": 2.182607412338257, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7285264134407043, + "num_tokens": 322958081.0, + "step": 12604 + }, + { + "epoch": 1.384252141445201, + "grad_norm": 2.1119580268859863, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7469222545623779, + "num_tokens": 322984257.0, + "step": 12605 + }, + { + "epoch": 1.3843619591478147, + "grad_norm": 2.4773240089416504, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.740491509437561, + "num_tokens": 323004718.0, + "step": 12606 + }, + { + "epoch": 1.3844717768504282, + "grad_norm": 2.0646138191223145, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7219763994216919, + "num_tokens": 323034078.0, + "step": 12607 + }, + { + "epoch": 1.384581594553042, + "grad_norm": 2.260622024536133, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6933092474937439, + "num_tokens": 323060799.0, + "step": 12608 + }, + { + "epoch": 1.3846914122556555, + "grad_norm": 1.978926181793213, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7062795162200928, + "num_tokens": 323093105.0, + "step": 12609 + }, + { + "epoch": 1.3848012299582693, + "grad_norm": 2.0976293087005615, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6861139535903931, + "num_tokens": 323122776.0, + "step": 12610 + }, + { + "epoch": 1.384911047660883, + "grad_norm": 1.9793586730957031, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7170450687408447, + "num_tokens": 323152228.0, + "step": 12611 + }, + { + "epoch": 1.3850208653634966, + "grad_norm": 2.4612057209014893, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7413473129272461, + "num_tokens": 323173104.0, + "step": 12612 + }, + { + "epoch": 1.3851306830661103, + "grad_norm": 2.2033700942993164, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7070953845977783, + "num_tokens": 323202248.0, + "step": 12613 + }, + { + "epoch": 1.3852405007687238, + "grad_norm": 2.3651211261749268, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7182994484901428, + "num_tokens": 323225785.0, + "step": 12614 + }, + { + "epoch": 1.3853503184713376, + "grad_norm": 2.067882776260376, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6996535062789917, + "num_tokens": 323257787.0, + "step": 12615 + }, + { + "epoch": 1.3854601361739514, + "grad_norm": 2.2628583908081055, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.716661810874939, + "num_tokens": 323283395.0, + "step": 12616 + }, + { + "epoch": 1.385569953876565, + "grad_norm": 2.4153895378112793, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7051773071289062, + "num_tokens": 323307426.0, + "step": 12617 + }, + { + "epoch": 1.3856797715791784, + "grad_norm": 2.145491361618042, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7089251279830933, + "num_tokens": 323335966.0, + "step": 12618 + }, + { + "epoch": 1.3857895892817922, + "grad_norm": 2.344151258468628, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7211159467697144, + "num_tokens": 323359283.0, + "step": 12619 + }, + { + "epoch": 1.385899406984406, + "grad_norm": 2.2005090713500977, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7155223488807678, + "num_tokens": 323385573.0, + "step": 12620 + }, + { + "epoch": 1.3860092246870195, + "grad_norm": 1.9301234483718872, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7068328857421875, + "num_tokens": 323418219.0, + "step": 12621 + }, + { + "epoch": 1.3861190423896332, + "grad_norm": 2.432919979095459, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7069606781005859, + "num_tokens": 323441972.0, + "step": 12622 + }, + { + "epoch": 1.3862288600922468, + "grad_norm": 2.191476583480835, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7286670207977295, + "num_tokens": 323469082.0, + "step": 12623 + }, + { + "epoch": 1.3863386777948605, + "grad_norm": 2.2471179962158203, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7012625932693481, + "num_tokens": 323495678.0, + "step": 12624 + }, + { + "epoch": 1.3864484954974743, + "grad_norm": 2.1763827800750732, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7125211954116821, + "num_tokens": 323523265.0, + "step": 12625 + }, + { + "epoch": 1.3865583132000878, + "grad_norm": 2.4089837074279785, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7364462614059448, + "num_tokens": 323543594.0, + "step": 12626 + }, + { + "epoch": 1.3866681309027016, + "grad_norm": 2.188720464706421, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6976326704025269, + "num_tokens": 323573658.0, + "step": 12627 + }, + { + "epoch": 1.3867779486053151, + "grad_norm": 2.464493751525879, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7494411468505859, + "num_tokens": 323595902.0, + "step": 12628 + }, + { + "epoch": 1.3868877663079289, + "grad_norm": 2.2748539447784424, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7350775003433228, + "num_tokens": 323620214.0, + "step": 12629 + }, + { + "epoch": 1.3869975840105426, + "grad_norm": 2.076525926589966, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7326017618179321, + "num_tokens": 323646610.0, + "step": 12630 + }, + { + "epoch": 1.3871074017131562, + "grad_norm": 2.1321418285369873, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7262989282608032, + "num_tokens": 323674347.0, + "step": 12631 + }, + { + "epoch": 1.3872172194157697, + "grad_norm": 2.048978328704834, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7167153358459473, + "num_tokens": 323703575.0, + "step": 12632 + }, + { + "epoch": 1.3873270371183835, + "grad_norm": 2.0579609870910645, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7011882066726685, + "num_tokens": 323735242.0, + "step": 12633 + }, + { + "epoch": 1.3874368548209972, + "grad_norm": 2.234452247619629, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7218589782714844, + "num_tokens": 323760234.0, + "step": 12634 + }, + { + "epoch": 1.3875466725236107, + "grad_norm": 2.452367067337036, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.71308434009552, + "num_tokens": 323782946.0, + "step": 12635 + }, + { + "epoch": 1.3876564902262245, + "grad_norm": 2.2673983573913574, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6829203367233276, + "num_tokens": 323810400.0, + "step": 12636 + }, + { + "epoch": 1.387766307928838, + "grad_norm": 2.652193069458008, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7197052240371704, + "num_tokens": 323830820.0, + "step": 12637 + }, + { + "epoch": 1.3878761256314518, + "grad_norm": 2.1962733268737793, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7118932008743286, + "num_tokens": 323860581.0, + "step": 12638 + }, + { + "epoch": 1.3879859433340656, + "grad_norm": 2.3300278186798096, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7281144857406616, + "num_tokens": 323884540.0, + "step": 12639 + }, + { + "epoch": 1.388095761036679, + "grad_norm": 2.1254265308380127, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7167209386825562, + "num_tokens": 323912813.0, + "step": 12640 + }, + { + "epoch": 1.3882055787392928, + "grad_norm": 2.4378371238708496, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7475194931030273, + "num_tokens": 323933753.0, + "step": 12641 + }, + { + "epoch": 1.3883153964419064, + "grad_norm": 2.4055426120758057, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7423213124275208, + "num_tokens": 323955709.0, + "step": 12642 + }, + { + "epoch": 1.3884252141445201, + "grad_norm": 2.2889368534088135, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7203150987625122, + "num_tokens": 323979833.0, + "step": 12643 + }, + { + "epoch": 1.388535031847134, + "grad_norm": 2.6515331268310547, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7241881489753723, + "num_tokens": 324000870.0, + "step": 12644 + }, + { + "epoch": 1.3886448495497474, + "grad_norm": 2.1601243019104004, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7371479868888855, + "num_tokens": 324028922.0, + "step": 12645 + }, + { + "epoch": 1.388754667252361, + "grad_norm": 2.2802865505218506, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.713059663772583, + "num_tokens": 324055159.0, + "step": 12646 + }, + { + "epoch": 1.3888644849549747, + "grad_norm": 2.430534839630127, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.722136378288269, + "num_tokens": 324077194.0, + "step": 12647 + }, + { + "epoch": 1.3889743026575885, + "grad_norm": 2.3434853553771973, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7243815660476685, + "num_tokens": 324101197.0, + "step": 12648 + }, + { + "epoch": 1.389084120360202, + "grad_norm": 2.63702130317688, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7431643009185791, + "num_tokens": 324121179.0, + "step": 12649 + }, + { + "epoch": 1.3891939380628158, + "grad_norm": 2.183312177658081, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6972149610519409, + "num_tokens": 324149476.0, + "step": 12650 + }, + { + "epoch": 1.3893037557654293, + "grad_norm": 2.0137782096862793, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7169512510299683, + "num_tokens": 324181360.0, + "step": 12651 + }, + { + "epoch": 1.389413573468043, + "grad_norm": 2.226369857788086, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7166421413421631, + "num_tokens": 324209589.0, + "step": 12652 + }, + { + "epoch": 1.3895233911706568, + "grad_norm": 2.0301432609558105, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7060228586196899, + "num_tokens": 324239488.0, + "step": 12653 + }, + { + "epoch": 1.3896332088732704, + "grad_norm": 2.3213484287261963, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7164429426193237, + "num_tokens": 324265247.0, + "step": 12654 + }, + { + "epoch": 1.389743026575884, + "grad_norm": 2.008462429046631, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.70356684923172, + "num_tokens": 324297428.0, + "step": 12655 + }, + { + "epoch": 1.3898528442784976, + "grad_norm": 2.584733486175537, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7162376046180725, + "num_tokens": 324317616.0, + "step": 12656 + }, + { + "epoch": 1.3899626619811114, + "grad_norm": 2.057859182357788, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7244428992271423, + "num_tokens": 324344734.0, + "step": 12657 + }, + { + "epoch": 1.390072479683725, + "grad_norm": 2.1292803287506104, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7083402872085571, + "num_tokens": 324375763.0, + "step": 12658 + }, + { + "epoch": 1.3901822973863387, + "grad_norm": 2.2755680084228516, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7220078110694885, + "num_tokens": 324399284.0, + "step": 12659 + }, + { + "epoch": 1.3902921150889522, + "grad_norm": 2.357328414916992, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.745391309261322, + "num_tokens": 324422382.0, + "step": 12660 + }, + { + "epoch": 1.390401932791566, + "grad_norm": 2.444969654083252, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7321940660476685, + "num_tokens": 324445791.0, + "step": 12661 + }, + { + "epoch": 1.3905117504941797, + "grad_norm": 2.2549901008605957, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7424415349960327, + "num_tokens": 324468808.0, + "step": 12662 + }, + { + "epoch": 1.3906215681967933, + "grad_norm": 2.2903964519500732, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7334446907043457, + "num_tokens": 324493022.0, + "step": 12663 + }, + { + "epoch": 1.390731385899407, + "grad_norm": 2.0828444957733154, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7098326683044434, + "num_tokens": 324522676.0, + "step": 12664 + }, + { + "epoch": 1.3908412036020206, + "grad_norm": 2.48598575592041, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.714666485786438, + "num_tokens": 324544970.0, + "step": 12665 + }, + { + "epoch": 1.3909510213046343, + "grad_norm": 2.355586051940918, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7259830236434937, + "num_tokens": 324567342.0, + "step": 12666 + }, + { + "epoch": 1.391060839007248, + "grad_norm": 2.1479499340057373, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.72552490234375, + "num_tokens": 324596282.0, + "step": 12667 + }, + { + "epoch": 1.3911706567098616, + "grad_norm": 2.1415863037109375, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7414320707321167, + "num_tokens": 324621638.0, + "step": 12668 + }, + { + "epoch": 1.3912804744124752, + "grad_norm": 2.1978468894958496, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6984865069389343, + "num_tokens": 324647827.0, + "step": 12669 + }, + { + "epoch": 1.391390292115089, + "grad_norm": 1.8202458620071411, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7031183838844299, + "num_tokens": 324680601.0, + "step": 12670 + }, + { + "epoch": 1.3915001098177027, + "grad_norm": 2.144252300262451, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7287702560424805, + "num_tokens": 324707459.0, + "step": 12671 + }, + { + "epoch": 1.3916099275203162, + "grad_norm": 2.442305564880371, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7225714921951294, + "num_tokens": 324731974.0, + "step": 12672 + }, + { + "epoch": 1.39171974522293, + "grad_norm": 2.0898280143737793, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7153652310371399, + "num_tokens": 324758567.0, + "step": 12673 + }, + { + "epoch": 1.3918295629255435, + "grad_norm": 2.404858112335205, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7433931231498718, + "num_tokens": 324778841.0, + "step": 12674 + }, + { + "epoch": 1.3919393806281573, + "grad_norm": 2.165741443634033, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7348107099533081, + "num_tokens": 324803891.0, + "step": 12675 + }, + { + "epoch": 1.392049198330771, + "grad_norm": 2.158480405807495, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7384604215621948, + "num_tokens": 324829861.0, + "step": 12676 + }, + { + "epoch": 1.3921590160333845, + "grad_norm": 2.2660670280456543, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7241925597190857, + "num_tokens": 324855137.0, + "step": 12677 + }, + { + "epoch": 1.3922688337359983, + "grad_norm": 2.469224214553833, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7148102521896362, + "num_tokens": 324876579.0, + "step": 12678 + }, + { + "epoch": 1.3923786514386118, + "grad_norm": 2.3426506519317627, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7133126258850098, + "num_tokens": 324899911.0, + "step": 12679 + }, + { + "epoch": 1.3924884691412256, + "grad_norm": 2.2449772357940674, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7402108311653137, + "num_tokens": 324922621.0, + "step": 12680 + }, + { + "epoch": 1.3925982868438394, + "grad_norm": 2.199223041534424, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7309750914573669, + "num_tokens": 324948490.0, + "step": 12681 + }, + { + "epoch": 1.3927081045464529, + "grad_norm": 2.2800405025482178, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7409117221832275, + "num_tokens": 324972702.0, + "step": 12682 + }, + { + "epoch": 1.3928179222490664, + "grad_norm": 2.4127628803253174, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7363878488540649, + "num_tokens": 324993513.0, + "step": 12683 + }, + { + "epoch": 1.3929277399516802, + "grad_norm": 2.2123799324035645, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7238960266113281, + "num_tokens": 325018567.0, + "step": 12684 + }, + { + "epoch": 1.393037557654294, + "grad_norm": 2.2949624061584473, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7530330419540405, + "num_tokens": 325040552.0, + "step": 12685 + }, + { + "epoch": 1.3931473753569075, + "grad_norm": 2.0636000633239746, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7071936130523682, + "num_tokens": 325071351.0, + "step": 12686 + }, + { + "epoch": 1.3932571930595212, + "grad_norm": 2.2814273834228516, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7274985313415527, + "num_tokens": 325098255.0, + "step": 12687 + }, + { + "epoch": 1.3933670107621348, + "grad_norm": 2.7113780975341797, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7385885119438171, + "num_tokens": 325116809.0, + "step": 12688 + }, + { + "epoch": 1.3934768284647485, + "grad_norm": 2.159364700317383, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7118816375732422, + "num_tokens": 325145607.0, + "step": 12689 + }, + { + "epoch": 1.3935866461673623, + "grad_norm": 2.50795841217041, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7328248023986816, + "num_tokens": 325166996.0, + "step": 12690 + }, + { + "epoch": 1.3936964638699758, + "grad_norm": 2.2483115196228027, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7133112549781799, + "num_tokens": 325193656.0, + "step": 12691 + }, + { + "epoch": 1.3938062815725896, + "grad_norm": 2.2706425189971924, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7256577014923096, + "num_tokens": 325217415.0, + "step": 12692 + }, + { + "epoch": 1.393916099275203, + "grad_norm": 2.237015962600708, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7392284870147705, + "num_tokens": 325242624.0, + "step": 12693 + }, + { + "epoch": 1.3940259169778169, + "grad_norm": 2.137990713119507, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7312066555023193, + "num_tokens": 325269681.0, + "step": 12694 + }, + { + "epoch": 1.3941357346804306, + "grad_norm": 2.202638864517212, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7245033979415894, + "num_tokens": 325294936.0, + "step": 12695 + }, + { + "epoch": 1.3942455523830442, + "grad_norm": 2.1770622730255127, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7211326956748962, + "num_tokens": 325321334.0, + "step": 12696 + }, + { + "epoch": 1.3943553700856577, + "grad_norm": 2.6267476081848145, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7456970810890198, + "num_tokens": 325341743.0, + "step": 12697 + }, + { + "epoch": 1.3944651877882714, + "grad_norm": 2.157820463180542, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7159410715103149, + "num_tokens": 325368705.0, + "step": 12698 + }, + { + "epoch": 1.3945750054908852, + "grad_norm": 2.718352794647217, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7251824140548706, + "num_tokens": 325387343.0, + "step": 12699 + }, + { + "epoch": 1.3946848231934987, + "grad_norm": 2.342116117477417, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7241570353507996, + "num_tokens": 325410535.0, + "step": 12700 + }, + { + "epoch": 1.3947946408961125, + "grad_norm": 2.7177977561950684, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7351292967796326, + "num_tokens": 325430161.0, + "step": 12701 + }, + { + "epoch": 1.394904458598726, + "grad_norm": 2.8355557918548584, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7337844371795654, + "num_tokens": 325448917.0, + "step": 12702 + }, + { + "epoch": 1.3950142763013398, + "grad_norm": 2.00673770904541, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6840290427207947, + "num_tokens": 325481282.0, + "step": 12703 + }, + { + "epoch": 1.3951240940039535, + "grad_norm": 2.0881450176239014, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7085773944854736, + "num_tokens": 325512355.0, + "step": 12704 + }, + { + "epoch": 1.395233911706567, + "grad_norm": 2.335662841796875, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7457431554794312, + "num_tokens": 325536519.0, + "step": 12705 + }, + { + "epoch": 1.3953437294091808, + "grad_norm": 1.96316397190094, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7378052473068237, + "num_tokens": 325564732.0, + "step": 12706 + }, + { + "epoch": 1.3954535471117944, + "grad_norm": 2.087268590927124, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6988476514816284, + "num_tokens": 325592997.0, + "step": 12707 + }, + { + "epoch": 1.3955633648144081, + "grad_norm": 2.3023853302001953, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7113622426986694, + "num_tokens": 325620286.0, + "step": 12708 + }, + { + "epoch": 1.3956731825170219, + "grad_norm": 2.130876064300537, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7037860155105591, + "num_tokens": 325648772.0, + "step": 12709 + }, + { + "epoch": 1.3957830002196354, + "grad_norm": 2.1842310428619385, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7323175668716431, + "num_tokens": 325675767.0, + "step": 12710 + }, + { + "epoch": 1.395892817922249, + "grad_norm": 2.30627179145813, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7210032939910889, + "num_tokens": 325701173.0, + "step": 12711 + }, + { + "epoch": 1.3960026356248627, + "grad_norm": 2.1151089668273926, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6946388483047485, + "num_tokens": 325730662.0, + "step": 12712 + }, + { + "epoch": 1.3961124533274765, + "grad_norm": 2.0145461559295654, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.703361451625824, + "num_tokens": 325760255.0, + "step": 12713 + }, + { + "epoch": 1.39622227103009, + "grad_norm": 2.082291841506958, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7155445218086243, + "num_tokens": 325790856.0, + "step": 12714 + }, + { + "epoch": 1.3963320887327038, + "grad_norm": 2.028238534927368, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7020397782325745, + "num_tokens": 325822667.0, + "step": 12715 + }, + { + "epoch": 1.3964419064353173, + "grad_norm": 2.5579984188079834, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7393687963485718, + "num_tokens": 325842498.0, + "step": 12716 + }, + { + "epoch": 1.396551724137931, + "grad_norm": 2.3191356658935547, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7167831659317017, + "num_tokens": 325868328.0, + "step": 12717 + }, + { + "epoch": 1.3966615418405448, + "grad_norm": 2.2010960578918457, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.6991188526153564, + "num_tokens": 325897295.0, + "step": 12718 + }, + { + "epoch": 1.3967713595431583, + "grad_norm": 2.1291635036468506, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7034292221069336, + "num_tokens": 325925304.0, + "step": 12719 + }, + { + "epoch": 1.396881177245772, + "grad_norm": 2.212223768234253, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7164263725280762, + "num_tokens": 325951259.0, + "step": 12720 + }, + { + "epoch": 1.3969909949483856, + "grad_norm": 1.9520360231399536, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7238832712173462, + "num_tokens": 325982599.0, + "step": 12721 + }, + { + "epoch": 1.3971008126509994, + "grad_norm": 2.3023087978363037, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7349621057510376, + "num_tokens": 326008398.0, + "step": 12722 + }, + { + "epoch": 1.397210630353613, + "grad_norm": 2.2697935104370117, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7219086289405823, + "num_tokens": 326033716.0, + "step": 12723 + }, + { + "epoch": 1.3973204480562267, + "grad_norm": 2.0266473293304443, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7190696597099304, + "num_tokens": 326062925.0, + "step": 12724 + }, + { + "epoch": 1.3974302657588402, + "grad_norm": 2.1309595108032227, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7155669927597046, + "num_tokens": 326091031.0, + "step": 12725 + }, + { + "epoch": 1.397540083461454, + "grad_norm": 2.0765228271484375, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.711540162563324, + "num_tokens": 326121451.0, + "step": 12726 + }, + { + "epoch": 1.3976499011640677, + "grad_norm": 2.282796859741211, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7128268480300903, + "num_tokens": 326147904.0, + "step": 12727 + }, + { + "epoch": 1.3977597188666813, + "grad_norm": 2.27418851852417, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6862095594406128, + "num_tokens": 326172809.0, + "step": 12728 + }, + { + "epoch": 1.397869536569295, + "grad_norm": 2.0730934143066406, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6972284317016602, + "num_tokens": 326201856.0, + "step": 12729 + }, + { + "epoch": 1.3979793542719086, + "grad_norm": 2.0785329341888428, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7076548337936401, + "num_tokens": 326232797.0, + "step": 12730 + }, + { + "epoch": 1.3980891719745223, + "grad_norm": 2.6625399589538574, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7316069602966309, + "num_tokens": 326252267.0, + "step": 12731 + }, + { + "epoch": 1.398198989677136, + "grad_norm": 2.3265509605407715, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7369803190231323, + "num_tokens": 326274588.0, + "step": 12732 + }, + { + "epoch": 1.3983088073797496, + "grad_norm": 2.240232467651367, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7159777879714966, + "num_tokens": 326300638.0, + "step": 12733 + }, + { + "epoch": 1.3984186250823631, + "grad_norm": 2.1943318843841553, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7381731271743774, + "num_tokens": 326326195.0, + "step": 12734 + }, + { + "epoch": 1.398528442784977, + "grad_norm": 2.4283440113067627, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7054153680801392, + "num_tokens": 326347996.0, + "step": 12735 + }, + { + "epoch": 1.3986382604875907, + "grad_norm": 2.1602649688720703, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6905278563499451, + "num_tokens": 326375824.0, + "step": 12736 + }, + { + "epoch": 1.3987480781902042, + "grad_norm": 2.2740566730499268, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7151889801025391, + "num_tokens": 326401120.0, + "step": 12737 + }, + { + "epoch": 1.398857895892818, + "grad_norm": 2.2192399501800537, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7172025442123413, + "num_tokens": 326427062.0, + "step": 12738 + }, + { + "epoch": 1.3989677135954315, + "grad_norm": 2.086857557296753, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7083687782287598, + "num_tokens": 326454736.0, + "step": 12739 + }, + { + "epoch": 1.3990775312980452, + "grad_norm": 2.472397804260254, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7208940982818604, + "num_tokens": 326477102.0, + "step": 12740 + }, + { + "epoch": 1.399187349000659, + "grad_norm": 2.361870050430298, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7089933156967163, + "num_tokens": 326501135.0, + "step": 12741 + }, + { + "epoch": 1.3992971667032725, + "grad_norm": 2.2312278747558594, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7052221894264221, + "num_tokens": 326526283.0, + "step": 12742 + }, + { + "epoch": 1.3994069844058863, + "grad_norm": 2.369668960571289, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7209019660949707, + "num_tokens": 326550339.0, + "step": 12743 + }, + { + "epoch": 1.3995168021084998, + "grad_norm": 2.3358206748962402, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7111610174179077, + "num_tokens": 326575548.0, + "step": 12744 + }, + { + "epoch": 1.3996266198111136, + "grad_norm": 2.3165440559387207, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.720975399017334, + "num_tokens": 326600839.0, + "step": 12745 + }, + { + "epoch": 1.3997364375137273, + "grad_norm": 2.4278743267059326, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7078050374984741, + "num_tokens": 326623541.0, + "step": 12746 + }, + { + "epoch": 1.3998462552163409, + "grad_norm": 1.9194544553756714, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7164609432220459, + "num_tokens": 326656143.0, + "step": 12747 + }, + { + "epoch": 1.3999560729189544, + "grad_norm": 2.119602918624878, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6851553916931152, + "num_tokens": 326683896.0, + "step": 12748 + }, + { + "epoch": 1.4000658906215682, + "grad_norm": 2.3163585662841797, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7241106033325195, + "num_tokens": 326708244.0, + "step": 12749 + }, + { + "epoch": 1.400175708324182, + "grad_norm": 2.341947555541992, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7227505445480347, + "num_tokens": 326730791.0, + "step": 12750 + }, + { + "epoch": 1.4002855260267955, + "grad_norm": 2.2857882976531982, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7155270576477051, + "num_tokens": 326754380.0, + "step": 12751 + }, + { + "epoch": 1.4003953437294092, + "grad_norm": 2.2866435050964355, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7391811609268188, + "num_tokens": 326777880.0, + "step": 12752 + }, + { + "epoch": 1.4005051614320227, + "grad_norm": 2.048736095428467, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7328104972839355, + "num_tokens": 326806138.0, + "step": 12753 + }, + { + "epoch": 1.4006149791346365, + "grad_norm": 2.1229195594787598, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7125403881072998, + "num_tokens": 326831742.0, + "step": 12754 + }, + { + "epoch": 1.4007247968372503, + "grad_norm": 2.305774688720703, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7439682483673096, + "num_tokens": 326854711.0, + "step": 12755 + }, + { + "epoch": 1.4008346145398638, + "grad_norm": 2.0961716175079346, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6944537162780762, + "num_tokens": 326885924.0, + "step": 12756 + }, + { + "epoch": 1.4009444322424776, + "grad_norm": 2.68359112739563, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7134039402008057, + "num_tokens": 326905342.0, + "step": 12757 + }, + { + "epoch": 1.401054249945091, + "grad_norm": 2.2458529472351074, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6969444155693054, + "num_tokens": 326932329.0, + "step": 12758 + }, + { + "epoch": 1.4011640676477048, + "grad_norm": 2.291452646255493, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7231366634368896, + "num_tokens": 326957064.0, + "step": 12759 + }, + { + "epoch": 1.4012738853503186, + "grad_norm": 2.20491886138916, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7468937635421753, + "num_tokens": 326982250.0, + "step": 12760 + }, + { + "epoch": 1.4013837030529321, + "grad_norm": 2.1751019954681396, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7057687640190125, + "num_tokens": 327010713.0, + "step": 12761 + }, + { + "epoch": 1.4014935207555457, + "grad_norm": 2.3913650512695312, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7174835205078125, + "num_tokens": 327032491.0, + "step": 12762 + }, + { + "epoch": 1.4016033384581594, + "grad_norm": 2.187042236328125, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6983370780944824, + "num_tokens": 327060277.0, + "step": 12763 + }, + { + "epoch": 1.4017131561607732, + "grad_norm": 2.445535659790039, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7125098705291748, + "num_tokens": 327082688.0, + "step": 12764 + }, + { + "epoch": 1.4018229738633867, + "grad_norm": 2.1686723232269287, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7228720784187317, + "num_tokens": 327111403.0, + "step": 12765 + }, + { + "epoch": 1.4019327915660005, + "grad_norm": 2.5010128021240234, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7262411117553711, + "num_tokens": 327131907.0, + "step": 12766 + }, + { + "epoch": 1.402042609268614, + "grad_norm": 2.3059089183807373, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7422634959220886, + "num_tokens": 327153922.0, + "step": 12767 + }, + { + "epoch": 1.4021524269712278, + "grad_norm": 2.3284506797790527, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7236461639404297, + "num_tokens": 327177243.0, + "step": 12768 + }, + { + "epoch": 1.4022622446738415, + "grad_norm": 2.1608200073242188, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7213988900184631, + "num_tokens": 327202757.0, + "step": 12769 + }, + { + "epoch": 1.402372062376455, + "grad_norm": 2.2692153453826904, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7236684560775757, + "num_tokens": 327227005.0, + "step": 12770 + }, + { + "epoch": 1.4024818800790688, + "grad_norm": 2.206003427505493, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7018495202064514, + "num_tokens": 327253493.0, + "step": 12771 + }, + { + "epoch": 1.4025916977816824, + "grad_norm": 2.3823347091674805, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7367942333221436, + "num_tokens": 327276726.0, + "step": 12772 + }, + { + "epoch": 1.4027015154842961, + "grad_norm": 2.311121702194214, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7322934865951538, + "num_tokens": 327301509.0, + "step": 12773 + }, + { + "epoch": 1.4028113331869096, + "grad_norm": 1.994694471359253, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7335413694381714, + "num_tokens": 327330549.0, + "step": 12774 + }, + { + "epoch": 1.4029211508895234, + "grad_norm": 2.232485055923462, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7308146953582764, + "num_tokens": 327355257.0, + "step": 12775 + }, + { + "epoch": 1.403030968592137, + "grad_norm": 2.3025858402252197, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7098702788352966, + "num_tokens": 327380275.0, + "step": 12776 + }, + { + "epoch": 1.4031407862947507, + "grad_norm": 2.1672792434692383, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7457950115203857, + "num_tokens": 327406009.0, + "step": 12777 + }, + { + "epoch": 1.4032506039973645, + "grad_norm": 2.188507080078125, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.716731071472168, + "num_tokens": 327432601.0, + "step": 12778 + }, + { + "epoch": 1.403360421699978, + "grad_norm": 2.014063835144043, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7037665247917175, + "num_tokens": 327463504.0, + "step": 12779 + }, + { + "epoch": 1.4034702394025917, + "grad_norm": 1.9236830472946167, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6883090734481812, + "num_tokens": 327496837.0, + "step": 12780 + }, + { + "epoch": 1.4035800571052053, + "grad_norm": 1.9779614210128784, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7199836373329163, + "num_tokens": 327527424.0, + "step": 12781 + }, + { + "epoch": 1.403689874807819, + "grad_norm": 2.007939577102661, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6966142058372498, + "num_tokens": 327557301.0, + "step": 12782 + }, + { + "epoch": 1.4037996925104328, + "grad_norm": 2.2993013858795166, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7139876484870911, + "num_tokens": 327582769.0, + "step": 12783 + }, + { + "epoch": 1.4039095102130463, + "grad_norm": 2.1821393966674805, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7001781463623047, + "num_tokens": 327611055.0, + "step": 12784 + }, + { + "epoch": 1.40401932791566, + "grad_norm": 2.110220193862915, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.6992220282554626, + "num_tokens": 327638573.0, + "step": 12785 + }, + { + "epoch": 1.4041291456182736, + "grad_norm": 2.0134623050689697, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.6992435455322266, + "num_tokens": 327669805.0, + "step": 12786 + }, + { + "epoch": 1.4042389633208874, + "grad_norm": 2.2612409591674805, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7167391777038574, + "num_tokens": 327694425.0, + "step": 12787 + }, + { + "epoch": 1.404348781023501, + "grad_norm": 2.421402931213379, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7388810515403748, + "num_tokens": 327716437.0, + "step": 12788 + }, + { + "epoch": 1.4044585987261147, + "grad_norm": 2.071716070175171, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7187231779098511, + "num_tokens": 327743732.0, + "step": 12789 + }, + { + "epoch": 1.4045684164287282, + "grad_norm": 2.0987329483032227, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7296971678733826, + "num_tokens": 327772957.0, + "step": 12790 + }, + { + "epoch": 1.404678234131342, + "grad_norm": 2.289764404296875, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7133962512016296, + "num_tokens": 327796450.0, + "step": 12791 + }, + { + "epoch": 1.4047880518339557, + "grad_norm": 2.1091625690460205, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7135976552963257, + "num_tokens": 327824855.0, + "step": 12792 + }, + { + "epoch": 1.4048978695365693, + "grad_norm": 2.2768325805664062, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7216528058052063, + "num_tokens": 327849719.0, + "step": 12793 + }, + { + "epoch": 1.405007687239183, + "grad_norm": 2.3481128215789795, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7064383625984192, + "num_tokens": 327873174.0, + "step": 12794 + }, + { + "epoch": 1.4051175049417965, + "grad_norm": 2.2373650074005127, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7150301933288574, + "num_tokens": 327899807.0, + "step": 12795 + }, + { + "epoch": 1.4052273226444103, + "grad_norm": 2.252781629562378, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7216324210166931, + "num_tokens": 327925327.0, + "step": 12796 + }, + { + "epoch": 1.405337140347024, + "grad_norm": 2.5352065563201904, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7279497385025024, + "num_tokens": 327945457.0, + "step": 12797 + }, + { + "epoch": 1.4054469580496376, + "grad_norm": 2.53711199760437, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7203285694122314, + "num_tokens": 327966459.0, + "step": 12798 + }, + { + "epoch": 1.4055567757522511, + "grad_norm": 2.2207469940185547, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6981911063194275, + "num_tokens": 327995110.0, + "step": 12799 + }, + { + "epoch": 1.4056665934548649, + "grad_norm": 2.4517314434051514, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7278906106948853, + "num_tokens": 328016946.0, + "step": 12800 + }, + { + "epoch": 1.4057764111574786, + "grad_norm": 2.3020036220550537, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.71504807472229, + "num_tokens": 328042475.0, + "step": 12801 + }, + { + "epoch": 1.4058862288600922, + "grad_norm": 2.178600549697876, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7072794437408447, + "num_tokens": 328070822.0, + "step": 12802 + }, + { + "epoch": 1.405996046562706, + "grad_norm": 2.4935355186462402, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7123409509658813, + "num_tokens": 328093065.0, + "step": 12803 + }, + { + "epoch": 1.4061058642653195, + "grad_norm": 2.176743268966675, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7148056030273438, + "num_tokens": 328120991.0, + "step": 12804 + }, + { + "epoch": 1.4062156819679332, + "grad_norm": 1.976735234260559, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7252902388572693, + "num_tokens": 328149568.0, + "step": 12805 + }, + { + "epoch": 1.406325499670547, + "grad_norm": 2.4940359592437744, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7320387363433838, + "num_tokens": 328171185.0, + "step": 12806 + }, + { + "epoch": 1.4064353173731605, + "grad_norm": 2.611733913421631, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7373712062835693, + "num_tokens": 328191229.0, + "step": 12807 + }, + { + "epoch": 1.4065451350757743, + "grad_norm": 2.0962142944335938, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7198828458786011, + "num_tokens": 328218111.0, + "step": 12808 + }, + { + "epoch": 1.4066549527783878, + "grad_norm": 2.259810209274292, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7216619253158569, + "num_tokens": 328242520.0, + "step": 12809 + }, + { + "epoch": 1.4067647704810016, + "grad_norm": 2.573127031326294, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7420228719711304, + "num_tokens": 328263084.0, + "step": 12810 + }, + { + "epoch": 1.4068745881836153, + "grad_norm": 2.247393846511841, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.6944763660430908, + "num_tokens": 328290534.0, + "step": 12811 + }, + { + "epoch": 1.4069844058862289, + "grad_norm": 2.5200400352478027, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.723753035068512, + "num_tokens": 328313270.0, + "step": 12812 + }, + { + "epoch": 1.4070942235888424, + "grad_norm": 2.355337381362915, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7187103033065796, + "num_tokens": 328336355.0, + "step": 12813 + }, + { + "epoch": 1.4072040412914562, + "grad_norm": 2.3009848594665527, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.709530770778656, + "num_tokens": 328363458.0, + "step": 12814 + }, + { + "epoch": 1.40731385899407, + "grad_norm": 2.4139089584350586, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7243598699569702, + "num_tokens": 328387181.0, + "step": 12815 + }, + { + "epoch": 1.4074236766966834, + "grad_norm": 2.1547980308532715, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7154726982116699, + "num_tokens": 328415291.0, + "step": 12816 + }, + { + "epoch": 1.4075334943992972, + "grad_norm": 2.381943941116333, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7163139581680298, + "num_tokens": 328439048.0, + "step": 12817 + }, + { + "epoch": 1.4076433121019107, + "grad_norm": 2.094064712524414, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7220135927200317, + "num_tokens": 328466686.0, + "step": 12818 + }, + { + "epoch": 1.4077531298045245, + "grad_norm": 2.0526175498962402, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7040693759918213, + "num_tokens": 328497734.0, + "step": 12819 + }, + { + "epoch": 1.4078629475071383, + "grad_norm": 2.4249634742736816, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7234650254249573, + "num_tokens": 328519474.0, + "step": 12820 + }, + { + "epoch": 1.4079727652097518, + "grad_norm": 2.316213607788086, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7385541200637817, + "num_tokens": 328541188.0, + "step": 12821 + }, + { + "epoch": 1.4080825829123655, + "grad_norm": 2.1331586837768555, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7309669256210327, + "num_tokens": 328566274.0, + "step": 12822 + }, + { + "epoch": 1.408192400614979, + "grad_norm": 2.2487103939056396, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7451717853546143, + "num_tokens": 328589338.0, + "step": 12823 + }, + { + "epoch": 1.4083022183175928, + "grad_norm": 2.551846981048584, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7414501309394836, + "num_tokens": 328610802.0, + "step": 12824 + }, + { + "epoch": 1.4084120360202066, + "grad_norm": 2.760171890258789, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7316135168075562, + "num_tokens": 328628724.0, + "step": 12825 + }, + { + "epoch": 1.4085218537228201, + "grad_norm": 2.191495180130005, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7195403575897217, + "num_tokens": 328655192.0, + "step": 12826 + }, + { + "epoch": 1.4086316714254337, + "grad_norm": 2.4492383003234863, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.711296796798706, + "num_tokens": 328677842.0, + "step": 12827 + }, + { + "epoch": 1.4087414891280474, + "grad_norm": 2.1404025554656982, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7168271541595459, + "num_tokens": 328706096.0, + "step": 12828 + }, + { + "epoch": 1.4088513068306612, + "grad_norm": 1.963774561882019, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7089481353759766, + "num_tokens": 328740102.0, + "step": 12829 + }, + { + "epoch": 1.4089611245332747, + "grad_norm": 2.365448474884033, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7385807633399963, + "num_tokens": 328763407.0, + "step": 12830 + }, + { + "epoch": 1.4090709422358885, + "grad_norm": 2.2267725467681885, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7064138650894165, + "num_tokens": 328789506.0, + "step": 12831 + }, + { + "epoch": 1.409180759938502, + "grad_norm": 2.321608781814575, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7187607884407043, + "num_tokens": 328813475.0, + "step": 12832 + }, + { + "epoch": 1.4092905776411158, + "grad_norm": 2.0393543243408203, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7135079503059387, + "num_tokens": 328842872.0, + "step": 12833 + }, + { + "epoch": 1.4094003953437295, + "grad_norm": 2.5921783447265625, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7239913940429688, + "num_tokens": 328862166.0, + "step": 12834 + }, + { + "epoch": 1.409510213046343, + "grad_norm": 2.3196969032287598, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7327460050582886, + "num_tokens": 328885898.0, + "step": 12835 + }, + { + "epoch": 1.4096200307489568, + "grad_norm": 2.160994052886963, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7296353578567505, + "num_tokens": 328911851.0, + "step": 12836 + }, + { + "epoch": 1.4097298484515703, + "grad_norm": 2.4526689052581787, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7206741571426392, + "num_tokens": 328933759.0, + "step": 12837 + }, + { + "epoch": 1.409839666154184, + "grad_norm": 2.027615785598755, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.718329906463623, + "num_tokens": 328963761.0, + "step": 12838 + }, + { + "epoch": 1.4099494838567976, + "grad_norm": 2.1856558322906494, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7482342720031738, + "num_tokens": 328988537.0, + "step": 12839 + }, + { + "epoch": 1.4100593015594114, + "grad_norm": 2.0739645957946777, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.7011173963546753, + "num_tokens": 329018978.0, + "step": 12840 + }, + { + "epoch": 1.410169119262025, + "grad_norm": 2.26969838142395, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7298858165740967, + "num_tokens": 329043307.0, + "step": 12841 + }, + { + "epoch": 1.4102789369646387, + "grad_norm": 2.0511016845703125, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7027648687362671, + "num_tokens": 329075202.0, + "step": 12842 + }, + { + "epoch": 1.4103887546672524, + "grad_norm": 2.065769672393799, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6957085132598877, + "num_tokens": 329103498.0, + "step": 12843 + }, + { + "epoch": 1.410498572369866, + "grad_norm": 2.2258944511413574, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.704876184463501, + "num_tokens": 329130103.0, + "step": 12844 + }, + { + "epoch": 1.4106083900724797, + "grad_norm": 2.0408082008361816, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7027681469917297, + "num_tokens": 329159568.0, + "step": 12845 + }, + { + "epoch": 1.4107182077750933, + "grad_norm": 2.248598098754883, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7099008560180664, + "num_tokens": 329186744.0, + "step": 12846 + }, + { + "epoch": 1.410828025477707, + "grad_norm": 2.0275814533233643, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7106290459632874, + "num_tokens": 329216730.0, + "step": 12847 + }, + { + "epoch": 1.4109378431803208, + "grad_norm": 2.2973897457122803, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7030967473983765, + "num_tokens": 329242389.0, + "step": 12848 + }, + { + "epoch": 1.4110476608829343, + "grad_norm": 2.1512465476989746, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.714311420917511, + "num_tokens": 329271612.0, + "step": 12849 + }, + { + "epoch": 1.4111574785855479, + "grad_norm": 2.121394157409668, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6968270540237427, + "num_tokens": 329301096.0, + "step": 12850 + }, + { + "epoch": 1.4112672962881616, + "grad_norm": 2.3184974193573, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7225092649459839, + "num_tokens": 329325556.0, + "step": 12851 + }, + { + "epoch": 1.4113771139907754, + "grad_norm": 2.4884071350097656, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7329445481300354, + "num_tokens": 329349845.0, + "step": 12852 + }, + { + "epoch": 1.411486931693389, + "grad_norm": 2.2095742225646973, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7107576131820679, + "num_tokens": 329378600.0, + "step": 12853 + }, + { + "epoch": 1.4115967493960027, + "grad_norm": 2.5182745456695557, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7395488023757935, + "num_tokens": 329399367.0, + "step": 12854 + }, + { + "epoch": 1.4117065670986162, + "grad_norm": 2.1046762466430664, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7322701215744019, + "num_tokens": 329427486.0, + "step": 12855 + }, + { + "epoch": 1.41181638480123, + "grad_norm": 1.9684852361679077, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7161688804626465, + "num_tokens": 329460513.0, + "step": 12856 + }, + { + "epoch": 1.4119262025038437, + "grad_norm": 1.8958690166473389, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.730344295501709, + "num_tokens": 329494147.0, + "step": 12857 + }, + { + "epoch": 1.4120360202064572, + "grad_norm": 2.4230949878692627, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7340878248214722, + "num_tokens": 329515496.0, + "step": 12858 + }, + { + "epoch": 1.412145837909071, + "grad_norm": 2.3286609649658203, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7203419208526611, + "num_tokens": 329540477.0, + "step": 12859 + }, + { + "epoch": 1.4122556556116845, + "grad_norm": 2.0738565921783447, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6882403492927551, + "num_tokens": 329572388.0, + "step": 12860 + }, + { + "epoch": 1.4123654733142983, + "grad_norm": 2.105780839920044, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.714674711227417, + "num_tokens": 329602141.0, + "step": 12861 + }, + { + "epoch": 1.412475291016912, + "grad_norm": 2.279528856277466, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7092690467834473, + "num_tokens": 329626950.0, + "step": 12862 + }, + { + "epoch": 1.4125851087195256, + "grad_norm": 2.4023356437683105, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7140393853187561, + "num_tokens": 329650451.0, + "step": 12863 + }, + { + "epoch": 1.4126949264221391, + "grad_norm": 2.1294546127319336, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7187913656234741, + "num_tokens": 329678753.0, + "step": 12864 + }, + { + "epoch": 1.4128047441247529, + "grad_norm": 2.4045639038085938, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7232250571250916, + "num_tokens": 329701384.0, + "step": 12865 + }, + { + "epoch": 1.4129145618273666, + "grad_norm": 2.2528023719787598, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7228749990463257, + "num_tokens": 329726305.0, + "step": 12866 + }, + { + "epoch": 1.4130243795299802, + "grad_norm": 2.367093801498413, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.73194420337677, + "num_tokens": 329751024.0, + "step": 12867 + }, + { + "epoch": 1.413134197232594, + "grad_norm": 2.202888250350952, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7317964434623718, + "num_tokens": 329778905.0, + "step": 12868 + }, + { + "epoch": 1.4132440149352075, + "grad_norm": 1.8124020099639893, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7170879244804382, + "num_tokens": 329817838.0, + "step": 12869 + }, + { + "epoch": 1.4133538326378212, + "grad_norm": 1.9647806882858276, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6881630420684814, + "num_tokens": 329852443.0, + "step": 12870 + }, + { + "epoch": 1.413463650340435, + "grad_norm": 2.2707571983337402, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7086259722709656, + "num_tokens": 329878042.0, + "step": 12871 + }, + { + "epoch": 1.4135734680430485, + "grad_norm": 2.5651140213012695, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7155933380126953, + "num_tokens": 329898294.0, + "step": 12872 + }, + { + "epoch": 1.4136832857456623, + "grad_norm": 2.309143543243408, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7251579761505127, + "num_tokens": 329922433.0, + "step": 12873 + }, + { + "epoch": 1.4137931034482758, + "grad_norm": 2.0943336486816406, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7452699542045593, + "num_tokens": 329949927.0, + "step": 12874 + }, + { + "epoch": 1.4139029211508896, + "grad_norm": 2.3806097507476807, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7263064384460449, + "num_tokens": 329972348.0, + "step": 12875 + }, + { + "epoch": 1.4140127388535033, + "grad_norm": 2.3350279331207275, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7315170764923096, + "num_tokens": 329994768.0, + "step": 12876 + }, + { + "epoch": 1.4141225565561168, + "grad_norm": 1.9617103338241577, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7302712202072144, + "num_tokens": 330024301.0, + "step": 12877 + }, + { + "epoch": 1.4142323742587304, + "grad_norm": 2.407707452774048, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7270081043243408, + "num_tokens": 330048387.0, + "step": 12878 + }, + { + "epoch": 1.4143421919613441, + "grad_norm": 2.311112880706787, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7110195755958557, + "num_tokens": 330074454.0, + "step": 12879 + }, + { + "epoch": 1.414452009663958, + "grad_norm": 2.0882866382598877, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7271741628646851, + "num_tokens": 330101861.0, + "step": 12880 + }, + { + "epoch": 1.4145618273665714, + "grad_norm": 1.886885166168213, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7015420794487, + "num_tokens": 330135658.0, + "step": 12881 + }, + { + "epoch": 1.4146716450691852, + "grad_norm": 2.1651480197906494, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7136998772621155, + "num_tokens": 330162400.0, + "step": 12882 + }, + { + "epoch": 1.4147814627717987, + "grad_norm": 2.436018705368042, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7200254797935486, + "num_tokens": 330185843.0, + "step": 12883 + }, + { + "epoch": 1.4148912804744125, + "grad_norm": 1.9577218294143677, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6989057064056396, + "num_tokens": 330219140.0, + "step": 12884 + }, + { + "epoch": 1.4150010981770262, + "grad_norm": 2.3774642944335938, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7346652746200562, + "num_tokens": 330242745.0, + "step": 12885 + }, + { + "epoch": 1.4151109158796398, + "grad_norm": 2.0829641819000244, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7044095993041992, + "num_tokens": 330273732.0, + "step": 12886 + }, + { + "epoch": 1.4152207335822535, + "grad_norm": 2.2839467525482178, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7217416763305664, + "num_tokens": 330299342.0, + "step": 12887 + }, + { + "epoch": 1.415330551284867, + "grad_norm": 2.094790458679199, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7153631448745728, + "num_tokens": 330327504.0, + "step": 12888 + }, + { + "epoch": 1.4154403689874808, + "grad_norm": 2.0719034671783447, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7561361789703369, + "num_tokens": 330355378.0, + "step": 12889 + }, + { + "epoch": 1.4155501866900946, + "grad_norm": 2.2611992359161377, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.732625424861908, + "num_tokens": 330380290.0, + "step": 12890 + }, + { + "epoch": 1.4156600043927081, + "grad_norm": 2.233279228210449, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7251273393630981, + "num_tokens": 330406788.0, + "step": 12891 + }, + { + "epoch": 1.4157698220953217, + "grad_norm": 1.8439048528671265, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.726090133190155, + "num_tokens": 330440974.0, + "step": 12892 + }, + { + "epoch": 1.4158796397979354, + "grad_norm": 2.2935562133789062, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7146384716033936, + "num_tokens": 330464549.0, + "step": 12893 + }, + { + "epoch": 1.4159894575005492, + "grad_norm": 2.1587345600128174, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7121098041534424, + "num_tokens": 330492060.0, + "step": 12894 + }, + { + "epoch": 1.4160992752031627, + "grad_norm": 2.438905715942383, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.710587739944458, + "num_tokens": 330515750.0, + "step": 12895 + }, + { + "epoch": 1.4162090929057765, + "grad_norm": 2.394857406616211, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7211891412734985, + "num_tokens": 330537981.0, + "step": 12896 + }, + { + "epoch": 1.41631891060839, + "grad_norm": 2.080078125, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7035365700721741, + "num_tokens": 330567643.0, + "step": 12897 + }, + { + "epoch": 1.4164287283110037, + "grad_norm": 2.315281629562378, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7331480979919434, + "num_tokens": 330590183.0, + "step": 12898 + }, + { + "epoch": 1.4165385460136175, + "grad_norm": 2.0005834102630615, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7068461179733276, + "num_tokens": 330623531.0, + "step": 12899 + }, + { + "epoch": 1.416648363716231, + "grad_norm": 2.280451774597168, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7022110223770142, + "num_tokens": 330648401.0, + "step": 12900 + }, + { + "epoch": 1.4167581814188448, + "grad_norm": 2.146256685256958, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.721255898475647, + "num_tokens": 330675190.0, + "step": 12901 + }, + { + "epoch": 1.4168679991214583, + "grad_norm": 2.23734450340271, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7064182162284851, + "num_tokens": 330708134.0, + "step": 12902 + }, + { + "epoch": 1.416977816824072, + "grad_norm": 2.165896415710449, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7046133875846863, + "num_tokens": 330735947.0, + "step": 12903 + }, + { + "epoch": 1.4170876345266856, + "grad_norm": 2.427680730819702, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7325842976570129, + "num_tokens": 330756943.0, + "step": 12904 + }, + { + "epoch": 1.4171974522292994, + "grad_norm": 2.1410794258117676, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7209230065345764, + "num_tokens": 330783784.0, + "step": 12905 + }, + { + "epoch": 1.417307269931913, + "grad_norm": 2.2044498920440674, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7245790958404541, + "num_tokens": 330808609.0, + "step": 12906 + }, + { + "epoch": 1.4174170876345267, + "grad_norm": 2.1183629035949707, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.746329665184021, + "num_tokens": 330836806.0, + "step": 12907 + }, + { + "epoch": 1.4175269053371404, + "grad_norm": 2.337660551071167, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.741622805595398, + "num_tokens": 330857737.0, + "step": 12908 + }, + { + "epoch": 1.417636723039754, + "grad_norm": 2.3324081897735596, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.713409423828125, + "num_tokens": 330881610.0, + "step": 12909 + }, + { + "epoch": 1.4177465407423677, + "grad_norm": 1.9317063093185425, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7167987823486328, + "num_tokens": 330915797.0, + "step": 12910 + }, + { + "epoch": 1.4178563584449813, + "grad_norm": 2.1168320178985596, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7185957431793213, + "num_tokens": 330946762.0, + "step": 12911 + }, + { + "epoch": 1.417966176147595, + "grad_norm": 2.2953402996063232, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7098944187164307, + "num_tokens": 330971402.0, + "step": 12912 + }, + { + "epoch": 1.4180759938502088, + "grad_norm": 2.498934030532837, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7160227298736572, + "num_tokens": 330991367.0, + "step": 12913 + }, + { + "epoch": 1.4181858115528223, + "grad_norm": 2.226433038711548, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7123165130615234, + "num_tokens": 331020783.0, + "step": 12914 + }, + { + "epoch": 1.4182956292554358, + "grad_norm": 2.126702070236206, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7082163691520691, + "num_tokens": 331048378.0, + "step": 12915 + }, + { + "epoch": 1.4184054469580496, + "grad_norm": 2.270273447036743, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7233648896217346, + "num_tokens": 331072528.0, + "step": 12916 + }, + { + "epoch": 1.4185152646606634, + "grad_norm": 2.298762321472168, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7316945195198059, + "num_tokens": 331097100.0, + "step": 12917 + }, + { + "epoch": 1.418625082363277, + "grad_norm": 2.287241220474243, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7176671028137207, + "num_tokens": 331123332.0, + "step": 12918 + }, + { + "epoch": 1.4187349000658906, + "grad_norm": 2.1132006645202637, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.723107635974884, + "num_tokens": 331150781.0, + "step": 12919 + }, + { + "epoch": 1.4188447177685042, + "grad_norm": 2.2970592975616455, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7388700246810913, + "num_tokens": 331175712.0, + "step": 12920 + }, + { + "epoch": 1.418954535471118, + "grad_norm": 2.085869550704956, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7100775241851807, + "num_tokens": 331203719.0, + "step": 12921 + }, + { + "epoch": 1.4190643531737317, + "grad_norm": 2.0589914321899414, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7338168621063232, + "num_tokens": 331230720.0, + "step": 12922 + }, + { + "epoch": 1.4191741708763452, + "grad_norm": 2.2707936763763428, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7240889668464661, + "num_tokens": 331255223.0, + "step": 12923 + }, + { + "epoch": 1.419283988578959, + "grad_norm": 2.0757951736450195, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6947693228721619, + "num_tokens": 331286594.0, + "step": 12924 + }, + { + "epoch": 1.4193938062815725, + "grad_norm": 2.113039493560791, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7364461421966553, + "num_tokens": 331312682.0, + "step": 12925 + }, + { + "epoch": 1.4195036239841863, + "grad_norm": 2.3633103370666504, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7214495539665222, + "num_tokens": 331335619.0, + "step": 12926 + }, + { + "epoch": 1.4196134416868, + "grad_norm": 2.062699556350708, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7182536125183105, + "num_tokens": 331364676.0, + "step": 12927 + }, + { + "epoch": 1.4197232593894136, + "grad_norm": 2.149390459060669, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7237467765808105, + "num_tokens": 331391366.0, + "step": 12928 + }, + { + "epoch": 1.419833077092027, + "grad_norm": 2.1765947341918945, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7152857184410095, + "num_tokens": 331417242.0, + "step": 12929 + }, + { + "epoch": 1.4199428947946409, + "grad_norm": 1.9512556791305542, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7201122045516968, + "num_tokens": 331448980.0, + "step": 12930 + }, + { + "epoch": 1.4200527124972546, + "grad_norm": 2.1721858978271484, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7302486896514893, + "num_tokens": 331473895.0, + "step": 12931 + }, + { + "epoch": 1.4201625301998682, + "grad_norm": 2.458913803100586, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7262975573539734, + "num_tokens": 331495023.0, + "step": 12932 + }, + { + "epoch": 1.420272347902482, + "grad_norm": 2.087066411972046, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6819030046463013, + "num_tokens": 331525932.0, + "step": 12933 + }, + { + "epoch": 1.4203821656050954, + "grad_norm": 2.213334083557129, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7150738835334778, + "num_tokens": 331553098.0, + "step": 12934 + }, + { + "epoch": 1.4204919833077092, + "grad_norm": 2.4342191219329834, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7167730331420898, + "num_tokens": 331575160.0, + "step": 12935 + }, + { + "epoch": 1.420601801010323, + "grad_norm": 2.4223437309265137, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7342685461044312, + "num_tokens": 331596830.0, + "step": 12936 + }, + { + "epoch": 1.4207116187129365, + "grad_norm": 2.2204341888427734, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7127376794815063, + "num_tokens": 331626372.0, + "step": 12937 + }, + { + "epoch": 1.4208214364155503, + "grad_norm": 2.9550554752349854, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7127806544303894, + "num_tokens": 331643992.0, + "step": 12938 + }, + { + "epoch": 1.4209312541181638, + "grad_norm": 2.5951905250549316, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7134870290756226, + "num_tokens": 331665623.0, + "step": 12939 + }, + { + "epoch": 1.4210410718207775, + "grad_norm": 2.147315263748169, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.704806387424469, + "num_tokens": 331692066.0, + "step": 12940 + }, + { + "epoch": 1.4211508895233913, + "grad_norm": 2.0444726943969727, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.6972932815551758, + "num_tokens": 331723918.0, + "step": 12941 + }, + { + "epoch": 1.4212607072260048, + "grad_norm": 2.3025219440460205, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7171858549118042, + "num_tokens": 331748358.0, + "step": 12942 + }, + { + "epoch": 1.4213705249286184, + "grad_norm": 2.2149648666381836, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.726453423500061, + "num_tokens": 331775447.0, + "step": 12943 + }, + { + "epoch": 1.4214803426312321, + "grad_norm": 2.1070783138275146, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7405281066894531, + "num_tokens": 331802845.0, + "step": 12944 + }, + { + "epoch": 1.4215901603338459, + "grad_norm": 2.3968000411987305, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7572619318962097, + "num_tokens": 331824955.0, + "step": 12945 + }, + { + "epoch": 1.4216999780364594, + "grad_norm": 2.2866580486297607, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6881216764450073, + "num_tokens": 331851660.0, + "step": 12946 + }, + { + "epoch": 1.4218097957390732, + "grad_norm": 2.3984014987945557, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7127094268798828, + "num_tokens": 331875541.0, + "step": 12947 + }, + { + "epoch": 1.4219196134416867, + "grad_norm": 2.420919895172119, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7358014583587646, + "num_tokens": 331897675.0, + "step": 12948 + }, + { + "epoch": 1.4220294311443005, + "grad_norm": 2.2875261306762695, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7070698738098145, + "num_tokens": 331924877.0, + "step": 12949 + }, + { + "epoch": 1.4221392488469142, + "grad_norm": 2.4853529930114746, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7030661106109619, + "num_tokens": 331947592.0, + "step": 12950 + }, + { + "epoch": 1.4222490665495278, + "grad_norm": 2.1779885292053223, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7230230569839478, + "num_tokens": 331972242.0, + "step": 12951 + }, + { + "epoch": 1.4223588842521415, + "grad_norm": 2.474602699279785, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7264647483825684, + "num_tokens": 331992558.0, + "step": 12952 + }, + { + "epoch": 1.422468701954755, + "grad_norm": 2.3505094051361084, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7139296531677246, + "num_tokens": 332016759.0, + "step": 12953 + }, + { + "epoch": 1.4225785196573688, + "grad_norm": 2.6580090522766113, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7415410280227661, + "num_tokens": 332034639.0, + "step": 12954 + }, + { + "epoch": 1.4226883373599823, + "grad_norm": 2.446871280670166, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7071225643157959, + "num_tokens": 332057113.0, + "step": 12955 + }, + { + "epoch": 1.422798155062596, + "grad_norm": 2.2650113105773926, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7143052220344543, + "num_tokens": 332085002.0, + "step": 12956 + }, + { + "epoch": 1.4229079727652096, + "grad_norm": 2.300114870071411, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7288526296615601, + "num_tokens": 332109478.0, + "step": 12957 + }, + { + "epoch": 1.4230177904678234, + "grad_norm": 2.2856414318084717, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7052481174468994, + "num_tokens": 332134402.0, + "step": 12958 + }, + { + "epoch": 1.4231276081704372, + "grad_norm": 2.391270637512207, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7244099378585815, + "num_tokens": 332157664.0, + "step": 12959 + }, + { + "epoch": 1.4232374258730507, + "grad_norm": 2.476468324661255, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7148251533508301, + "num_tokens": 332179664.0, + "step": 12960 + }, + { + "epoch": 1.4233472435756644, + "grad_norm": 2.187398672103882, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7163519859313965, + "num_tokens": 332206309.0, + "step": 12961 + }, + { + "epoch": 1.423457061278278, + "grad_norm": 2.316638469696045, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.728103518486023, + "num_tokens": 332230872.0, + "step": 12962 + }, + { + "epoch": 1.4235668789808917, + "grad_norm": 2.392338752746582, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7323106527328491, + "num_tokens": 332253244.0, + "step": 12963 + }, + { + "epoch": 1.4236766966835055, + "grad_norm": 2.1785130500793457, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.733483076095581, + "num_tokens": 332278633.0, + "step": 12964 + }, + { + "epoch": 1.423786514386119, + "grad_norm": 2.434422731399536, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7116315960884094, + "num_tokens": 332302636.0, + "step": 12965 + }, + { + "epoch": 1.4238963320887328, + "grad_norm": 2.462369441986084, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7430330514907837, + "num_tokens": 332324165.0, + "step": 12966 + }, + { + "epoch": 1.4240061497913463, + "grad_norm": 2.298804998397827, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.723081111907959, + "num_tokens": 332349152.0, + "step": 12967 + }, + { + "epoch": 1.42411596749396, + "grad_norm": 2.1839942932128906, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.713478684425354, + "num_tokens": 332376754.0, + "step": 12968 + }, + { + "epoch": 1.4242257851965736, + "grad_norm": 2.282543659210205, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7149653434753418, + "num_tokens": 332401556.0, + "step": 12969 + }, + { + "epoch": 1.4243356028991874, + "grad_norm": 2.1549158096313477, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7015534043312073, + "num_tokens": 332427845.0, + "step": 12970 + }, + { + "epoch": 1.424445420601801, + "grad_norm": 2.3532166481018066, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7027897834777832, + "num_tokens": 332452171.0, + "step": 12971 + }, + { + "epoch": 1.4245552383044147, + "grad_norm": 2.2975234985351562, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7187628746032715, + "num_tokens": 332475412.0, + "step": 12972 + }, + { + "epoch": 1.4246650560070284, + "grad_norm": 2.452939748764038, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7319416403770447, + "num_tokens": 332497138.0, + "step": 12973 + }, + { + "epoch": 1.424774873709642, + "grad_norm": 1.8541860580444336, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7183222770690918, + "num_tokens": 332530136.0, + "step": 12974 + }, + { + "epoch": 1.4248846914122557, + "grad_norm": 2.492971181869507, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7216611504554749, + "num_tokens": 332554449.0, + "step": 12975 + }, + { + "epoch": 1.4249945091148692, + "grad_norm": 2.5738487243652344, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7054523825645447, + "num_tokens": 332578098.0, + "step": 12976 + }, + { + "epoch": 1.425104326817483, + "grad_norm": 2.3376991748809814, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.700157642364502, + "num_tokens": 332602580.0, + "step": 12977 + }, + { + "epoch": 1.4252141445200968, + "grad_norm": 2.3478260040283203, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7110011577606201, + "num_tokens": 332626614.0, + "step": 12978 + }, + { + "epoch": 1.4253239622227103, + "grad_norm": 2.485957145690918, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7250489592552185, + "num_tokens": 332648204.0, + "step": 12979 + }, + { + "epoch": 1.4254337799253238, + "grad_norm": 2.476280450820923, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7287440896034241, + "num_tokens": 332670079.0, + "step": 12980 + }, + { + "epoch": 1.4255435976279376, + "grad_norm": 2.465620517730713, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.714230477809906, + "num_tokens": 332691444.0, + "step": 12981 + }, + { + "epoch": 1.4256534153305513, + "grad_norm": 2.602062225341797, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7338844537734985, + "num_tokens": 332712113.0, + "step": 12982 + }, + { + "epoch": 1.4257632330331649, + "grad_norm": 2.1925952434539795, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7250633239746094, + "num_tokens": 332738740.0, + "step": 12983 + }, + { + "epoch": 1.4258730507357786, + "grad_norm": 2.2251076698303223, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7240927219390869, + "num_tokens": 332765743.0, + "step": 12984 + }, + { + "epoch": 1.4259828684383922, + "grad_norm": 2.2100608348846436, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6953346729278564, + "num_tokens": 332794090.0, + "step": 12985 + }, + { + "epoch": 1.426092686141006, + "grad_norm": 2.075908660888672, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7093111276626587, + "num_tokens": 332825108.0, + "step": 12986 + }, + { + "epoch": 1.4262025038436197, + "grad_norm": 2.320915699005127, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7351188063621521, + "num_tokens": 332849742.0, + "step": 12987 + }, + { + "epoch": 1.4263123215462332, + "grad_norm": 2.227341890335083, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7241226434707642, + "num_tokens": 332875542.0, + "step": 12988 + }, + { + "epoch": 1.426422139248847, + "grad_norm": 2.555251359939575, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7281574010848999, + "num_tokens": 332896165.0, + "step": 12989 + }, + { + "epoch": 1.4265319569514605, + "grad_norm": 2.078639268875122, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7024201154708862, + "num_tokens": 332925625.0, + "step": 12990 + }, + { + "epoch": 1.4266417746540743, + "grad_norm": 2.3923728466033936, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.6916499733924866, + "num_tokens": 332950706.0, + "step": 12991 + }, + { + "epoch": 1.426751592356688, + "grad_norm": 2.3194379806518555, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7299533486366272, + "num_tokens": 332974794.0, + "step": 12992 + }, + { + "epoch": 1.4268614100593016, + "grad_norm": 2.2582077980041504, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7235645055770874, + "num_tokens": 332999316.0, + "step": 12993 + }, + { + "epoch": 1.426971227761915, + "grad_norm": 2.2076306343078613, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7121370434761047, + "num_tokens": 333024732.0, + "step": 12994 + }, + { + "epoch": 1.4270810454645289, + "grad_norm": 2.605437755584717, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7290868759155273, + "num_tokens": 333043008.0, + "step": 12995 + }, + { + "epoch": 1.4271908631671426, + "grad_norm": 2.3588876724243164, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7256897687911987, + "num_tokens": 333066611.0, + "step": 12996 + }, + { + "epoch": 1.4273006808697561, + "grad_norm": 2.332587957382202, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.6761088371276855, + "num_tokens": 333094194.0, + "step": 12997 + }, + { + "epoch": 1.42741049857237, + "grad_norm": 2.16341233253479, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7185201644897461, + "num_tokens": 333123836.0, + "step": 12998 + }, + { + "epoch": 1.4275203162749834, + "grad_norm": 2.415870428085327, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7058985233306885, + "num_tokens": 333147420.0, + "step": 12999 + }, + { + "epoch": 1.4276301339775972, + "grad_norm": 2.2633235454559326, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7289053201675415, + "num_tokens": 333171417.0, + "step": 13000 + }, + { + "epoch": 1.427739951680211, + "grad_norm": 2.0917837619781494, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7164192199707031, + "num_tokens": 333198940.0, + "step": 13001 + }, + { + "epoch": 1.4278497693828245, + "grad_norm": 2.2334516048431396, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7290512323379517, + "num_tokens": 333222806.0, + "step": 13002 + }, + { + "epoch": 1.4279595870854382, + "grad_norm": 2.2566585540771484, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7299486398696899, + "num_tokens": 333246247.0, + "step": 13003 + }, + { + "epoch": 1.4280694047880518, + "grad_norm": 2.3824996948242188, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7055086493492126, + "num_tokens": 333270649.0, + "step": 13004 + }, + { + "epoch": 1.4281792224906655, + "grad_norm": 2.102050304412842, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7074968218803406, + "num_tokens": 333299157.0, + "step": 13005 + }, + { + "epoch": 1.4282890401932793, + "grad_norm": 2.1931660175323486, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.700095534324646, + "num_tokens": 333326542.0, + "step": 13006 + }, + { + "epoch": 1.4283988578958928, + "grad_norm": 2.0228819847106934, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7120296955108643, + "num_tokens": 333356845.0, + "step": 13007 + }, + { + "epoch": 1.4285086755985064, + "grad_norm": 2.3848419189453125, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7206276655197144, + "num_tokens": 333379217.0, + "step": 13008 + }, + { + "epoch": 1.4286184933011201, + "grad_norm": 2.143155813217163, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7136980295181274, + "num_tokens": 333407250.0, + "step": 13009 + }, + { + "epoch": 1.4287283110037339, + "grad_norm": 2.2416799068450928, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.6989672183990479, + "num_tokens": 333433368.0, + "step": 13010 + }, + { + "epoch": 1.4288381287063474, + "grad_norm": 2.1988937854766846, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.724990963935852, + "num_tokens": 333459638.0, + "step": 13011 + }, + { + "epoch": 1.4289479464089612, + "grad_norm": 2.2221601009368896, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7117221355438232, + "num_tokens": 333484884.0, + "step": 13012 + }, + { + "epoch": 1.4290577641115747, + "grad_norm": 2.13633394241333, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6974729895591736, + "num_tokens": 333514845.0, + "step": 13013 + }, + { + "epoch": 1.4291675818141885, + "grad_norm": 2.339059352874756, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7046339511871338, + "num_tokens": 333540148.0, + "step": 13014 + }, + { + "epoch": 1.4292773995168022, + "grad_norm": 2.3551251888275146, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.713870644569397, + "num_tokens": 333562933.0, + "step": 13015 + }, + { + "epoch": 1.4293872172194158, + "grad_norm": 2.3432929515838623, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7174255847930908, + "num_tokens": 333588240.0, + "step": 13016 + }, + { + "epoch": 1.4294970349220295, + "grad_norm": 2.1702651977539062, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7093199491500854, + "num_tokens": 333617343.0, + "step": 13017 + }, + { + "epoch": 1.429606852624643, + "grad_norm": 2.5363264083862305, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7343149185180664, + "num_tokens": 333637936.0, + "step": 13018 + }, + { + "epoch": 1.4297166703272568, + "grad_norm": 2.3652536869049072, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7435014843940735, + "num_tokens": 333659706.0, + "step": 13019 + }, + { + "epoch": 1.4298264880298703, + "grad_norm": 2.0781755447387695, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7173027992248535, + "num_tokens": 333688697.0, + "step": 13020 + }, + { + "epoch": 1.429936305732484, + "grad_norm": 2.5098116397857666, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.716079831123352, + "num_tokens": 333711513.0, + "step": 13021 + }, + { + "epoch": 1.4300461234350976, + "grad_norm": 2.0720012187957764, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7210511565208435, + "num_tokens": 333741755.0, + "step": 13022 + }, + { + "epoch": 1.4301559411377114, + "grad_norm": 2.3308651447296143, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7039222717285156, + "num_tokens": 333767666.0, + "step": 13023 + }, + { + "epoch": 1.4302657588403251, + "grad_norm": 2.421931266784668, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7544949054718018, + "num_tokens": 333787881.0, + "step": 13024 + }, + { + "epoch": 1.4303755765429387, + "grad_norm": 2.399836778640747, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.722672700881958, + "num_tokens": 333811374.0, + "step": 13025 + }, + { + "epoch": 1.4304853942455524, + "grad_norm": 2.0561513900756836, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.712308943271637, + "num_tokens": 333841220.0, + "step": 13026 + }, + { + "epoch": 1.430595211948166, + "grad_norm": 2.3001585006713867, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7348387241363525, + "num_tokens": 333865141.0, + "step": 13027 + }, + { + "epoch": 1.4307050296507797, + "grad_norm": 2.1662886142730713, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7252624034881592, + "num_tokens": 333892382.0, + "step": 13028 + }, + { + "epoch": 1.4308148473533935, + "grad_norm": 2.022813558578491, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6964053511619568, + "num_tokens": 333924980.0, + "step": 13029 + }, + { + "epoch": 1.430924665056007, + "grad_norm": 2.0596630573272705, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7056683301925659, + "num_tokens": 333954665.0, + "step": 13030 + }, + { + "epoch": 1.4310344827586206, + "grad_norm": 2.352385997772217, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7195001840591431, + "num_tokens": 333977930.0, + "step": 13031 + }, + { + "epoch": 1.4311443004612343, + "grad_norm": 2.0947024822235107, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.733944296836853, + "num_tokens": 334005323.0, + "step": 13032 + }, + { + "epoch": 1.431254118163848, + "grad_norm": 2.1515278816223145, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7058092355728149, + "num_tokens": 334032410.0, + "step": 13033 + }, + { + "epoch": 1.4313639358664616, + "grad_norm": 2.2437570095062256, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7078330516815186, + "num_tokens": 334056274.0, + "step": 13034 + }, + { + "epoch": 1.4314737535690754, + "grad_norm": 2.4892451763153076, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7221403121948242, + "num_tokens": 334079434.0, + "step": 13035 + }, + { + "epoch": 1.431583571271689, + "grad_norm": 2.440246343612671, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7240147590637207, + "num_tokens": 334102483.0, + "step": 13036 + }, + { + "epoch": 1.4316933889743026, + "grad_norm": 2.0720701217651367, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7176991105079651, + "num_tokens": 334131240.0, + "step": 13037 + }, + { + "epoch": 1.4318032066769164, + "grad_norm": 2.098392963409424, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7381857633590698, + "num_tokens": 334159586.0, + "step": 13038 + }, + { + "epoch": 1.43191302437953, + "grad_norm": 2.225486993789673, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6983679533004761, + "num_tokens": 334185883.0, + "step": 13039 + }, + { + "epoch": 1.4320228420821437, + "grad_norm": 2.115116834640503, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7295509576797485, + "num_tokens": 334213739.0, + "step": 13040 + }, + { + "epoch": 1.4321326597847572, + "grad_norm": 2.2209835052490234, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6947207450866699, + "num_tokens": 334242409.0, + "step": 13041 + }, + { + "epoch": 1.432242477487371, + "grad_norm": 2.17213773727417, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7187250256538391, + "num_tokens": 334270032.0, + "step": 13042 + }, + { + "epoch": 1.4323522951899847, + "grad_norm": 2.1400349140167236, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7369229197502136, + "num_tokens": 334295796.0, + "step": 13043 + }, + { + "epoch": 1.4324621128925983, + "grad_norm": 2.340087890625, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7149131894111633, + "num_tokens": 334319538.0, + "step": 13044 + }, + { + "epoch": 1.4325719305952118, + "grad_norm": 2.320106029510498, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7241164445877075, + "num_tokens": 334341892.0, + "step": 13045 + }, + { + "epoch": 1.4326817482978256, + "grad_norm": 1.786565899848938, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7199446558952332, + "num_tokens": 334376921.0, + "step": 13046 + }, + { + "epoch": 1.4327915660004393, + "grad_norm": 2.1097588539123535, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7031658291816711, + "num_tokens": 334404047.0, + "step": 13047 + }, + { + "epoch": 1.4329013837030529, + "grad_norm": 2.3067164421081543, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7334357500076294, + "num_tokens": 334428965.0, + "step": 13048 + }, + { + "epoch": 1.4330112014056666, + "grad_norm": 2.581573009490967, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7160580158233643, + "num_tokens": 334450177.0, + "step": 13049 + }, + { + "epoch": 1.4331210191082802, + "grad_norm": 1.8959295749664307, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7242043018341064, + "num_tokens": 334484643.0, + "step": 13050 + }, + { + "epoch": 1.433230836810894, + "grad_norm": 2.2742581367492676, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7268426418304443, + "num_tokens": 334510174.0, + "step": 13051 + }, + { + "epoch": 1.4333406545135077, + "grad_norm": 2.228428363800049, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7124556303024292, + "num_tokens": 334534895.0, + "step": 13052 + }, + { + "epoch": 1.4334504722161212, + "grad_norm": 2.397733688354492, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7230443954467773, + "num_tokens": 334558494.0, + "step": 13053 + }, + { + "epoch": 1.433560289918735, + "grad_norm": 2.645477056503296, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7383465766906738, + "num_tokens": 334576626.0, + "step": 13054 + }, + { + "epoch": 1.4336701076213485, + "grad_norm": 2.2823596000671387, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7312912940979004, + "num_tokens": 334600730.0, + "step": 13055 + }, + { + "epoch": 1.4337799253239623, + "grad_norm": 2.3217086791992188, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7087501287460327, + "num_tokens": 334625364.0, + "step": 13056 + }, + { + "epoch": 1.433889743026576, + "grad_norm": 2.270327091217041, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7267550826072693, + "num_tokens": 334649627.0, + "step": 13057 + }, + { + "epoch": 1.4339995607291895, + "grad_norm": 2.4042203426361084, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7108832597732544, + "num_tokens": 334672500.0, + "step": 13058 + }, + { + "epoch": 1.434109378431803, + "grad_norm": 2.251185417175293, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7192513346672058, + "num_tokens": 334697973.0, + "step": 13059 + }, + { + "epoch": 1.4342191961344168, + "grad_norm": 2.2578043937683105, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7128753662109375, + "num_tokens": 334727029.0, + "step": 13060 + }, + { + "epoch": 1.4343290138370306, + "grad_norm": 2.443390130996704, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7177492380142212, + "num_tokens": 334751427.0, + "step": 13061 + }, + { + "epoch": 1.4344388315396441, + "grad_norm": 2.475367307662964, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7246352434158325, + "num_tokens": 334775551.0, + "step": 13062 + }, + { + "epoch": 1.434548649242258, + "grad_norm": 2.3691701889038086, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7016624212265015, + "num_tokens": 334801437.0, + "step": 13063 + }, + { + "epoch": 1.4346584669448714, + "grad_norm": 2.4915714263916016, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.734359085559845, + "num_tokens": 334819834.0, + "step": 13064 + }, + { + "epoch": 1.4347682846474852, + "grad_norm": 2.3375320434570312, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.717824399471283, + "num_tokens": 334844192.0, + "step": 13065 + }, + { + "epoch": 1.434878102350099, + "grad_norm": 2.314870595932007, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7233333587646484, + "num_tokens": 334869042.0, + "step": 13066 + }, + { + "epoch": 1.4349879200527125, + "grad_norm": 2.3217110633850098, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7075108885765076, + "num_tokens": 334892896.0, + "step": 13067 + }, + { + "epoch": 1.4350977377553262, + "grad_norm": 2.2613418102264404, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7184876799583435, + "num_tokens": 334917434.0, + "step": 13068 + }, + { + "epoch": 1.4352075554579398, + "grad_norm": 2.843414545059204, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7451988458633423, + "num_tokens": 334936344.0, + "step": 13069 + }, + { + "epoch": 1.4353173731605535, + "grad_norm": 2.057689666748047, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7313725352287292, + "num_tokens": 334965583.0, + "step": 13070 + }, + { + "epoch": 1.4354271908631673, + "grad_norm": 2.105755567550659, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.712691068649292, + "num_tokens": 334995910.0, + "step": 13071 + }, + { + "epoch": 1.4355370085657808, + "grad_norm": 2.4382851123809814, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7207428216934204, + "num_tokens": 335019202.0, + "step": 13072 + }, + { + "epoch": 1.4356468262683943, + "grad_norm": 2.2960681915283203, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7138064503669739, + "num_tokens": 335045202.0, + "step": 13073 + }, + { + "epoch": 1.435756643971008, + "grad_norm": 2.326878786087036, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7340940833091736, + "num_tokens": 335066371.0, + "step": 13074 + }, + { + "epoch": 1.4358664616736219, + "grad_norm": 2.2118523120880127, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7096248865127563, + "num_tokens": 335093208.0, + "step": 13075 + }, + { + "epoch": 1.4359762793762354, + "grad_norm": 1.9806492328643799, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7326226830482483, + "num_tokens": 335124769.0, + "step": 13076 + }, + { + "epoch": 1.4360860970788492, + "grad_norm": 2.0532357692718506, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6894453763961792, + "num_tokens": 335155369.0, + "step": 13077 + }, + { + "epoch": 1.4361959147814627, + "grad_norm": 2.5843117237091064, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7308892011642456, + "num_tokens": 335176499.0, + "step": 13078 + }, + { + "epoch": 1.4363057324840764, + "grad_norm": 2.2044472694396973, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7072946429252625, + "num_tokens": 335202836.0, + "step": 13079 + }, + { + "epoch": 1.4364155501866902, + "grad_norm": 2.5176963806152344, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7213178873062134, + "num_tokens": 335225416.0, + "step": 13080 + }, + { + "epoch": 1.4365253678893037, + "grad_norm": 2.563141107559204, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7290240526199341, + "num_tokens": 335246814.0, + "step": 13081 + }, + { + "epoch": 1.4366351855919175, + "grad_norm": 2.591202974319458, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7240633964538574, + "num_tokens": 335265840.0, + "step": 13082 + }, + { + "epoch": 1.436745003294531, + "grad_norm": 2.3805670738220215, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7146391868591309, + "num_tokens": 335290982.0, + "step": 13083 + }, + { + "epoch": 1.4368548209971448, + "grad_norm": 2.150472640991211, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7077491283416748, + "num_tokens": 335318755.0, + "step": 13084 + }, + { + "epoch": 1.4369646386997583, + "grad_norm": 2.17842173576355, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7265689373016357, + "num_tokens": 335345128.0, + "step": 13085 + }, + { + "epoch": 1.437074456402372, + "grad_norm": 2.3083133697509766, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7240451574325562, + "num_tokens": 335371329.0, + "step": 13086 + }, + { + "epoch": 1.4371842741049856, + "grad_norm": 2.159259796142578, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7108657360076904, + "num_tokens": 335396578.0, + "step": 13087 + }, + { + "epoch": 1.4372940918075994, + "grad_norm": 2.5755159854888916, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7057127356529236, + "num_tokens": 335416499.0, + "step": 13088 + }, + { + "epoch": 1.4374039095102131, + "grad_norm": 2.114504337310791, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7255832552909851, + "num_tokens": 335444035.0, + "step": 13089 + }, + { + "epoch": 1.4375137272128267, + "grad_norm": 2.1378426551818848, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7139080166816711, + "num_tokens": 335473390.0, + "step": 13090 + }, + { + "epoch": 1.4376235449154404, + "grad_norm": 2.4318511486053467, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7397278547286987, + "num_tokens": 335495938.0, + "step": 13091 + }, + { + "epoch": 1.437733362618054, + "grad_norm": 2.4135396480560303, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7151980400085449, + "num_tokens": 335518478.0, + "step": 13092 + }, + { + "epoch": 1.4378431803206677, + "grad_norm": 2.3366148471832275, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7060999870300293, + "num_tokens": 335541411.0, + "step": 13093 + }, + { + "epoch": 1.4379529980232815, + "grad_norm": 2.129276990890503, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7184590101242065, + "num_tokens": 335569903.0, + "step": 13094 + }, + { + "epoch": 1.438062815725895, + "grad_norm": 2.1686744689941406, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7334895133972168, + "num_tokens": 335597044.0, + "step": 13095 + }, + { + "epoch": 1.4381726334285085, + "grad_norm": 2.317861557006836, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7235338687896729, + "num_tokens": 335620726.0, + "step": 13096 + }, + { + "epoch": 1.4382824511311223, + "grad_norm": 2.1246378421783447, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7045458555221558, + "num_tokens": 335649979.0, + "step": 13097 + }, + { + "epoch": 1.438392268833736, + "grad_norm": 2.361562967300415, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7144224643707275, + "num_tokens": 335673592.0, + "step": 13098 + }, + { + "epoch": 1.4385020865363496, + "grad_norm": 2.781067132949829, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.727081835269928, + "num_tokens": 335692933.0, + "step": 13099 + }, + { + "epoch": 1.4386119042389633, + "grad_norm": 2.362994909286499, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.727557361125946, + "num_tokens": 335715820.0, + "step": 13100 + }, + { + "epoch": 1.4387217219415769, + "grad_norm": 2.2393150329589844, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7218614816665649, + "num_tokens": 335740761.0, + "step": 13101 + }, + { + "epoch": 1.4388315396441906, + "grad_norm": 2.1341307163238525, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7195636034011841, + "num_tokens": 335769428.0, + "step": 13102 + }, + { + "epoch": 1.4389413573468044, + "grad_norm": 2.228848695755005, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7173947095870972, + "num_tokens": 335795087.0, + "step": 13103 + }, + { + "epoch": 1.439051175049418, + "grad_norm": 1.9654704332351685, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.736892819404602, + "num_tokens": 335825824.0, + "step": 13104 + }, + { + "epoch": 1.4391609927520317, + "grad_norm": 1.9336882829666138, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7395353317260742, + "num_tokens": 335857560.0, + "step": 13105 + }, + { + "epoch": 1.4392708104546452, + "grad_norm": 1.9904392957687378, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7126114368438721, + "num_tokens": 335890650.0, + "step": 13106 + }, + { + "epoch": 1.439380628157259, + "grad_norm": 2.1271650791168213, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7234249711036682, + "num_tokens": 335919698.0, + "step": 13107 + }, + { + "epoch": 1.4394904458598727, + "grad_norm": 2.0197746753692627, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7016699314117432, + "num_tokens": 335948951.0, + "step": 13108 + }, + { + "epoch": 1.4396002635624863, + "grad_norm": 2.124368190765381, + "learning_rate": 1e-06, + "loss": 1.08, + "mean_token_accuracy": 0.6800880432128906, + "num_tokens": 335977142.0, + "step": 13109 + }, + { + "epoch": 1.4397100812650998, + "grad_norm": 2.1553597450256348, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7070934772491455, + "num_tokens": 336004999.0, + "step": 13110 + }, + { + "epoch": 1.4398198989677136, + "grad_norm": 2.1624491214752197, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7377867102622986, + "num_tokens": 336029263.0, + "step": 13111 + }, + { + "epoch": 1.4399297166703273, + "grad_norm": 2.172389507293701, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7158821821212769, + "num_tokens": 336057425.0, + "step": 13112 + }, + { + "epoch": 1.4400395343729409, + "grad_norm": 2.195462226867676, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.712500274181366, + "num_tokens": 336082601.0, + "step": 13113 + }, + { + "epoch": 1.4401493520755546, + "grad_norm": 2.292634963989258, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7110496163368225, + "num_tokens": 336108772.0, + "step": 13114 + }, + { + "epoch": 1.4402591697781681, + "grad_norm": 2.1072559356689453, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7092642784118652, + "num_tokens": 336136011.0, + "step": 13115 + }, + { + "epoch": 1.440368987480782, + "grad_norm": 2.3659791946411133, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7235569357872009, + "num_tokens": 336159916.0, + "step": 13116 + }, + { + "epoch": 1.4404788051833957, + "grad_norm": 2.396324396133423, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7379721403121948, + "num_tokens": 336181725.0, + "step": 13117 + }, + { + "epoch": 1.4405886228860092, + "grad_norm": 2.2150583267211914, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7200478315353394, + "num_tokens": 336206521.0, + "step": 13118 + }, + { + "epoch": 1.440698440588623, + "grad_norm": 2.2208306789398193, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6957430243492126, + "num_tokens": 336235297.0, + "step": 13119 + }, + { + "epoch": 1.4408082582912365, + "grad_norm": 2.5934431552886963, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7213964462280273, + "num_tokens": 336256374.0, + "step": 13120 + }, + { + "epoch": 1.4409180759938502, + "grad_norm": 2.3553073406219482, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7143687605857849, + "num_tokens": 336280738.0, + "step": 13121 + }, + { + "epoch": 1.441027893696464, + "grad_norm": 2.213146924972534, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7096419334411621, + "num_tokens": 336306781.0, + "step": 13122 + }, + { + "epoch": 1.4411377113990775, + "grad_norm": 2.3869850635528564, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7247347831726074, + "num_tokens": 336328606.0, + "step": 13123 + }, + { + "epoch": 1.441247529101691, + "grad_norm": 2.1069209575653076, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7136778831481934, + "num_tokens": 336356061.0, + "step": 13124 + }, + { + "epoch": 1.4413573468043048, + "grad_norm": 2.3327646255493164, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7024103403091431, + "num_tokens": 336380448.0, + "step": 13125 + }, + { + "epoch": 1.4414671645069186, + "grad_norm": 2.090733528137207, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7150281071662903, + "num_tokens": 336408572.0, + "step": 13126 + }, + { + "epoch": 1.4415769822095321, + "grad_norm": 2.4973814487457275, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7202091217041016, + "num_tokens": 336432643.0, + "step": 13127 + }, + { + "epoch": 1.4416867999121459, + "grad_norm": 2.214315891265869, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7068827748298645, + "num_tokens": 336458553.0, + "step": 13128 + }, + { + "epoch": 1.4417966176147594, + "grad_norm": 2.1502034664154053, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7020117044448853, + "num_tokens": 336489107.0, + "step": 13129 + }, + { + "epoch": 1.4419064353173732, + "grad_norm": 2.3690855503082275, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7338857650756836, + "num_tokens": 336512573.0, + "step": 13130 + }, + { + "epoch": 1.442016253019987, + "grad_norm": 2.1519744396209717, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.71455317735672, + "num_tokens": 336539610.0, + "step": 13131 + }, + { + "epoch": 1.4421260707226005, + "grad_norm": 2.296522617340088, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7238541841506958, + "num_tokens": 336562545.0, + "step": 13132 + }, + { + "epoch": 1.4422358884252142, + "grad_norm": 2.3417184352874756, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7258354425430298, + "num_tokens": 336585721.0, + "step": 13133 + }, + { + "epoch": 1.4423457061278278, + "grad_norm": 2.2208707332611084, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7061094045639038, + "num_tokens": 336612719.0, + "step": 13134 + }, + { + "epoch": 1.4424555238304415, + "grad_norm": 2.2177894115448, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7286413908004761, + "num_tokens": 336637759.0, + "step": 13135 + }, + { + "epoch": 1.442565341533055, + "grad_norm": 2.105397939682007, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7203611135482788, + "num_tokens": 336665704.0, + "step": 13136 + }, + { + "epoch": 1.4426751592356688, + "grad_norm": 2.1455063819885254, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7082667946815491, + "num_tokens": 336692509.0, + "step": 13137 + }, + { + "epoch": 1.4427849769382823, + "grad_norm": 2.4077537059783936, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7255454063415527, + "num_tokens": 336714584.0, + "step": 13138 + }, + { + "epoch": 1.442894794640896, + "grad_norm": 2.171180486679077, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7348254919052124, + "num_tokens": 336741023.0, + "step": 13139 + }, + { + "epoch": 1.4430046123435099, + "grad_norm": 2.3959126472473145, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7206305861473083, + "num_tokens": 336764467.0, + "step": 13140 + }, + { + "epoch": 1.4431144300461234, + "grad_norm": 2.6414945125579834, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7334253191947937, + "num_tokens": 336783499.0, + "step": 13141 + }, + { + "epoch": 1.4432242477487371, + "grad_norm": 2.0056920051574707, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6991861462593079, + "num_tokens": 336817328.0, + "step": 13142 + }, + { + "epoch": 1.4433340654513507, + "grad_norm": 2.5194289684295654, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7249246835708618, + "num_tokens": 336839185.0, + "step": 13143 + }, + { + "epoch": 1.4434438831539644, + "grad_norm": 2.3415286540985107, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7178283929824829, + "num_tokens": 336865040.0, + "step": 13144 + }, + { + "epoch": 1.4435537008565782, + "grad_norm": 1.92408287525177, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7060112953186035, + "num_tokens": 336901531.0, + "step": 13145 + }, + { + "epoch": 1.4436635185591917, + "grad_norm": 2.3099355697631836, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7093421220779419, + "num_tokens": 336927097.0, + "step": 13146 + }, + { + "epoch": 1.4437733362618055, + "grad_norm": 2.544612169265747, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7532448768615723, + "num_tokens": 336946187.0, + "step": 13147 + }, + { + "epoch": 1.443883153964419, + "grad_norm": 2.0614748001098633, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6913047432899475, + "num_tokens": 336974220.0, + "step": 13148 + }, + { + "epoch": 1.4439929716670328, + "grad_norm": 2.085322856903076, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.733708381652832, + "num_tokens": 337002528.0, + "step": 13149 + }, + { + "epoch": 1.4441027893696463, + "grad_norm": 2.3670554161071777, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7060046792030334, + "num_tokens": 337026072.0, + "step": 13150 + }, + { + "epoch": 1.44421260707226, + "grad_norm": 2.103389024734497, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7255258560180664, + "num_tokens": 337053795.0, + "step": 13151 + }, + { + "epoch": 1.4443224247748736, + "grad_norm": 2.1069769859313965, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7230318784713745, + "num_tokens": 337082649.0, + "step": 13152 + }, + { + "epoch": 1.4444322424774874, + "grad_norm": 2.3010077476501465, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7308945655822754, + "num_tokens": 337107395.0, + "step": 13153 + }, + { + "epoch": 1.4445420601801011, + "grad_norm": 2.799462080001831, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7486931085586548, + "num_tokens": 337125528.0, + "step": 13154 + }, + { + "epoch": 1.4446518778827147, + "grad_norm": 2.493762731552124, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7246061563491821, + "num_tokens": 337148174.0, + "step": 13155 + }, + { + "epoch": 1.4447616955853284, + "grad_norm": 2.062227964401245, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6930533647537231, + "num_tokens": 337178753.0, + "step": 13156 + }, + { + "epoch": 1.444871513287942, + "grad_norm": 2.405299425125122, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7353506088256836, + "num_tokens": 337201147.0, + "step": 13157 + }, + { + "epoch": 1.4449813309905557, + "grad_norm": 2.1409831047058105, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7324764728546143, + "num_tokens": 337229671.0, + "step": 13158 + }, + { + "epoch": 1.4450911486931695, + "grad_norm": 2.493337869644165, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7425309419631958, + "num_tokens": 337252841.0, + "step": 13159 + }, + { + "epoch": 1.445200966395783, + "grad_norm": 2.0973446369171143, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7329533100128174, + "num_tokens": 337278858.0, + "step": 13160 + }, + { + "epoch": 1.4453107840983965, + "grad_norm": 2.2497448921203613, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7373077273368835, + "num_tokens": 337303031.0, + "step": 13161 + }, + { + "epoch": 1.4454206018010103, + "grad_norm": 2.456225633621216, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7326111793518066, + "num_tokens": 337325966.0, + "step": 13162 + }, + { + "epoch": 1.445530419503624, + "grad_norm": 2.154967784881592, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7223960161209106, + "num_tokens": 337355422.0, + "step": 13163 + }, + { + "epoch": 1.4456402372062376, + "grad_norm": 2.1314306259155273, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7275525331497192, + "num_tokens": 337385501.0, + "step": 13164 + }, + { + "epoch": 1.4457500549088513, + "grad_norm": 2.169219493865967, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7268328666687012, + "num_tokens": 337412965.0, + "step": 13165 + }, + { + "epoch": 1.4458598726114649, + "grad_norm": 2.612870693206787, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7459548711776733, + "num_tokens": 337432370.0, + "step": 13166 + }, + { + "epoch": 1.4459696903140786, + "grad_norm": 2.3066647052764893, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7255585789680481, + "num_tokens": 337455858.0, + "step": 13167 + }, + { + "epoch": 1.4460795080166924, + "grad_norm": 2.2746973037719727, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7132354974746704, + "num_tokens": 337481489.0, + "step": 13168 + }, + { + "epoch": 1.446189325719306, + "grad_norm": 2.1735329627990723, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.72633957862854, + "num_tokens": 337506682.0, + "step": 13169 + }, + { + "epoch": 1.4462991434219197, + "grad_norm": 2.1293785572052, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7046800851821899, + "num_tokens": 337536640.0, + "step": 13170 + }, + { + "epoch": 1.4464089611245332, + "grad_norm": 2.511899948120117, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7082295417785645, + "num_tokens": 337558047.0, + "step": 13171 + }, + { + "epoch": 1.446518778827147, + "grad_norm": 2.389287233352661, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7273986339569092, + "num_tokens": 337581363.0, + "step": 13172 + }, + { + "epoch": 1.4466285965297607, + "grad_norm": 2.1631343364715576, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7333481907844543, + "num_tokens": 337607476.0, + "step": 13173 + }, + { + "epoch": 1.4467384142323743, + "grad_norm": 2.637097120285034, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7448105812072754, + "num_tokens": 337625509.0, + "step": 13174 + }, + { + "epoch": 1.4468482319349878, + "grad_norm": 2.178086996078491, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7118039131164551, + "num_tokens": 337655550.0, + "step": 13175 + }, + { + "epoch": 1.4469580496376016, + "grad_norm": 2.0770881175994873, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7346563935279846, + "num_tokens": 337685360.0, + "step": 13176 + }, + { + "epoch": 1.4470678673402153, + "grad_norm": 2.3685500621795654, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7140802145004272, + "num_tokens": 337707896.0, + "step": 13177 + }, + { + "epoch": 1.4471776850428288, + "grad_norm": 2.2282276153564453, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7256872057914734, + "num_tokens": 337733051.0, + "step": 13178 + }, + { + "epoch": 1.4472875027454426, + "grad_norm": 2.6630189418792725, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7253309488296509, + "num_tokens": 337753578.0, + "step": 13179 + }, + { + "epoch": 1.4473973204480561, + "grad_norm": 2.1816372871398926, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7150553464889526, + "num_tokens": 337782059.0, + "step": 13180 + }, + { + "epoch": 1.44750713815067, + "grad_norm": 2.77759051322937, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7457209825515747, + "num_tokens": 337800557.0, + "step": 13181 + }, + { + "epoch": 1.4476169558532836, + "grad_norm": 2.341782331466675, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.737879753112793, + "num_tokens": 337822477.0, + "step": 13182 + }, + { + "epoch": 1.4477267735558972, + "grad_norm": 2.272888660430908, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.712894856929779, + "num_tokens": 337848099.0, + "step": 13183 + }, + { + "epoch": 1.447836591258511, + "grad_norm": 2.179020404815674, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.74761962890625, + "num_tokens": 337873576.0, + "step": 13184 + }, + { + "epoch": 1.4479464089611245, + "grad_norm": 2.0639612674713135, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7407724857330322, + "num_tokens": 337901014.0, + "step": 13185 + }, + { + "epoch": 1.4480562266637382, + "grad_norm": 2.266542911529541, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7241838574409485, + "num_tokens": 337927208.0, + "step": 13186 + }, + { + "epoch": 1.448166044366352, + "grad_norm": 2.301375389099121, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7320943474769592, + "num_tokens": 337951018.0, + "step": 13187 + }, + { + "epoch": 1.4482758620689655, + "grad_norm": 2.250396966934204, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7121689319610596, + "num_tokens": 337978105.0, + "step": 13188 + }, + { + "epoch": 1.448385679771579, + "grad_norm": 2.0108044147491455, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7102459073066711, + "num_tokens": 338006619.0, + "step": 13189 + }, + { + "epoch": 1.4484954974741928, + "grad_norm": 2.1635005474090576, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7243125438690186, + "num_tokens": 338033703.0, + "step": 13190 + }, + { + "epoch": 1.4486053151768066, + "grad_norm": 2.1455814838409424, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7042487859725952, + "num_tokens": 338059919.0, + "step": 13191 + }, + { + "epoch": 1.44871513287942, + "grad_norm": 2.25186824798584, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7278248071670532, + "num_tokens": 338084731.0, + "step": 13192 + }, + { + "epoch": 1.4488249505820339, + "grad_norm": 2.242067813873291, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.727496862411499, + "num_tokens": 338109783.0, + "step": 13193 + }, + { + "epoch": 1.4489347682846474, + "grad_norm": 2.1489248275756836, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6892426013946533, + "num_tokens": 338138776.0, + "step": 13194 + }, + { + "epoch": 1.4490445859872612, + "grad_norm": 2.0290284156799316, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7039008140563965, + "num_tokens": 338169957.0, + "step": 13195 + }, + { + "epoch": 1.449154403689875, + "grad_norm": 2.394197702407837, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7234299182891846, + "num_tokens": 338192876.0, + "step": 13196 + }, + { + "epoch": 1.4492642213924884, + "grad_norm": 2.1535346508026123, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7090436220169067, + "num_tokens": 338220192.0, + "step": 13197 + }, + { + "epoch": 1.4493740390951022, + "grad_norm": 2.1909031867980957, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7424596548080444, + "num_tokens": 338247961.0, + "step": 13198 + }, + { + "epoch": 1.4494838567977157, + "grad_norm": 2.2011966705322266, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7132634520530701, + "num_tokens": 338274099.0, + "step": 13199 + }, + { + "epoch": 1.4495936745003295, + "grad_norm": 1.9667340517044067, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7598361968994141, + "num_tokens": 338302037.0, + "step": 13200 + }, + { + "epoch": 1.449703492202943, + "grad_norm": 2.331765651702881, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7281146049499512, + "num_tokens": 338326806.0, + "step": 13201 + }, + { + "epoch": 1.4498133099055568, + "grad_norm": 2.255601644515991, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.6970036029815674, + "num_tokens": 338352577.0, + "step": 13202 + }, + { + "epoch": 1.4499231276081703, + "grad_norm": 2.1076600551605225, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.702280580997467, + "num_tokens": 338381376.0, + "step": 13203 + }, + { + "epoch": 1.450032945310784, + "grad_norm": 1.997736930847168, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7139669060707092, + "num_tokens": 338412550.0, + "step": 13204 + }, + { + "epoch": 1.4501427630133978, + "grad_norm": 2.3498260974884033, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7034659385681152, + "num_tokens": 338436862.0, + "step": 13205 + }, + { + "epoch": 1.4502525807160114, + "grad_norm": 2.5360703468322754, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7501620650291443, + "num_tokens": 338461202.0, + "step": 13206 + }, + { + "epoch": 1.4503623984186251, + "grad_norm": 2.2320098876953125, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6942625045776367, + "num_tokens": 338487518.0, + "step": 13207 + }, + { + "epoch": 1.4504722161212387, + "grad_norm": 2.0620498657226562, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7297723293304443, + "num_tokens": 338516955.0, + "step": 13208 + }, + { + "epoch": 1.4505820338238524, + "grad_norm": 2.018636465072632, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7300313115119934, + "num_tokens": 338545940.0, + "step": 13209 + }, + { + "epoch": 1.4506918515264662, + "grad_norm": 2.3375461101531982, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7301740646362305, + "num_tokens": 338568709.0, + "step": 13210 + }, + { + "epoch": 1.4508016692290797, + "grad_norm": 2.2676730155944824, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7150940895080566, + "num_tokens": 338593292.0, + "step": 13211 + }, + { + "epoch": 1.4509114869316933, + "grad_norm": 2.287790060043335, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7104452848434448, + "num_tokens": 338620611.0, + "step": 13212 + }, + { + "epoch": 1.451021304634307, + "grad_norm": 2.3379454612731934, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7159985303878784, + "num_tokens": 338645169.0, + "step": 13213 + }, + { + "epoch": 1.4511311223369208, + "grad_norm": 2.1836771965026855, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7216797471046448, + "num_tokens": 338672528.0, + "step": 13214 + }, + { + "epoch": 1.4512409400395343, + "grad_norm": 2.0894315242767334, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6785072088241577, + "num_tokens": 338702447.0, + "step": 13215 + }, + { + "epoch": 1.451350757742148, + "grad_norm": 2.4671432971954346, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.722143292427063, + "num_tokens": 338723690.0, + "step": 13216 + }, + { + "epoch": 1.4514605754447616, + "grad_norm": 2.052618980407715, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7201225757598877, + "num_tokens": 338752510.0, + "step": 13217 + }, + { + "epoch": 1.4515703931473753, + "grad_norm": 2.3302724361419678, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7224665880203247, + "num_tokens": 338775980.0, + "step": 13218 + }, + { + "epoch": 1.451680210849989, + "grad_norm": 2.3390731811523438, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7581412196159363, + "num_tokens": 338798652.0, + "step": 13219 + }, + { + "epoch": 1.4517900285526026, + "grad_norm": 2.6067938804626465, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7209060788154602, + "num_tokens": 338819342.0, + "step": 13220 + }, + { + "epoch": 1.4518998462552164, + "grad_norm": 2.197460889816284, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7162691950798035, + "num_tokens": 338846210.0, + "step": 13221 + }, + { + "epoch": 1.45200966395783, + "grad_norm": 2.4735732078552246, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6905731558799744, + "num_tokens": 338870090.0, + "step": 13222 + }, + { + "epoch": 1.4521194816604437, + "grad_norm": 2.297137498855591, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7112714052200317, + "num_tokens": 338894226.0, + "step": 13223 + }, + { + "epoch": 1.4522292993630574, + "grad_norm": 2.103100299835205, + "learning_rate": 1e-06, + "loss": 1.0719, + "mean_token_accuracy": 0.6781941652297974, + "num_tokens": 338923561.0, + "step": 13224 + }, + { + "epoch": 1.452339117065671, + "grad_norm": 2.157057523727417, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6973470449447632, + "num_tokens": 338952349.0, + "step": 13225 + }, + { + "epoch": 1.4524489347682845, + "grad_norm": 2.5964062213897705, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7260457277297974, + "num_tokens": 338973994.0, + "step": 13226 + }, + { + "epoch": 1.4525587524708983, + "grad_norm": 2.2890660762786865, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7309213280677795, + "num_tokens": 338996447.0, + "step": 13227 + }, + { + "epoch": 1.452668570173512, + "grad_norm": 2.0375993251800537, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7259589433670044, + "num_tokens": 339026241.0, + "step": 13228 + }, + { + "epoch": 1.4527783878761256, + "grad_norm": 2.184262275695801, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7194109559059143, + "num_tokens": 339053093.0, + "step": 13229 + }, + { + "epoch": 1.4528882055787393, + "grad_norm": 2.3575987815856934, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7164293527603149, + "num_tokens": 339078181.0, + "step": 13230 + }, + { + "epoch": 1.4529980232813529, + "grad_norm": 2.4396309852600098, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7134872674942017, + "num_tokens": 339100718.0, + "step": 13231 + }, + { + "epoch": 1.4531078409839666, + "grad_norm": 2.1645870208740234, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7023921012878418, + "num_tokens": 339127416.0, + "step": 13232 + }, + { + "epoch": 1.4532176586865804, + "grad_norm": 2.2967045307159424, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.735482394695282, + "num_tokens": 339150032.0, + "step": 13233 + }, + { + "epoch": 1.453327476389194, + "grad_norm": 1.9506241083145142, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7278350591659546, + "num_tokens": 339181674.0, + "step": 13234 + }, + { + "epoch": 1.4534372940918077, + "grad_norm": 2.309943675994873, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7121268510818481, + "num_tokens": 339207262.0, + "step": 13235 + }, + { + "epoch": 1.4535471117944212, + "grad_norm": 2.019733428955078, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7313020825386047, + "num_tokens": 339235962.0, + "step": 13236 + }, + { + "epoch": 1.453656929497035, + "grad_norm": 2.064004421234131, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7031521797180176, + "num_tokens": 339267335.0, + "step": 13237 + }, + { + "epoch": 1.4537667471996487, + "grad_norm": 1.9283902645111084, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7309308052062988, + "num_tokens": 339300772.0, + "step": 13238 + }, + { + "epoch": 1.4538765649022622, + "grad_norm": 2.006523370742798, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6898066997528076, + "num_tokens": 339333139.0, + "step": 13239 + }, + { + "epoch": 1.4539863826048758, + "grad_norm": 2.130531072616577, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7321170568466187, + "num_tokens": 339360344.0, + "step": 13240 + }, + { + "epoch": 1.4540962003074895, + "grad_norm": 2.2142584323883057, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7227567434310913, + "num_tokens": 339386956.0, + "step": 13241 + }, + { + "epoch": 1.4542060180101033, + "grad_norm": 2.1543538570404053, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6949729919433594, + "num_tokens": 339415075.0, + "step": 13242 + }, + { + "epoch": 1.4543158357127168, + "grad_norm": 2.320713996887207, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7404552698135376, + "num_tokens": 339439252.0, + "step": 13243 + }, + { + "epoch": 1.4544256534153306, + "grad_norm": 2.1687049865722656, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.710713267326355, + "num_tokens": 339467974.0, + "step": 13244 + }, + { + "epoch": 1.4545354711179441, + "grad_norm": 2.3635094165802, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7348753809928894, + "num_tokens": 339491969.0, + "step": 13245 + }, + { + "epoch": 1.4546452888205579, + "grad_norm": 2.096477508544922, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7178244590759277, + "num_tokens": 339521110.0, + "step": 13246 + }, + { + "epoch": 1.4547551065231716, + "grad_norm": 2.3139710426330566, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7299931049346924, + "num_tokens": 339545868.0, + "step": 13247 + }, + { + "epoch": 1.4548649242257852, + "grad_norm": 2.364715099334717, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.71083003282547, + "num_tokens": 339570141.0, + "step": 13248 + }, + { + "epoch": 1.454974741928399, + "grad_norm": 2.0320191383361816, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7119612693786621, + "num_tokens": 339601498.0, + "step": 13249 + }, + { + "epoch": 1.4550845596310125, + "grad_norm": 2.2968273162841797, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7151224613189697, + "num_tokens": 339626560.0, + "step": 13250 + }, + { + "epoch": 1.4551943773336262, + "grad_norm": 2.071357011795044, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7226199507713318, + "num_tokens": 339655666.0, + "step": 13251 + }, + { + "epoch": 1.45530419503624, + "grad_norm": 2.1222734451293945, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7061275243759155, + "num_tokens": 339683093.0, + "step": 13252 + }, + { + "epoch": 1.4554140127388535, + "grad_norm": 2.282289981842041, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7507168054580688, + "num_tokens": 339705408.0, + "step": 13253 + }, + { + "epoch": 1.455523830441467, + "grad_norm": 1.9334535598754883, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6955394744873047, + "num_tokens": 339738986.0, + "step": 13254 + }, + { + "epoch": 1.4556336481440808, + "grad_norm": 2.0714263916015625, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7046408653259277, + "num_tokens": 339769767.0, + "step": 13255 + }, + { + "epoch": 1.4557434658466946, + "grad_norm": 2.3349738121032715, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7245453596115112, + "num_tokens": 339792431.0, + "step": 13256 + }, + { + "epoch": 1.455853283549308, + "grad_norm": 2.108776569366455, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7401248216629028, + "num_tokens": 339818668.0, + "step": 13257 + }, + { + "epoch": 1.4559631012519219, + "grad_norm": 2.2065114974975586, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7166451215744019, + "num_tokens": 339845648.0, + "step": 13258 + }, + { + "epoch": 1.4560729189545354, + "grad_norm": 2.3705830574035645, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7334370613098145, + "num_tokens": 339869140.0, + "step": 13259 + }, + { + "epoch": 1.4561827366571491, + "grad_norm": 2.4465320110321045, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.751839816570282, + "num_tokens": 339891365.0, + "step": 13260 + }, + { + "epoch": 1.456292554359763, + "grad_norm": 2.469425678253174, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7236372828483582, + "num_tokens": 339915902.0, + "step": 13261 + }, + { + "epoch": 1.4564023720623764, + "grad_norm": 2.270663022994995, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.722183346748352, + "num_tokens": 339942590.0, + "step": 13262 + }, + { + "epoch": 1.4565121897649902, + "grad_norm": 2.640902042388916, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7276175022125244, + "num_tokens": 339963133.0, + "step": 13263 + }, + { + "epoch": 1.4566220074676037, + "grad_norm": 2.272249698638916, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7367017865180969, + "num_tokens": 339987233.0, + "step": 13264 + }, + { + "epoch": 1.4567318251702175, + "grad_norm": 2.2760226726531982, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6944526433944702, + "num_tokens": 340012563.0, + "step": 13265 + }, + { + "epoch": 1.456841642872831, + "grad_norm": 2.4283041954040527, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7164199352264404, + "num_tokens": 340035269.0, + "step": 13266 + }, + { + "epoch": 1.4569514605754448, + "grad_norm": 2.231761932373047, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.734498143196106, + "num_tokens": 340059558.0, + "step": 13267 + }, + { + "epoch": 1.4570612782780583, + "grad_norm": 1.8575267791748047, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6943050622940063, + "num_tokens": 340098006.0, + "step": 13268 + }, + { + "epoch": 1.457171095980672, + "grad_norm": 2.2465178966522217, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7256060838699341, + "num_tokens": 340123157.0, + "step": 13269 + }, + { + "epoch": 1.4572809136832858, + "grad_norm": 2.139981508255005, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7179129123687744, + "num_tokens": 340151341.0, + "step": 13270 + }, + { + "epoch": 1.4573907313858994, + "grad_norm": 2.1942763328552246, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7251090407371521, + "num_tokens": 340176720.0, + "step": 13271 + }, + { + "epoch": 1.4575005490885131, + "grad_norm": 2.118898391723633, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7144948840141296, + "num_tokens": 340204163.0, + "step": 13272 + }, + { + "epoch": 1.4576103667911267, + "grad_norm": 2.3407459259033203, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.705703854560852, + "num_tokens": 340230881.0, + "step": 13273 + }, + { + "epoch": 1.4577201844937404, + "grad_norm": 2.321206569671631, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7060443758964539, + "num_tokens": 340256905.0, + "step": 13274 + }, + { + "epoch": 1.4578300021963542, + "grad_norm": 2.1388375759124756, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.727927565574646, + "num_tokens": 340286163.0, + "step": 13275 + }, + { + "epoch": 1.4579398198989677, + "grad_norm": 2.095616102218628, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7261191606521606, + "num_tokens": 340312652.0, + "step": 13276 + }, + { + "epoch": 1.4580496376015812, + "grad_norm": 1.9870774745941162, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7150278091430664, + "num_tokens": 340345727.0, + "step": 13277 + }, + { + "epoch": 1.458159455304195, + "grad_norm": 2.0868234634399414, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7398571968078613, + "num_tokens": 340374066.0, + "step": 13278 + }, + { + "epoch": 1.4582692730068088, + "grad_norm": 2.3610827922821045, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6982492208480835, + "num_tokens": 340402149.0, + "step": 13279 + }, + { + "epoch": 1.4583790907094223, + "grad_norm": 2.4819421768188477, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7217092514038086, + "num_tokens": 340424251.0, + "step": 13280 + }, + { + "epoch": 1.458488908412036, + "grad_norm": 2.123134136199951, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7054296731948853, + "num_tokens": 340453326.0, + "step": 13281 + }, + { + "epoch": 1.4585987261146496, + "grad_norm": 2.370879650115967, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7041557431221008, + "num_tokens": 340477459.0, + "step": 13282 + }, + { + "epoch": 1.4587085438172633, + "grad_norm": 2.389024257659912, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7150553464889526, + "num_tokens": 340501982.0, + "step": 13283 + }, + { + "epoch": 1.458818361519877, + "grad_norm": 2.396430730819702, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.728798508644104, + "num_tokens": 340525958.0, + "step": 13284 + }, + { + "epoch": 1.4589281792224906, + "grad_norm": 2.508453130722046, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7287682294845581, + "num_tokens": 340546361.0, + "step": 13285 + }, + { + "epoch": 1.4590379969251044, + "grad_norm": 2.553609609603882, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7203160524368286, + "num_tokens": 340567893.0, + "step": 13286 + }, + { + "epoch": 1.459147814627718, + "grad_norm": 2.370023727416992, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7010996341705322, + "num_tokens": 340593015.0, + "step": 13287 + }, + { + "epoch": 1.4592576323303317, + "grad_norm": 2.2142136096954346, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7206450700759888, + "num_tokens": 340618089.0, + "step": 13288 + }, + { + "epoch": 1.4593674500329454, + "grad_norm": 2.223361015319824, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7229354381561279, + "num_tokens": 340645771.0, + "step": 13289 + }, + { + "epoch": 1.459477267735559, + "grad_norm": 2.307927370071411, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6999531388282776, + "num_tokens": 340671886.0, + "step": 13290 + }, + { + "epoch": 1.4595870854381725, + "grad_norm": 2.0112946033477783, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6958238482475281, + "num_tokens": 340701734.0, + "step": 13291 + }, + { + "epoch": 1.4596969031407863, + "grad_norm": 2.1862664222717285, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.730962336063385, + "num_tokens": 340728197.0, + "step": 13292 + }, + { + "epoch": 1.4598067208434, + "grad_norm": 2.3485729694366455, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7197468280792236, + "num_tokens": 340750468.0, + "step": 13293 + }, + { + "epoch": 1.4599165385460136, + "grad_norm": 2.0147557258605957, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7371327877044678, + "num_tokens": 340777886.0, + "step": 13294 + }, + { + "epoch": 1.4600263562486273, + "grad_norm": 2.4898688793182373, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7505149841308594, + "num_tokens": 340796660.0, + "step": 13295 + }, + { + "epoch": 1.4601361739512408, + "grad_norm": 2.066532611846924, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.685556948184967, + "num_tokens": 340828993.0, + "step": 13296 + }, + { + "epoch": 1.4602459916538546, + "grad_norm": 2.0935163497924805, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7070653438568115, + "num_tokens": 340858211.0, + "step": 13297 + }, + { + "epoch": 1.4603558093564684, + "grad_norm": 2.1872990131378174, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7308788895606995, + "num_tokens": 340887630.0, + "step": 13298 + }, + { + "epoch": 1.460465627059082, + "grad_norm": 1.869981288909912, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7280592918395996, + "num_tokens": 340923843.0, + "step": 13299 + }, + { + "epoch": 1.4605754447616957, + "grad_norm": 2.47221302986145, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7303066253662109, + "num_tokens": 340945169.0, + "step": 13300 + }, + { + "epoch": 1.4606852624643092, + "grad_norm": 2.008239984512329, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7174968719482422, + "num_tokens": 340974187.0, + "step": 13301 + }, + { + "epoch": 1.460795080166923, + "grad_norm": 2.4175093173980713, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7244270443916321, + "num_tokens": 340998549.0, + "step": 13302 + }, + { + "epoch": 1.4609048978695367, + "grad_norm": 2.183018207550049, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7254547476768494, + "num_tokens": 341026251.0, + "step": 13303 + }, + { + "epoch": 1.4610147155721502, + "grad_norm": 2.4851059913635254, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7252171039581299, + "num_tokens": 341047628.0, + "step": 13304 + }, + { + "epoch": 1.4611245332747638, + "grad_norm": 2.3487837314605713, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7192747592926025, + "num_tokens": 341072651.0, + "step": 13305 + }, + { + "epoch": 1.4612343509773775, + "grad_norm": 2.2438883781433105, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6930994987487793, + "num_tokens": 341102674.0, + "step": 13306 + }, + { + "epoch": 1.4613441686799913, + "grad_norm": 2.162836790084839, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7211356163024902, + "num_tokens": 341129700.0, + "step": 13307 + }, + { + "epoch": 1.4614539863826048, + "grad_norm": 2.08294677734375, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7399905323982239, + "num_tokens": 341157342.0, + "step": 13308 + }, + { + "epoch": 1.4615638040852186, + "grad_norm": 2.566854238510132, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7460942268371582, + "num_tokens": 341176377.0, + "step": 13309 + }, + { + "epoch": 1.4616736217878321, + "grad_norm": 2.174954891204834, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7144700288772583, + "num_tokens": 341201148.0, + "step": 13310 + }, + { + "epoch": 1.4617834394904459, + "grad_norm": 2.0988516807556152, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.702768862247467, + "num_tokens": 341231521.0, + "step": 13311 + }, + { + "epoch": 1.4618932571930596, + "grad_norm": 2.1762378215789795, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7158777713775635, + "num_tokens": 341259345.0, + "step": 13312 + }, + { + "epoch": 1.4620030748956732, + "grad_norm": 2.2492947578430176, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7171658277511597, + "num_tokens": 341285761.0, + "step": 13313 + }, + { + "epoch": 1.462112892598287, + "grad_norm": 2.4098050594329834, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6930249929428101, + "num_tokens": 341309461.0, + "step": 13314 + }, + { + "epoch": 1.4622227103009005, + "grad_norm": 2.1607770919799805, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.713971734046936, + "num_tokens": 341336687.0, + "step": 13315 + }, + { + "epoch": 1.4623325280035142, + "grad_norm": 2.344731569290161, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.718149721622467, + "num_tokens": 341360309.0, + "step": 13316 + }, + { + "epoch": 1.462442345706128, + "grad_norm": 2.3915843963623047, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7194221019744873, + "num_tokens": 341384589.0, + "step": 13317 + }, + { + "epoch": 1.4625521634087415, + "grad_norm": 2.34848690032959, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7123634219169617, + "num_tokens": 341409018.0, + "step": 13318 + }, + { + "epoch": 1.462661981111355, + "grad_norm": 2.4089772701263428, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7153266668319702, + "num_tokens": 341434113.0, + "step": 13319 + }, + { + "epoch": 1.4627717988139688, + "grad_norm": 2.2604339122772217, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7241392135620117, + "num_tokens": 341461957.0, + "step": 13320 + }, + { + "epoch": 1.4628816165165826, + "grad_norm": 2.3150923252105713, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7314668893814087, + "num_tokens": 341485661.0, + "step": 13321 + }, + { + "epoch": 1.462991434219196, + "grad_norm": 2.038555860519409, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.70987468957901, + "num_tokens": 341518415.0, + "step": 13322 + }, + { + "epoch": 1.4631012519218098, + "grad_norm": 2.324533700942993, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7194662094116211, + "num_tokens": 341543036.0, + "step": 13323 + }, + { + "epoch": 1.4632110696244234, + "grad_norm": 2.4364356994628906, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7440943717956543, + "num_tokens": 341566192.0, + "step": 13324 + }, + { + "epoch": 1.4633208873270371, + "grad_norm": 2.4392764568328857, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7289309501647949, + "num_tokens": 341589680.0, + "step": 13325 + }, + { + "epoch": 1.463430705029651, + "grad_norm": 2.493476152420044, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7322699427604675, + "num_tokens": 341610757.0, + "step": 13326 + }, + { + "epoch": 1.4635405227322644, + "grad_norm": 2.233431100845337, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7146488428115845, + "num_tokens": 341638092.0, + "step": 13327 + }, + { + "epoch": 1.4636503404348782, + "grad_norm": 2.1709516048431396, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7088959217071533, + "num_tokens": 341667020.0, + "step": 13328 + }, + { + "epoch": 1.4637601581374917, + "grad_norm": 2.263624429702759, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7142363786697388, + "num_tokens": 341691603.0, + "step": 13329 + }, + { + "epoch": 1.4638699758401055, + "grad_norm": 2.4073526859283447, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7378339767456055, + "num_tokens": 341713675.0, + "step": 13330 + }, + { + "epoch": 1.463979793542719, + "grad_norm": 2.3235230445861816, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7165600061416626, + "num_tokens": 341740616.0, + "step": 13331 + }, + { + "epoch": 1.4640896112453328, + "grad_norm": 2.195164203643799, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7121113538742065, + "num_tokens": 341767688.0, + "step": 13332 + }, + { + "epoch": 1.4641994289479463, + "grad_norm": 2.5463030338287354, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7197749614715576, + "num_tokens": 341788987.0, + "step": 13333 + }, + { + "epoch": 1.46430924665056, + "grad_norm": 2.3485641479492188, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7351333498954773, + "num_tokens": 341814247.0, + "step": 13334 + }, + { + "epoch": 1.4644190643531738, + "grad_norm": 2.3700592517852783, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6996474266052246, + "num_tokens": 341838421.0, + "step": 13335 + }, + { + "epoch": 1.4645288820557874, + "grad_norm": 2.5211191177368164, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7360930442810059, + "num_tokens": 341859514.0, + "step": 13336 + }, + { + "epoch": 1.464638699758401, + "grad_norm": 2.237819194793701, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7282172441482544, + "num_tokens": 341885166.0, + "step": 13337 + }, + { + "epoch": 1.4647485174610146, + "grad_norm": 2.368006944656372, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7474111318588257, + "num_tokens": 341908637.0, + "step": 13338 + }, + { + "epoch": 1.4648583351636284, + "grad_norm": 2.1187305450439453, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7184761166572571, + "num_tokens": 341938513.0, + "step": 13339 + }, + { + "epoch": 1.4649681528662422, + "grad_norm": 2.5934488773345947, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7156251072883606, + "num_tokens": 341959327.0, + "step": 13340 + }, + { + "epoch": 1.4650779705688557, + "grad_norm": 2.5437941551208496, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.724027156829834, + "num_tokens": 341981031.0, + "step": 13341 + }, + { + "epoch": 1.4651877882714692, + "grad_norm": 2.5302047729492188, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7029384970664978, + "num_tokens": 342003507.0, + "step": 13342 + }, + { + "epoch": 1.465297605974083, + "grad_norm": 2.253248453140259, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7108863592147827, + "num_tokens": 342030179.0, + "step": 13343 + }, + { + "epoch": 1.4654074236766967, + "grad_norm": 2.2149972915649414, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.721269965171814, + "num_tokens": 342057925.0, + "step": 13344 + }, + { + "epoch": 1.4655172413793103, + "grad_norm": 2.063483953475952, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6946780681610107, + "num_tokens": 342088721.0, + "step": 13345 + }, + { + "epoch": 1.465627059081924, + "grad_norm": 2.1739108562469482, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7221255898475647, + "num_tokens": 342116625.0, + "step": 13346 + }, + { + "epoch": 1.4657368767845376, + "grad_norm": 2.503898859024048, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7458441257476807, + "num_tokens": 342136254.0, + "step": 13347 + }, + { + "epoch": 1.4658466944871513, + "grad_norm": 2.071481943130493, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7154195308685303, + "num_tokens": 342164844.0, + "step": 13348 + }, + { + "epoch": 1.465956512189765, + "grad_norm": 2.164268970489502, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7215471863746643, + "num_tokens": 342191937.0, + "step": 13349 + }, + { + "epoch": 1.4660663298923786, + "grad_norm": 2.184429168701172, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7431004643440247, + "num_tokens": 342217202.0, + "step": 13350 + }, + { + "epoch": 1.4661761475949924, + "grad_norm": 1.9928712844848633, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7100255489349365, + "num_tokens": 342247756.0, + "step": 13351 + }, + { + "epoch": 1.466285965297606, + "grad_norm": 2.296231746673584, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7138426899909973, + "num_tokens": 342273138.0, + "step": 13352 + }, + { + "epoch": 1.4663957830002197, + "grad_norm": 2.503028392791748, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7264962196350098, + "num_tokens": 342294267.0, + "step": 13353 + }, + { + "epoch": 1.4665056007028334, + "grad_norm": 2.2148585319519043, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7177005410194397, + "num_tokens": 342322190.0, + "step": 13354 + }, + { + "epoch": 1.466615418405447, + "grad_norm": 2.179928779602051, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7333348989486694, + "num_tokens": 342350953.0, + "step": 13355 + }, + { + "epoch": 1.4667252361080605, + "grad_norm": 2.2504425048828125, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7208777070045471, + "num_tokens": 342377429.0, + "step": 13356 + }, + { + "epoch": 1.4668350538106742, + "grad_norm": 2.183026075363159, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6981236934661865, + "num_tokens": 342407803.0, + "step": 13357 + }, + { + "epoch": 1.466944871513288, + "grad_norm": 2.1805615425109863, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7117908000946045, + "num_tokens": 342434373.0, + "step": 13358 + }, + { + "epoch": 1.4670546892159015, + "grad_norm": 2.454646110534668, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7395813465118408, + "num_tokens": 342455526.0, + "step": 13359 + }, + { + "epoch": 1.4671645069185153, + "grad_norm": 2.311844825744629, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7086009979248047, + "num_tokens": 342479793.0, + "step": 13360 + }, + { + "epoch": 1.4672743246211288, + "grad_norm": 2.459821939468384, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7044696807861328, + "num_tokens": 342501484.0, + "step": 13361 + }, + { + "epoch": 1.4673841423237426, + "grad_norm": 2.3309500217437744, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7379794120788574, + "num_tokens": 342523740.0, + "step": 13362 + }, + { + "epoch": 1.4674939600263563, + "grad_norm": 2.0350217819213867, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.693191409111023, + "num_tokens": 342552113.0, + "step": 13363 + }, + { + "epoch": 1.4676037777289699, + "grad_norm": 2.262169599533081, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7265679836273193, + "num_tokens": 342577397.0, + "step": 13364 + }, + { + "epoch": 1.4677135954315836, + "grad_norm": 1.9372270107269287, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7286997437477112, + "num_tokens": 342608370.0, + "step": 13365 + }, + { + "epoch": 1.4678234131341972, + "grad_norm": 2.1244852542877197, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7024527192115784, + "num_tokens": 342636678.0, + "step": 13366 + }, + { + "epoch": 1.467933230836811, + "grad_norm": 2.2848165035247803, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7198450565338135, + "num_tokens": 342661672.0, + "step": 13367 + }, + { + "epoch": 1.4680430485394247, + "grad_norm": 2.1796185970306396, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7094932794570923, + "num_tokens": 342687571.0, + "step": 13368 + }, + { + "epoch": 1.4681528662420382, + "grad_norm": 2.567277431488037, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.704382061958313, + "num_tokens": 342708531.0, + "step": 13369 + }, + { + "epoch": 1.4682626839446518, + "grad_norm": 2.2476608753204346, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.722387433052063, + "num_tokens": 342734924.0, + "step": 13370 + }, + { + "epoch": 1.4683725016472655, + "grad_norm": 2.2215418815612793, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7284067273139954, + "num_tokens": 342761396.0, + "step": 13371 + }, + { + "epoch": 1.4684823193498793, + "grad_norm": 2.417635679244995, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7244926691055298, + "num_tokens": 342783455.0, + "step": 13372 + }, + { + "epoch": 1.4685921370524928, + "grad_norm": 2.1151223182678223, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7054056525230408, + "num_tokens": 342812282.0, + "step": 13373 + }, + { + "epoch": 1.4687019547551066, + "grad_norm": 2.5015742778778076, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7145397663116455, + "num_tokens": 342833624.0, + "step": 13374 + }, + { + "epoch": 1.46881177245772, + "grad_norm": 2.2947230339050293, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7147542834281921, + "num_tokens": 342859043.0, + "step": 13375 + }, + { + "epoch": 1.4689215901603339, + "grad_norm": 2.3174049854278564, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7063790559768677, + "num_tokens": 342885440.0, + "step": 13376 + }, + { + "epoch": 1.4690314078629476, + "grad_norm": 2.0219709873199463, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.715474009513855, + "num_tokens": 342916307.0, + "step": 13377 + }, + { + "epoch": 1.4691412255655611, + "grad_norm": 2.346608877182007, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7493573427200317, + "num_tokens": 342937862.0, + "step": 13378 + }, + { + "epoch": 1.469251043268175, + "grad_norm": 2.1934688091278076, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.719575047492981, + "num_tokens": 342963520.0, + "step": 13379 + }, + { + "epoch": 1.4693608609707884, + "grad_norm": 2.2182624340057373, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7373636960983276, + "num_tokens": 342989332.0, + "step": 13380 + }, + { + "epoch": 1.4694706786734022, + "grad_norm": 2.248854637145996, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.721942663192749, + "num_tokens": 343015094.0, + "step": 13381 + }, + { + "epoch": 1.4695804963760157, + "grad_norm": 2.3570213317871094, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7273504137992859, + "num_tokens": 343039125.0, + "step": 13382 + }, + { + "epoch": 1.4696903140786295, + "grad_norm": 1.9918053150177002, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7148575782775879, + "num_tokens": 343071886.0, + "step": 13383 + }, + { + "epoch": 1.469800131781243, + "grad_norm": 1.8848990201950073, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7236169576644897, + "num_tokens": 343106075.0, + "step": 13384 + }, + { + "epoch": 1.4699099494838568, + "grad_norm": 2.124746799468994, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7188825607299805, + "num_tokens": 343132528.0, + "step": 13385 + }, + { + "epoch": 1.4700197671864705, + "grad_norm": 2.0443146228790283, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6979450583457947, + "num_tokens": 343164274.0, + "step": 13386 + }, + { + "epoch": 1.470129584889084, + "grad_norm": 2.2708399295806885, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7176390886306763, + "num_tokens": 343190697.0, + "step": 13387 + }, + { + "epoch": 1.4702394025916978, + "grad_norm": 2.222748279571533, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7322624921798706, + "num_tokens": 343215064.0, + "step": 13388 + }, + { + "epoch": 1.4703492202943114, + "grad_norm": 2.045128107070923, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7056183218955994, + "num_tokens": 343248125.0, + "step": 13389 + }, + { + "epoch": 1.4704590379969251, + "grad_norm": 2.4032235145568848, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7248862981796265, + "num_tokens": 343272914.0, + "step": 13390 + }, + { + "epoch": 1.4705688556995389, + "grad_norm": 2.4236018657684326, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7108218669891357, + "num_tokens": 343298210.0, + "step": 13391 + }, + { + "epoch": 1.4706786734021524, + "grad_norm": 2.136493682861328, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7214096784591675, + "num_tokens": 343326658.0, + "step": 13392 + }, + { + "epoch": 1.4707884911047662, + "grad_norm": 2.3205783367156982, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7127682566642761, + "num_tokens": 343350361.0, + "step": 13393 + }, + { + "epoch": 1.4708983088073797, + "grad_norm": 2.0365612506866455, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.694230318069458, + "num_tokens": 343381398.0, + "step": 13394 + }, + { + "epoch": 1.4710081265099935, + "grad_norm": 2.722961902618408, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7370971441268921, + "num_tokens": 343400061.0, + "step": 13395 + }, + { + "epoch": 1.471117944212607, + "grad_norm": 2.1551706790924072, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7126896381378174, + "num_tokens": 343428487.0, + "step": 13396 + }, + { + "epoch": 1.4712277619152208, + "grad_norm": 2.5772855281829834, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7181013822555542, + "num_tokens": 343449431.0, + "step": 13397 + }, + { + "epoch": 1.4713375796178343, + "grad_norm": 2.1672475337982178, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.715738832950592, + "num_tokens": 343476077.0, + "step": 13398 + }, + { + "epoch": 1.471447397320448, + "grad_norm": 2.6452789306640625, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7223615050315857, + "num_tokens": 343494253.0, + "step": 13399 + }, + { + "epoch": 1.4715572150230618, + "grad_norm": 1.9964871406555176, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7141716480255127, + "num_tokens": 343525755.0, + "step": 13400 + }, + { + "epoch": 1.4716670327256753, + "grad_norm": 2.4283363819122314, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7340749502182007, + "num_tokens": 343548020.0, + "step": 13401 + }, + { + "epoch": 1.471776850428289, + "grad_norm": 2.1665775775909424, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7243804931640625, + "num_tokens": 343574824.0, + "step": 13402 + }, + { + "epoch": 1.4718866681309026, + "grad_norm": 2.3916754722595215, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.766597330570221, + "num_tokens": 343595977.0, + "step": 13403 + }, + { + "epoch": 1.4719964858335164, + "grad_norm": 2.4472219944000244, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7075291872024536, + "num_tokens": 343620782.0, + "step": 13404 + }, + { + "epoch": 1.4721063035361301, + "grad_norm": 2.32588529586792, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7117691040039062, + "num_tokens": 343645463.0, + "step": 13405 + }, + { + "epoch": 1.4722161212387437, + "grad_norm": 2.6205289363861084, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7137766480445862, + "num_tokens": 343666286.0, + "step": 13406 + }, + { + "epoch": 1.4723259389413572, + "grad_norm": 2.3604540824890137, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7261698246002197, + "num_tokens": 343690034.0, + "step": 13407 + }, + { + "epoch": 1.472435756643971, + "grad_norm": 2.218538284301758, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6963987350463867, + "num_tokens": 343720854.0, + "step": 13408 + }, + { + "epoch": 1.4725455743465847, + "grad_norm": 2.04190731048584, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7298065423965454, + "num_tokens": 343748910.0, + "step": 13409 + }, + { + "epoch": 1.4726553920491983, + "grad_norm": 2.0189223289489746, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7135165929794312, + "num_tokens": 343779560.0, + "step": 13410 + }, + { + "epoch": 1.472765209751812, + "grad_norm": 2.3358986377716064, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7201623916625977, + "num_tokens": 343802512.0, + "step": 13411 + }, + { + "epoch": 1.4728750274544256, + "grad_norm": 1.8414456844329834, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7114944458007812, + "num_tokens": 343835478.0, + "step": 13412 + }, + { + "epoch": 1.4729848451570393, + "grad_norm": 2.1263623237609863, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7266354560852051, + "num_tokens": 343863421.0, + "step": 13413 + }, + { + "epoch": 1.473094662859653, + "grad_norm": 2.053973436355591, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7185863256454468, + "num_tokens": 343891239.0, + "step": 13414 + }, + { + "epoch": 1.4732044805622666, + "grad_norm": 2.084472417831421, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7127154469490051, + "num_tokens": 343921773.0, + "step": 13415 + }, + { + "epoch": 1.4733142982648804, + "grad_norm": 2.249295949935913, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7452731728553772, + "num_tokens": 343946800.0, + "step": 13416 + }, + { + "epoch": 1.473424115967494, + "grad_norm": 2.3635153770446777, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7406679391860962, + "num_tokens": 343969575.0, + "step": 13417 + }, + { + "epoch": 1.4735339336701077, + "grad_norm": 2.1116998195648193, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7171032428741455, + "num_tokens": 343996460.0, + "step": 13418 + }, + { + "epoch": 1.4736437513727214, + "grad_norm": 2.0201165676116943, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7135211229324341, + "num_tokens": 344028052.0, + "step": 13419 + }, + { + "epoch": 1.473753569075335, + "grad_norm": 2.124216079711914, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7159989476203918, + "num_tokens": 344054370.0, + "step": 13420 + }, + { + "epoch": 1.4738633867779485, + "grad_norm": 2.375872850418091, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7162574529647827, + "num_tokens": 344078238.0, + "step": 13421 + }, + { + "epoch": 1.4739732044805622, + "grad_norm": 2.0803122520446777, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7079735398292542, + "num_tokens": 344108395.0, + "step": 13422 + }, + { + "epoch": 1.474083022183176, + "grad_norm": 2.7166781425476074, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7393602132797241, + "num_tokens": 344124864.0, + "step": 13423 + }, + { + "epoch": 1.4741928398857895, + "grad_norm": 2.448127508163452, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7194297313690186, + "num_tokens": 344147866.0, + "step": 13424 + }, + { + "epoch": 1.4743026575884033, + "grad_norm": 1.990061640739441, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6823997497558594, + "num_tokens": 344180759.0, + "step": 13425 + }, + { + "epoch": 1.4744124752910168, + "grad_norm": 2.2785534858703613, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7032122611999512, + "num_tokens": 344207243.0, + "step": 13426 + }, + { + "epoch": 1.4745222929936306, + "grad_norm": 2.6914613246917725, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.712459921836853, + "num_tokens": 344229750.0, + "step": 13427 + }, + { + "epoch": 1.4746321106962443, + "grad_norm": 2.130235433578491, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7162998914718628, + "num_tokens": 344258234.0, + "step": 13428 + }, + { + "epoch": 1.4747419283988579, + "grad_norm": 2.0644750595092773, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7231465578079224, + "num_tokens": 344286454.0, + "step": 13429 + }, + { + "epoch": 1.4748517461014716, + "grad_norm": 2.281843900680542, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7049593329429626, + "num_tokens": 344313043.0, + "step": 13430 + }, + { + "epoch": 1.4749615638040852, + "grad_norm": 2.2678050994873047, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7221188545227051, + "num_tokens": 344337735.0, + "step": 13431 + }, + { + "epoch": 1.475071381506699, + "grad_norm": 2.3464598655700684, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7218205332756042, + "num_tokens": 344361651.0, + "step": 13432 + }, + { + "epoch": 1.4751811992093127, + "grad_norm": 2.280010223388672, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7129842042922974, + "num_tokens": 344386413.0, + "step": 13433 + }, + { + "epoch": 1.4752910169119262, + "grad_norm": 2.0066285133361816, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6959813833236694, + "num_tokens": 344416733.0, + "step": 13434 + }, + { + "epoch": 1.4754008346145397, + "grad_norm": 2.48897123336792, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7455263733863831, + "num_tokens": 344437252.0, + "step": 13435 + }, + { + "epoch": 1.4755106523171535, + "grad_norm": 2.4621028900146484, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7335084676742554, + "num_tokens": 344458604.0, + "step": 13436 + }, + { + "epoch": 1.4756204700197673, + "grad_norm": 2.2034435272216797, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.716827392578125, + "num_tokens": 344485041.0, + "step": 13437 + }, + { + "epoch": 1.4757302877223808, + "grad_norm": 2.44621205329895, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7431000471115112, + "num_tokens": 344506405.0, + "step": 13438 + }, + { + "epoch": 1.4758401054249946, + "grad_norm": 2.0924642086029053, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7032596468925476, + "num_tokens": 344534869.0, + "step": 13439 + }, + { + "epoch": 1.475949923127608, + "grad_norm": 2.0326478481292725, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7227815389633179, + "num_tokens": 344566551.0, + "step": 13440 + }, + { + "epoch": 1.4760597408302218, + "grad_norm": 2.506418466567993, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.720106840133667, + "num_tokens": 344587863.0, + "step": 13441 + }, + { + "epoch": 1.4761695585328356, + "grad_norm": 2.114130735397339, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6895737648010254, + "num_tokens": 344618578.0, + "step": 13442 + }, + { + "epoch": 1.4762793762354491, + "grad_norm": 2.275926113128662, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7181038856506348, + "num_tokens": 344644332.0, + "step": 13443 + }, + { + "epoch": 1.476389193938063, + "grad_norm": 2.0816421508789062, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7299176454544067, + "num_tokens": 344673567.0, + "step": 13444 + }, + { + "epoch": 1.4764990116406764, + "grad_norm": 2.4546031951904297, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7074024677276611, + "num_tokens": 344697820.0, + "step": 13445 + }, + { + "epoch": 1.4766088293432902, + "grad_norm": 2.3235225677490234, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7176553606987, + "num_tokens": 344724365.0, + "step": 13446 + }, + { + "epoch": 1.4767186470459037, + "grad_norm": 2.5471417903900146, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7308824062347412, + "num_tokens": 344744477.0, + "step": 13447 + }, + { + "epoch": 1.4768284647485175, + "grad_norm": 2.2744524478912354, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7219046354293823, + "num_tokens": 344770189.0, + "step": 13448 + }, + { + "epoch": 1.476938282451131, + "grad_norm": 2.2138185501098633, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7270633578300476, + "num_tokens": 344796809.0, + "step": 13449 + }, + { + "epoch": 1.4770481001537448, + "grad_norm": 2.7587006092071533, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7462701797485352, + "num_tokens": 344814496.0, + "step": 13450 + }, + { + "epoch": 1.4771579178563585, + "grad_norm": 2.4222233295440674, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7178509831428528, + "num_tokens": 344838560.0, + "step": 13451 + }, + { + "epoch": 1.477267735558972, + "grad_norm": 2.494080066680908, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7189679741859436, + "num_tokens": 344860128.0, + "step": 13452 + }, + { + "epoch": 1.4773775532615858, + "grad_norm": 2.5287978649139404, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.729119062423706, + "num_tokens": 344880742.0, + "step": 13453 + }, + { + "epoch": 1.4774873709641994, + "grad_norm": 2.1143903732299805, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7022159695625305, + "num_tokens": 344909308.0, + "step": 13454 + }, + { + "epoch": 1.477597188666813, + "grad_norm": 2.6475937366485596, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7259501218795776, + "num_tokens": 344928610.0, + "step": 13455 + }, + { + "epoch": 1.4777070063694269, + "grad_norm": 2.6262404918670654, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7059327363967896, + "num_tokens": 344948833.0, + "step": 13456 + }, + { + "epoch": 1.4778168240720404, + "grad_norm": 1.9794665575027466, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.711563229560852, + "num_tokens": 344979476.0, + "step": 13457 + }, + { + "epoch": 1.477926641774654, + "grad_norm": 2.1786181926727295, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7290016412734985, + "num_tokens": 345005604.0, + "step": 13458 + }, + { + "epoch": 1.4780364594772677, + "grad_norm": 2.2170448303222656, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7216947078704834, + "num_tokens": 345029490.0, + "step": 13459 + }, + { + "epoch": 1.4781462771798815, + "grad_norm": 1.8201693296432495, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7050244212150574, + "num_tokens": 345065599.0, + "step": 13460 + }, + { + "epoch": 1.478256094882495, + "grad_norm": 2.192014455795288, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7013569474220276, + "num_tokens": 345092365.0, + "step": 13461 + }, + { + "epoch": 1.4783659125851087, + "grad_norm": 2.2219996452331543, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7042893171310425, + "num_tokens": 345118722.0, + "step": 13462 + }, + { + "epoch": 1.4784757302877223, + "grad_norm": 2.070903778076172, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7182538509368896, + "num_tokens": 345146892.0, + "step": 13463 + }, + { + "epoch": 1.478585547990336, + "grad_norm": 2.308798313140869, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7087583541870117, + "num_tokens": 345172636.0, + "step": 13464 + }, + { + "epoch": 1.4786953656929498, + "grad_norm": 2.2611188888549805, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7286619544029236, + "num_tokens": 345198731.0, + "step": 13465 + }, + { + "epoch": 1.4788051833955633, + "grad_norm": 2.148200511932373, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6942543387413025, + "num_tokens": 345226748.0, + "step": 13466 + }, + { + "epoch": 1.478915001098177, + "grad_norm": 2.1402766704559326, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7392047643661499, + "num_tokens": 345253140.0, + "step": 13467 + }, + { + "epoch": 1.4790248188007906, + "grad_norm": 2.2967615127563477, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7220531702041626, + "num_tokens": 345278129.0, + "step": 13468 + }, + { + "epoch": 1.4791346365034044, + "grad_norm": 2.1938228607177734, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7067638039588928, + "num_tokens": 345305317.0, + "step": 13469 + }, + { + "epoch": 1.4792444542060181, + "grad_norm": 2.0685813426971436, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.706800103187561, + "num_tokens": 345337541.0, + "step": 13470 + }, + { + "epoch": 1.4793542719086317, + "grad_norm": 2.1081409454345703, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7286081910133362, + "num_tokens": 345364339.0, + "step": 13471 + }, + { + "epoch": 1.4794640896112452, + "grad_norm": 2.4296650886535645, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7049721479415894, + "num_tokens": 345390366.0, + "step": 13472 + }, + { + "epoch": 1.479573907313859, + "grad_norm": 2.4316964149475098, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7385344505310059, + "num_tokens": 345413769.0, + "step": 13473 + }, + { + "epoch": 1.4796837250164727, + "grad_norm": 2.4248008728027344, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7416900396347046, + "num_tokens": 345434406.0, + "step": 13474 + }, + { + "epoch": 1.4797935427190863, + "grad_norm": 2.086818218231201, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.722527027130127, + "num_tokens": 345466315.0, + "step": 13475 + }, + { + "epoch": 1.4799033604217, + "grad_norm": 2.169443368911743, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7313789129257202, + "num_tokens": 345491557.0, + "step": 13476 + }, + { + "epoch": 1.4800131781243135, + "grad_norm": 2.7703638076782227, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7066438794136047, + "num_tokens": 345511363.0, + "step": 13477 + }, + { + "epoch": 1.4801229958269273, + "grad_norm": 2.2137954235076904, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7210457921028137, + "num_tokens": 345540213.0, + "step": 13478 + }, + { + "epoch": 1.480232813529541, + "grad_norm": 2.179368495941162, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7236279845237732, + "num_tokens": 345565553.0, + "step": 13479 + }, + { + "epoch": 1.4803426312321546, + "grad_norm": 2.5759458541870117, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7362884283065796, + "num_tokens": 345585192.0, + "step": 13480 + }, + { + "epoch": 1.4804524489347684, + "grad_norm": 2.56211519241333, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7290710210800171, + "num_tokens": 345605840.0, + "step": 13481 + }, + { + "epoch": 1.4805622666373819, + "grad_norm": 2.1284892559051514, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6923783421516418, + "num_tokens": 345636040.0, + "step": 13482 + }, + { + "epoch": 1.4806720843399956, + "grad_norm": 2.314164400100708, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.723501980304718, + "num_tokens": 345660054.0, + "step": 13483 + }, + { + "epoch": 1.4807819020426094, + "grad_norm": 2.2031872272491455, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7216150164604187, + "num_tokens": 345689905.0, + "step": 13484 + }, + { + "epoch": 1.480891719745223, + "grad_norm": 2.346625328063965, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7393959760665894, + "num_tokens": 345714400.0, + "step": 13485 + }, + { + "epoch": 1.4810015374478365, + "grad_norm": 2.4259564876556396, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.730918288230896, + "num_tokens": 345736653.0, + "step": 13486 + }, + { + "epoch": 1.4811113551504502, + "grad_norm": 2.1751787662506104, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7058454155921936, + "num_tokens": 345765028.0, + "step": 13487 + }, + { + "epoch": 1.481221172853064, + "grad_norm": 2.2571752071380615, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7352246046066284, + "num_tokens": 345789598.0, + "step": 13488 + }, + { + "epoch": 1.4813309905556775, + "grad_norm": 2.2229743003845215, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7231050729751587, + "num_tokens": 345813072.0, + "step": 13489 + }, + { + "epoch": 1.4814408082582913, + "grad_norm": 2.205385684967041, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.720325767993927, + "num_tokens": 345836821.0, + "step": 13490 + }, + { + "epoch": 1.4815506259609048, + "grad_norm": 2.3317623138427734, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7249942421913147, + "num_tokens": 345862565.0, + "step": 13491 + }, + { + "epoch": 1.4816604436635186, + "grad_norm": 2.3588051795959473, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7232940793037415, + "num_tokens": 345886091.0, + "step": 13492 + }, + { + "epoch": 1.4817702613661323, + "grad_norm": 2.190727949142456, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7135258913040161, + "num_tokens": 345913000.0, + "step": 13493 + }, + { + "epoch": 1.4818800790687459, + "grad_norm": 2.544530153274536, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7246842384338379, + "num_tokens": 345934421.0, + "step": 13494 + }, + { + "epoch": 1.4819898967713596, + "grad_norm": 2.4091150760650635, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7583059072494507, + "num_tokens": 345953828.0, + "step": 13495 + }, + { + "epoch": 1.4820997144739732, + "grad_norm": 2.15779972076416, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.716402530670166, + "num_tokens": 345980779.0, + "step": 13496 + }, + { + "epoch": 1.482209532176587, + "grad_norm": 2.1592166423797607, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7284271717071533, + "num_tokens": 346007699.0, + "step": 13497 + }, + { + "epoch": 1.4823193498792007, + "grad_norm": 2.362882375717163, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7399884462356567, + "num_tokens": 346030512.0, + "step": 13498 + }, + { + "epoch": 1.4824291675818142, + "grad_norm": 2.3102798461914062, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7174543142318726, + "num_tokens": 346056792.0, + "step": 13499 + }, + { + "epoch": 1.4825389852844277, + "grad_norm": 2.2126095294952393, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7076691389083862, + "num_tokens": 346084426.0, + "step": 13500 + }, + { + "epoch": 1.4826488029870415, + "grad_norm": 2.3404908180236816, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7204462885856628, + "num_tokens": 346108397.0, + "step": 13501 + }, + { + "epoch": 1.4827586206896552, + "grad_norm": 2.194014549255371, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7232152819633484, + "num_tokens": 346136122.0, + "step": 13502 + }, + { + "epoch": 1.4828684383922688, + "grad_norm": 2.405771493911743, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7223927974700928, + "num_tokens": 346159843.0, + "step": 13503 + }, + { + "epoch": 1.4829782560948825, + "grad_norm": 2.1372334957122803, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7084993124008179, + "num_tokens": 346188990.0, + "step": 13504 + }, + { + "epoch": 1.483088073797496, + "grad_norm": 2.010791063308716, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.716913104057312, + "num_tokens": 346219792.0, + "step": 13505 + }, + { + "epoch": 1.4831978915001098, + "grad_norm": 2.0580310821533203, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7105967998504639, + "num_tokens": 346248935.0, + "step": 13506 + }, + { + "epoch": 1.4833077092027236, + "grad_norm": 2.0860674381256104, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6950681209564209, + "num_tokens": 346280508.0, + "step": 13507 + }, + { + "epoch": 1.4834175269053371, + "grad_norm": 2.2901103496551514, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7102125883102417, + "num_tokens": 346304320.0, + "step": 13508 + }, + { + "epoch": 1.4835273446079509, + "grad_norm": 2.0489721298217773, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7110515832901001, + "num_tokens": 346334835.0, + "step": 13509 + }, + { + "epoch": 1.4836371623105644, + "grad_norm": 2.23386549949646, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7358074188232422, + "num_tokens": 346360984.0, + "step": 13510 + }, + { + "epoch": 1.4837469800131782, + "grad_norm": 2.003544330596924, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.72010737657547, + "num_tokens": 346391023.0, + "step": 13511 + }, + { + "epoch": 1.4838567977157917, + "grad_norm": 2.860591173171997, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.730861246585846, + "num_tokens": 346409548.0, + "step": 13512 + }, + { + "epoch": 1.4839666154184055, + "grad_norm": 2.4617199897766113, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7325038909912109, + "num_tokens": 346432620.0, + "step": 13513 + }, + { + "epoch": 1.484076433121019, + "grad_norm": 2.206127166748047, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7328180074691772, + "num_tokens": 346459034.0, + "step": 13514 + }, + { + "epoch": 1.4841862508236328, + "grad_norm": 2.2454235553741455, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7188822627067566, + "num_tokens": 346484483.0, + "step": 13515 + }, + { + "epoch": 1.4842960685262465, + "grad_norm": 2.505340099334717, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.716396689414978, + "num_tokens": 346507785.0, + "step": 13516 + }, + { + "epoch": 1.48440588622886, + "grad_norm": 2.5432605743408203, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7389947175979614, + "num_tokens": 346529138.0, + "step": 13517 + }, + { + "epoch": 1.4845157039314738, + "grad_norm": 2.2211649417877197, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7162866592407227, + "num_tokens": 346554489.0, + "step": 13518 + }, + { + "epoch": 1.4846255216340873, + "grad_norm": 2.281785726547241, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7237344980239868, + "num_tokens": 346578717.0, + "step": 13519 + }, + { + "epoch": 1.484735339336701, + "grad_norm": 2.414639711380005, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.735806941986084, + "num_tokens": 346601464.0, + "step": 13520 + }, + { + "epoch": 1.4848451570393149, + "grad_norm": 2.4461002349853516, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7315983772277832, + "num_tokens": 346623327.0, + "step": 13521 + }, + { + "epoch": 1.4849549747419284, + "grad_norm": 2.1152071952819824, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7212837934494019, + "num_tokens": 346651179.0, + "step": 13522 + }, + { + "epoch": 1.485064792444542, + "grad_norm": 2.088606834411621, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7149873971939087, + "num_tokens": 346679090.0, + "step": 13523 + }, + { + "epoch": 1.4851746101471557, + "grad_norm": 2.1049182415008545, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7149631381034851, + "num_tokens": 346705771.0, + "step": 13524 + }, + { + "epoch": 1.4852844278497694, + "grad_norm": 2.281113624572754, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7005813121795654, + "num_tokens": 346733438.0, + "step": 13525 + }, + { + "epoch": 1.485394245552383, + "grad_norm": 2.312380075454712, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7103585004806519, + "num_tokens": 346756886.0, + "step": 13526 + }, + { + "epoch": 1.4855040632549967, + "grad_norm": 2.3436572551727295, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7284954786300659, + "num_tokens": 346779758.0, + "step": 13527 + }, + { + "epoch": 1.4856138809576103, + "grad_norm": 2.2805991172790527, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.704790472984314, + "num_tokens": 346805696.0, + "step": 13528 + }, + { + "epoch": 1.485723698660224, + "grad_norm": 2.30092191696167, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7097287178039551, + "num_tokens": 346831244.0, + "step": 13529 + }, + { + "epoch": 1.4858335163628378, + "grad_norm": 2.2951102256774902, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7071963548660278, + "num_tokens": 346856273.0, + "step": 13530 + }, + { + "epoch": 1.4859433340654513, + "grad_norm": 2.376488447189331, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7156329154968262, + "num_tokens": 346879476.0, + "step": 13531 + }, + { + "epoch": 1.486053151768065, + "grad_norm": 2.337815284729004, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7216623425483704, + "num_tokens": 346902572.0, + "step": 13532 + }, + { + "epoch": 1.4861629694706786, + "grad_norm": 2.413090467453003, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7108417749404907, + "num_tokens": 346925005.0, + "step": 13533 + }, + { + "epoch": 1.4862727871732924, + "grad_norm": 2.157158613204956, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7180352210998535, + "num_tokens": 346954716.0, + "step": 13534 + }, + { + "epoch": 1.4863826048759061, + "grad_norm": 2.3365557193756104, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7378724813461304, + "num_tokens": 346979195.0, + "step": 13535 + }, + { + "epoch": 1.4864924225785197, + "grad_norm": 2.210981607437134, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7036967277526855, + "num_tokens": 347007958.0, + "step": 13536 + }, + { + "epoch": 1.4866022402811332, + "grad_norm": 2.2484307289123535, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7332456707954407, + "num_tokens": 347030780.0, + "step": 13537 + }, + { + "epoch": 1.486712057983747, + "grad_norm": 2.22951340675354, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.759246826171875, + "num_tokens": 347056231.0, + "step": 13538 + }, + { + "epoch": 1.4868218756863607, + "grad_norm": 2.1128017902374268, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7230627536773682, + "num_tokens": 347085523.0, + "step": 13539 + }, + { + "epoch": 1.4869316933889742, + "grad_norm": 2.0720133781433105, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7037693858146667, + "num_tokens": 347116948.0, + "step": 13540 + }, + { + "epoch": 1.487041511091588, + "grad_norm": 2.0640640258789062, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6943036913871765, + "num_tokens": 347146294.0, + "step": 13541 + }, + { + "epoch": 1.4871513287942015, + "grad_norm": 2.2478671073913574, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7155945301055908, + "num_tokens": 347170290.0, + "step": 13542 + }, + { + "epoch": 1.4872611464968153, + "grad_norm": 1.9604871273040771, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7163605690002441, + "num_tokens": 347203036.0, + "step": 13543 + }, + { + "epoch": 1.487370964199429, + "grad_norm": 2.5203800201416016, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.751582145690918, + "num_tokens": 347223518.0, + "step": 13544 + }, + { + "epoch": 1.4874807819020426, + "grad_norm": 2.3613357543945312, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7521252632141113, + "num_tokens": 347246152.0, + "step": 13545 + }, + { + "epoch": 1.4875905996046563, + "grad_norm": 2.0348172187805176, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7128344774246216, + "num_tokens": 347275809.0, + "step": 13546 + }, + { + "epoch": 1.4877004173072699, + "grad_norm": 2.2417070865631104, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7090474367141724, + "num_tokens": 347303538.0, + "step": 13547 + }, + { + "epoch": 1.4878102350098836, + "grad_norm": 2.626265048980713, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7195156812667847, + "num_tokens": 347325012.0, + "step": 13548 + }, + { + "epoch": 1.4879200527124974, + "grad_norm": 2.3434345722198486, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7096149325370789, + "num_tokens": 347350423.0, + "step": 13549 + }, + { + "epoch": 1.488029870415111, + "grad_norm": 2.4796950817108154, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7237759232521057, + "num_tokens": 347372223.0, + "step": 13550 + }, + { + "epoch": 1.4881396881177245, + "grad_norm": 2.1953771114349365, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7195022106170654, + "num_tokens": 347398001.0, + "step": 13551 + }, + { + "epoch": 1.4882495058203382, + "grad_norm": 2.211735963821411, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.717496395111084, + "num_tokens": 347422083.0, + "step": 13552 + }, + { + "epoch": 1.488359323522952, + "grad_norm": 2.2336041927337646, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7338259220123291, + "num_tokens": 347446865.0, + "step": 13553 + }, + { + "epoch": 1.4884691412255655, + "grad_norm": 2.477738618850708, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7259268760681152, + "num_tokens": 347468462.0, + "step": 13554 + }, + { + "epoch": 1.4885789589281793, + "grad_norm": 2.414306163787842, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7224846482276917, + "num_tokens": 347493660.0, + "step": 13555 + }, + { + "epoch": 1.4886887766307928, + "grad_norm": 2.314882755279541, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7164114713668823, + "num_tokens": 347518413.0, + "step": 13556 + }, + { + "epoch": 1.4887985943334066, + "grad_norm": 1.9622327089309692, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7027788162231445, + "num_tokens": 347553209.0, + "step": 13557 + }, + { + "epoch": 1.4889084120360203, + "grad_norm": 2.3124518394470215, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7144664525985718, + "num_tokens": 347577281.0, + "step": 13558 + }, + { + "epoch": 1.4890182297386338, + "grad_norm": 2.360501527786255, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7175527215003967, + "num_tokens": 347600438.0, + "step": 13559 + }, + { + "epoch": 1.4891280474412476, + "grad_norm": 2.354416847229004, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7183536887168884, + "num_tokens": 347623814.0, + "step": 13560 + }, + { + "epoch": 1.4892378651438611, + "grad_norm": 2.2667784690856934, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6903221011161804, + "num_tokens": 347650755.0, + "step": 13561 + }, + { + "epoch": 1.489347682846475, + "grad_norm": 2.1627848148345947, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.707821249961853, + "num_tokens": 347677416.0, + "step": 13562 + }, + { + "epoch": 1.4894575005490884, + "grad_norm": 2.4535715579986572, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7047665119171143, + "num_tokens": 347701641.0, + "step": 13563 + }, + { + "epoch": 1.4895673182517022, + "grad_norm": 2.5314254760742188, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7207553386688232, + "num_tokens": 347724727.0, + "step": 13564 + }, + { + "epoch": 1.4896771359543157, + "grad_norm": 2.0883140563964844, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6973913908004761, + "num_tokens": 347754121.0, + "step": 13565 + }, + { + "epoch": 1.4897869536569295, + "grad_norm": 2.304584503173828, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7339381575584412, + "num_tokens": 347778768.0, + "step": 13566 + }, + { + "epoch": 1.4898967713595432, + "grad_norm": 2.3082337379455566, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7168434858322144, + "num_tokens": 347803054.0, + "step": 13567 + }, + { + "epoch": 1.4900065890621568, + "grad_norm": 2.089559316635132, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7307049632072449, + "num_tokens": 347834468.0, + "step": 13568 + }, + { + "epoch": 1.4901164067647705, + "grad_norm": 2.0563621520996094, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7206677794456482, + "num_tokens": 347864794.0, + "step": 13569 + }, + { + "epoch": 1.490226224467384, + "grad_norm": 2.2583155632019043, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7210446000099182, + "num_tokens": 347890582.0, + "step": 13570 + }, + { + "epoch": 1.4903360421699978, + "grad_norm": 2.220243453979492, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7164026498794556, + "num_tokens": 347916966.0, + "step": 13571 + }, + { + "epoch": 1.4904458598726116, + "grad_norm": 2.6372716426849365, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7547260522842407, + "num_tokens": 347935858.0, + "step": 13572 + }, + { + "epoch": 1.4905556775752251, + "grad_norm": 1.9201024770736694, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7226410508155823, + "num_tokens": 347967170.0, + "step": 13573 + }, + { + "epoch": 1.4906654952778389, + "grad_norm": 2.3121070861816406, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7131656408309937, + "num_tokens": 347993084.0, + "step": 13574 + }, + { + "epoch": 1.4907753129804524, + "grad_norm": 2.6461613178253174, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7085696458816528, + "num_tokens": 348012423.0, + "step": 13575 + }, + { + "epoch": 1.4908851306830662, + "grad_norm": 2.2245681285858154, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7133358716964722, + "num_tokens": 348037123.0, + "step": 13576 + }, + { + "epoch": 1.4909949483856797, + "grad_norm": 2.1704909801483154, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7064085602760315, + "num_tokens": 348065239.0, + "step": 13577 + }, + { + "epoch": 1.4911047660882935, + "grad_norm": 2.050123929977417, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7092674374580383, + "num_tokens": 348097744.0, + "step": 13578 + }, + { + "epoch": 1.491214583790907, + "grad_norm": 1.9565181732177734, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7128316760063171, + "num_tokens": 348130089.0, + "step": 13579 + }, + { + "epoch": 1.4913244014935207, + "grad_norm": 2.3450052738189697, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6995112895965576, + "num_tokens": 348153517.0, + "step": 13580 + }, + { + "epoch": 1.4914342191961345, + "grad_norm": 2.581735610961914, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.750954270362854, + "num_tokens": 348172478.0, + "step": 13581 + }, + { + "epoch": 1.491544036898748, + "grad_norm": 2.4298436641693115, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7356880307197571, + "num_tokens": 348194506.0, + "step": 13582 + }, + { + "epoch": 1.4916538546013618, + "grad_norm": 2.28587007522583, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7212984561920166, + "num_tokens": 348218861.0, + "step": 13583 + }, + { + "epoch": 1.4917636723039753, + "grad_norm": 2.4130120277404785, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7368031740188599, + "num_tokens": 348241759.0, + "step": 13584 + }, + { + "epoch": 1.491873490006589, + "grad_norm": 2.264556884765625, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6933289766311646, + "num_tokens": 348268177.0, + "step": 13585 + }, + { + "epoch": 1.4919833077092028, + "grad_norm": 2.1739084720611572, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7410246133804321, + "num_tokens": 348294557.0, + "step": 13586 + }, + { + "epoch": 1.4920931254118164, + "grad_norm": 2.4320178031921387, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.721455454826355, + "num_tokens": 348318014.0, + "step": 13587 + }, + { + "epoch": 1.49220294311443, + "grad_norm": 2.4111618995666504, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7215995788574219, + "num_tokens": 348342094.0, + "step": 13588 + }, + { + "epoch": 1.4923127608170437, + "grad_norm": 2.2322700023651123, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7091203927993774, + "num_tokens": 348368881.0, + "step": 13589 + }, + { + "epoch": 1.4924225785196574, + "grad_norm": 2.2796707153320312, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7273231744766235, + "num_tokens": 348394260.0, + "step": 13590 + }, + { + "epoch": 1.492532396222271, + "grad_norm": 2.048159599304199, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7238431572914124, + "num_tokens": 348422897.0, + "step": 13591 + }, + { + "epoch": 1.4926422139248847, + "grad_norm": 2.3954906463623047, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.742015540599823, + "num_tokens": 348445441.0, + "step": 13592 + }, + { + "epoch": 1.4927520316274983, + "grad_norm": 2.475959062576294, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7218334674835205, + "num_tokens": 348467216.0, + "step": 13593 + }, + { + "epoch": 1.492861849330112, + "grad_norm": 2.091153860092163, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7532058358192444, + "num_tokens": 348493288.0, + "step": 13594 + }, + { + "epoch": 1.4929716670327258, + "grad_norm": 2.324934244155884, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7250083684921265, + "num_tokens": 348518723.0, + "step": 13595 + }, + { + "epoch": 1.4930814847353393, + "grad_norm": 2.2214975357055664, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7147573232650757, + "num_tokens": 348543941.0, + "step": 13596 + }, + { + "epoch": 1.493191302437953, + "grad_norm": 2.3133456707000732, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.738720178604126, + "num_tokens": 348567581.0, + "step": 13597 + }, + { + "epoch": 1.4933011201405666, + "grad_norm": 2.2899742126464844, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7335380911827087, + "num_tokens": 348590805.0, + "step": 13598 + }, + { + "epoch": 1.4934109378431804, + "grad_norm": 2.219231605529785, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7050448656082153, + "num_tokens": 348619490.0, + "step": 13599 + }, + { + "epoch": 1.493520755545794, + "grad_norm": 2.1452133655548096, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6927197575569153, + "num_tokens": 348647823.0, + "step": 13600 + }, + { + "epoch": 1.4936305732484076, + "grad_norm": 2.600740909576416, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.725835382938385, + "num_tokens": 348667832.0, + "step": 13601 + }, + { + "epoch": 1.4937403909510212, + "grad_norm": 2.0325160026550293, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7329437732696533, + "num_tokens": 348697651.0, + "step": 13602 + }, + { + "epoch": 1.493850208653635, + "grad_norm": 2.1280174255371094, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7091274857521057, + "num_tokens": 348728408.0, + "step": 13603 + }, + { + "epoch": 1.4939600263562487, + "grad_norm": 2.476775884628296, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7281635999679565, + "num_tokens": 348747988.0, + "step": 13604 + }, + { + "epoch": 1.4940698440588622, + "grad_norm": 2.115269184112549, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7194854021072388, + "num_tokens": 348775901.0, + "step": 13605 + }, + { + "epoch": 1.494179661761476, + "grad_norm": 2.2271487712860107, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7565129995346069, + "num_tokens": 348800263.0, + "step": 13606 + }, + { + "epoch": 1.4942894794640895, + "grad_norm": 2.2534236907958984, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7316182851791382, + "num_tokens": 348824537.0, + "step": 13607 + }, + { + "epoch": 1.4943992971667033, + "grad_norm": 2.2919206619262695, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7158992290496826, + "num_tokens": 348851329.0, + "step": 13608 + }, + { + "epoch": 1.494509114869317, + "grad_norm": 2.320516586303711, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7263658046722412, + "num_tokens": 348878802.0, + "step": 13609 + }, + { + "epoch": 1.4946189325719306, + "grad_norm": 2.3180694580078125, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7107951045036316, + "num_tokens": 348903369.0, + "step": 13610 + }, + { + "epoch": 1.4947287502745443, + "grad_norm": 2.4862678050994873, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7127587199211121, + "num_tokens": 348923865.0, + "step": 13611 + }, + { + "epoch": 1.4948385679771579, + "grad_norm": 2.0958383083343506, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6915559768676758, + "num_tokens": 348953391.0, + "step": 13612 + }, + { + "epoch": 1.4949483856797716, + "grad_norm": 2.3766419887542725, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7032105326652527, + "num_tokens": 348983195.0, + "step": 13613 + }, + { + "epoch": 1.4950582033823854, + "grad_norm": 2.2541513442993164, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.719007134437561, + "num_tokens": 349008085.0, + "step": 13614 + }, + { + "epoch": 1.495168021084999, + "grad_norm": 2.1556320190429688, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6985934972763062, + "num_tokens": 349036204.0, + "step": 13615 + }, + { + "epoch": 1.4952778387876124, + "grad_norm": 2.3349945545196533, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7244348526000977, + "num_tokens": 349060522.0, + "step": 13616 + }, + { + "epoch": 1.4953876564902262, + "grad_norm": 2.199247360229492, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6981786489486694, + "num_tokens": 349088167.0, + "step": 13617 + }, + { + "epoch": 1.49549747419284, + "grad_norm": 2.2588891983032227, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.731656014919281, + "num_tokens": 349114214.0, + "step": 13618 + }, + { + "epoch": 1.4956072918954535, + "grad_norm": 2.0799810886383057, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7042326331138611, + "num_tokens": 349142755.0, + "step": 13619 + }, + { + "epoch": 1.4957171095980673, + "grad_norm": 2.2258145809173584, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7023518085479736, + "num_tokens": 349170469.0, + "step": 13620 + }, + { + "epoch": 1.4958269273006808, + "grad_norm": 2.348592519760132, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7195578217506409, + "num_tokens": 349195971.0, + "step": 13621 + }, + { + "epoch": 1.4959367450032945, + "grad_norm": 2.31238055229187, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7354587316513062, + "num_tokens": 349220290.0, + "step": 13622 + }, + { + "epoch": 1.4960465627059083, + "grad_norm": 2.112814426422119, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7261412143707275, + "num_tokens": 349248063.0, + "step": 13623 + }, + { + "epoch": 1.4961563804085218, + "grad_norm": 2.302351713180542, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.741503119468689, + "num_tokens": 349272242.0, + "step": 13624 + }, + { + "epoch": 1.4962661981111356, + "grad_norm": 2.03432297706604, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6856857538223267, + "num_tokens": 349304745.0, + "step": 13625 + }, + { + "epoch": 1.4963760158137491, + "grad_norm": 2.068613052368164, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7145065665245056, + "num_tokens": 349334379.0, + "step": 13626 + }, + { + "epoch": 1.4964858335163629, + "grad_norm": 2.0949997901916504, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7105827331542969, + "num_tokens": 349365038.0, + "step": 13627 + }, + { + "epoch": 1.4965956512189764, + "grad_norm": 2.4091806411743164, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7197481393814087, + "num_tokens": 349388302.0, + "step": 13628 + }, + { + "epoch": 1.4967054689215902, + "grad_norm": 2.337829113006592, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7035188674926758, + "num_tokens": 349412982.0, + "step": 13629 + }, + { + "epoch": 1.4968152866242037, + "grad_norm": 2.72139573097229, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7281463742256165, + "num_tokens": 349431096.0, + "step": 13630 + }, + { + "epoch": 1.4969251043268175, + "grad_norm": 2.079684257507324, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6839116811752319, + "num_tokens": 349460006.0, + "step": 13631 + }, + { + "epoch": 1.4970349220294312, + "grad_norm": 2.330853223800659, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7539482712745667, + "num_tokens": 349481570.0, + "step": 13632 + }, + { + "epoch": 1.4971447397320448, + "grad_norm": 2.5647082328796387, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7202435731887817, + "num_tokens": 349503467.0, + "step": 13633 + }, + { + "epoch": 1.4972545574346585, + "grad_norm": 2.651031255722046, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7411876916885376, + "num_tokens": 349522657.0, + "step": 13634 + }, + { + "epoch": 1.497364375137272, + "grad_norm": 2.061183214187622, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7211456298828125, + "num_tokens": 349553570.0, + "step": 13635 + }, + { + "epoch": 1.4974741928398858, + "grad_norm": 2.354072332382202, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.734898567199707, + "num_tokens": 349576381.0, + "step": 13636 + }, + { + "epoch": 1.4975840105424996, + "grad_norm": 2.429330587387085, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7259976863861084, + "num_tokens": 349599046.0, + "step": 13637 + }, + { + "epoch": 1.497693828245113, + "grad_norm": 2.1130192279815674, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7189285755157471, + "num_tokens": 349628963.0, + "step": 13638 + }, + { + "epoch": 1.4978036459477266, + "grad_norm": 2.106292247772217, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7000782489776611, + "num_tokens": 349658017.0, + "step": 13639 + }, + { + "epoch": 1.4979134636503404, + "grad_norm": 2.201472282409668, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7359874844551086, + "num_tokens": 349685711.0, + "step": 13640 + }, + { + "epoch": 1.4980232813529542, + "grad_norm": 1.9665240049362183, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7128942012786865, + "num_tokens": 349719509.0, + "step": 13641 + }, + { + "epoch": 1.4981330990555677, + "grad_norm": 2.7164156436920166, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7090095281600952, + "num_tokens": 349739033.0, + "step": 13642 + }, + { + "epoch": 1.4982429167581814, + "grad_norm": 2.6784982681274414, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7381710410118103, + "num_tokens": 349758364.0, + "step": 13643 + }, + { + "epoch": 1.498352734460795, + "grad_norm": 2.3027868270874023, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7189959287643433, + "num_tokens": 349780864.0, + "step": 13644 + }, + { + "epoch": 1.4984625521634087, + "grad_norm": 2.1236441135406494, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7305899858474731, + "num_tokens": 349807850.0, + "step": 13645 + }, + { + "epoch": 1.4985723698660225, + "grad_norm": 2.4940571784973145, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7376705408096313, + "num_tokens": 349827191.0, + "step": 13646 + }, + { + "epoch": 1.498682187568636, + "grad_norm": 2.208756446838379, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7333736419677734, + "num_tokens": 349851880.0, + "step": 13647 + }, + { + "epoch": 1.4987920052712498, + "grad_norm": 2.1810758113861084, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.70099937915802, + "num_tokens": 349881084.0, + "step": 13648 + }, + { + "epoch": 1.4989018229738633, + "grad_norm": 2.7316529750823975, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7188610434532166, + "num_tokens": 349898907.0, + "step": 13649 + }, + { + "epoch": 1.499011640676477, + "grad_norm": 2.4478559494018555, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7249339818954468, + "num_tokens": 349919938.0, + "step": 13650 + }, + { + "epoch": 1.4991214583790908, + "grad_norm": 2.1414284706115723, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7242171168327332, + "num_tokens": 349946957.0, + "step": 13651 + }, + { + "epoch": 1.4992312760817044, + "grad_norm": 2.2700586318969727, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7069635987281799, + "num_tokens": 349974668.0, + "step": 13652 + }, + { + "epoch": 1.499341093784318, + "grad_norm": 2.5626585483551025, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7198338508605957, + "num_tokens": 349998079.0, + "step": 13653 + }, + { + "epoch": 1.4994509114869317, + "grad_norm": 2.0184309482574463, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7132389545440674, + "num_tokens": 350025877.0, + "step": 13654 + }, + { + "epoch": 1.4995607291895454, + "grad_norm": 2.4480695724487305, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7286258935928345, + "num_tokens": 350047599.0, + "step": 13655 + }, + { + "epoch": 1.499670546892159, + "grad_norm": 2.3542404174804688, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7278542518615723, + "num_tokens": 350070926.0, + "step": 13656 + }, + { + "epoch": 1.4997803645947727, + "grad_norm": 2.2284464836120605, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7175152897834778, + "num_tokens": 350096292.0, + "step": 13657 + }, + { + "epoch": 1.4998901822973862, + "grad_norm": 2.3219332695007324, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7176194190979004, + "num_tokens": 350120458.0, + "step": 13658 + }, + { + "epoch": 1.5, + "grad_norm": 2.3798418045043945, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7166775465011597, + "num_tokens": 350145521.0, + "step": 13659 + }, + { + "epoch": 1.5001098177026138, + "grad_norm": 2.39156436920166, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7081862092018127, + "num_tokens": 350169096.0, + "step": 13660 + }, + { + "epoch": 1.5002196354052273, + "grad_norm": 2.5265824794769287, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7292016744613647, + "num_tokens": 350190458.0, + "step": 13661 + }, + { + "epoch": 1.5003294531078408, + "grad_norm": 1.8198223114013672, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.725296676158905, + "num_tokens": 350226710.0, + "step": 13662 + }, + { + "epoch": 1.5004392708104546, + "grad_norm": 2.235719919204712, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7266099452972412, + "num_tokens": 350251209.0, + "step": 13663 + }, + { + "epoch": 1.5005490885130683, + "grad_norm": 2.1666016578674316, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7075353860855103, + "num_tokens": 350278595.0, + "step": 13664 + }, + { + "epoch": 1.500658906215682, + "grad_norm": 2.0563504695892334, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7153687477111816, + "num_tokens": 350308385.0, + "step": 13665 + }, + { + "epoch": 1.5007687239182956, + "grad_norm": 1.9903903007507324, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.712319016456604, + "num_tokens": 350340093.0, + "step": 13666 + }, + { + "epoch": 1.5008785416209092, + "grad_norm": 2.2815487384796143, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6976439952850342, + "num_tokens": 350367478.0, + "step": 13667 + }, + { + "epoch": 1.500988359323523, + "grad_norm": 1.9778233766555786, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7106781601905823, + "num_tokens": 350400232.0, + "step": 13668 + }, + { + "epoch": 1.5010981770261367, + "grad_norm": 2.4979660511016846, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7271597385406494, + "num_tokens": 350420564.0, + "step": 13669 + }, + { + "epoch": 1.5012079947287504, + "grad_norm": 2.2816832065582275, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7490123510360718, + "num_tokens": 350445088.0, + "step": 13670 + }, + { + "epoch": 1.501317812431364, + "grad_norm": 2.426561117172241, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7266404628753662, + "num_tokens": 350467030.0, + "step": 13671 + }, + { + "epoch": 1.5014276301339775, + "grad_norm": 2.29179310798645, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7273600697517395, + "num_tokens": 350493345.0, + "step": 13672 + }, + { + "epoch": 1.5015374478365913, + "grad_norm": 2.444511890411377, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7073320150375366, + "num_tokens": 350517035.0, + "step": 13673 + }, + { + "epoch": 1.501647265539205, + "grad_norm": 2.1396403312683105, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7210807800292969, + "num_tokens": 350544928.0, + "step": 13674 + }, + { + "epoch": 1.5017570832418186, + "grad_norm": 2.822723150253296, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7360749244689941, + "num_tokens": 350561823.0, + "step": 13675 + }, + { + "epoch": 1.501866900944432, + "grad_norm": 2.1973657608032227, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7225779294967651, + "num_tokens": 350587283.0, + "step": 13676 + }, + { + "epoch": 1.5019767186470458, + "grad_norm": 2.324690103530884, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7113702893257141, + "num_tokens": 350611477.0, + "step": 13677 + }, + { + "epoch": 1.5020865363496596, + "grad_norm": 2.532222032546997, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7343212366104126, + "num_tokens": 350631971.0, + "step": 13678 + }, + { + "epoch": 1.5021963540522734, + "grad_norm": 2.3984522819519043, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7190766930580139, + "num_tokens": 350656487.0, + "step": 13679 + }, + { + "epoch": 1.502306171754887, + "grad_norm": 2.0642874240875244, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6896209716796875, + "num_tokens": 350687870.0, + "step": 13680 + }, + { + "epoch": 1.5024159894575004, + "grad_norm": 2.753789186477661, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7229934334754944, + "num_tokens": 350705771.0, + "step": 13681 + }, + { + "epoch": 1.5025258071601142, + "grad_norm": 1.9882590770721436, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7352025508880615, + "num_tokens": 350736210.0, + "step": 13682 + }, + { + "epoch": 1.502635624862728, + "grad_norm": 2.685694456100464, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7559148073196411, + "num_tokens": 350753798.0, + "step": 13683 + }, + { + "epoch": 1.5027454425653415, + "grad_norm": 2.2568068504333496, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7210578918457031, + "num_tokens": 350778229.0, + "step": 13684 + }, + { + "epoch": 1.5028552602679552, + "grad_norm": 2.7132108211517334, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7145066261291504, + "num_tokens": 350798836.0, + "step": 13685 + }, + { + "epoch": 1.5029650779705688, + "grad_norm": 2.12257719039917, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7430869340896606, + "num_tokens": 350826204.0, + "step": 13686 + }, + { + "epoch": 1.5030748956731825, + "grad_norm": 2.367485284805298, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7185519933700562, + "num_tokens": 350849981.0, + "step": 13687 + }, + { + "epoch": 1.5031847133757963, + "grad_norm": 2.647101640701294, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7190070152282715, + "num_tokens": 350870562.0, + "step": 13688 + }, + { + "epoch": 1.5032945310784098, + "grad_norm": 2.5136334896087646, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7335773706436157, + "num_tokens": 350892593.0, + "step": 13689 + }, + { + "epoch": 1.5034043487810234, + "grad_norm": 2.4376394748687744, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7312713861465454, + "num_tokens": 350913674.0, + "step": 13690 + }, + { + "epoch": 1.5035141664836371, + "grad_norm": 2.278779983520508, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7295594215393066, + "num_tokens": 350937360.0, + "step": 13691 + }, + { + "epoch": 1.5036239841862509, + "grad_norm": 1.9975159168243408, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.720400869846344, + "num_tokens": 350968247.0, + "step": 13692 + }, + { + "epoch": 1.5037338018888646, + "grad_norm": 2.0537877082824707, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7081603407859802, + "num_tokens": 350999240.0, + "step": 13693 + }, + { + "epoch": 1.5038436195914782, + "grad_norm": 2.305152654647827, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7163900136947632, + "num_tokens": 351023392.0, + "step": 13694 + }, + { + "epoch": 1.5039534372940917, + "grad_norm": 2.294633150100708, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7296308279037476, + "num_tokens": 351047810.0, + "step": 13695 + }, + { + "epoch": 1.5040632549967055, + "grad_norm": 2.0904910564422607, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7096205949783325, + "num_tokens": 351076615.0, + "step": 13696 + }, + { + "epoch": 1.5041730726993192, + "grad_norm": 2.2514076232910156, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7181969285011292, + "num_tokens": 351103439.0, + "step": 13697 + }, + { + "epoch": 1.5042828904019327, + "grad_norm": 2.0947299003601074, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7089961171150208, + "num_tokens": 351133526.0, + "step": 13698 + }, + { + "epoch": 1.5043927081045465, + "grad_norm": 2.1696534156799316, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7302846908569336, + "num_tokens": 351160307.0, + "step": 13699 + }, + { + "epoch": 1.50450252580716, + "grad_norm": 2.1806204319000244, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7125535011291504, + "num_tokens": 351186920.0, + "step": 13700 + }, + { + "epoch": 1.5046123435097738, + "grad_norm": 1.9707518815994263, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7036647796630859, + "num_tokens": 351218888.0, + "step": 13701 + }, + { + "epoch": 1.5047221612123876, + "grad_norm": 2.368561267852783, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.720599353313446, + "num_tokens": 351243328.0, + "step": 13702 + }, + { + "epoch": 1.504831978915001, + "grad_norm": 2.131620168685913, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7293257117271423, + "num_tokens": 351270961.0, + "step": 13703 + }, + { + "epoch": 1.5049417966176146, + "grad_norm": 2.43463397026062, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7460494041442871, + "num_tokens": 351291567.0, + "step": 13704 + }, + { + "epoch": 1.5050516143202284, + "grad_norm": 2.583489418029785, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7289683818817139, + "num_tokens": 351311083.0, + "step": 13705 + }, + { + "epoch": 1.5051614320228421, + "grad_norm": 2.3484811782836914, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.740304172039032, + "num_tokens": 351333068.0, + "step": 13706 + }, + { + "epoch": 1.505271249725456, + "grad_norm": 2.487673759460449, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7224358320236206, + "num_tokens": 351355431.0, + "step": 13707 + }, + { + "epoch": 1.5053810674280694, + "grad_norm": 2.4164366722106934, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7444904446601868, + "num_tokens": 351377028.0, + "step": 13708 + }, + { + "epoch": 1.505490885130683, + "grad_norm": 2.442213773727417, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7199922204017639, + "num_tokens": 351399894.0, + "step": 13709 + }, + { + "epoch": 1.5056007028332967, + "grad_norm": 2.609590768814087, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7212457656860352, + "num_tokens": 351419898.0, + "step": 13710 + }, + { + "epoch": 1.5057105205359105, + "grad_norm": 2.440371036529541, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7345494031906128, + "num_tokens": 351440528.0, + "step": 13711 + }, + { + "epoch": 1.505820338238524, + "grad_norm": 2.1269381046295166, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7330343127250671, + "num_tokens": 351467274.0, + "step": 13712 + }, + { + "epoch": 1.5059301559411375, + "grad_norm": 2.3825759887695312, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7316992282867432, + "num_tokens": 351491236.0, + "step": 13713 + }, + { + "epoch": 1.5060399736437513, + "grad_norm": 2.229426622390747, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7069628834724426, + "num_tokens": 351517633.0, + "step": 13714 + }, + { + "epoch": 1.506149791346365, + "grad_norm": 2.115694284439087, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7292096614837646, + "num_tokens": 351543561.0, + "step": 13715 + }, + { + "epoch": 1.5062596090489788, + "grad_norm": 2.2424535751342773, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7465705871582031, + "num_tokens": 351569121.0, + "step": 13716 + }, + { + "epoch": 1.5063694267515924, + "grad_norm": 2.5723025798797607, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7194755673408508, + "num_tokens": 351590543.0, + "step": 13717 + }, + { + "epoch": 1.506479244454206, + "grad_norm": 1.988301157951355, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.738786518573761, + "num_tokens": 351619456.0, + "step": 13718 + }, + { + "epoch": 1.5065890621568196, + "grad_norm": 2.152265787124634, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7145954966545105, + "num_tokens": 351645658.0, + "step": 13719 + }, + { + "epoch": 1.5066988798594334, + "grad_norm": 2.260861873626709, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7151477932929993, + "num_tokens": 351670803.0, + "step": 13720 + }, + { + "epoch": 1.5068086975620472, + "grad_norm": 2.2303144931793213, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.720223069190979, + "num_tokens": 351696305.0, + "step": 13721 + }, + { + "epoch": 1.5069185152646607, + "grad_norm": 1.9284217357635498, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7069259881973267, + "num_tokens": 351728273.0, + "step": 13722 + }, + { + "epoch": 1.5070283329672742, + "grad_norm": 2.7138826847076416, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7220898866653442, + "num_tokens": 351748122.0, + "step": 13723 + }, + { + "epoch": 1.507138150669888, + "grad_norm": 2.2755210399627686, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7134721875190735, + "num_tokens": 351774484.0, + "step": 13724 + }, + { + "epoch": 1.5072479683725017, + "grad_norm": 2.082096576690674, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7138465642929077, + "num_tokens": 351801840.0, + "step": 13725 + }, + { + "epoch": 1.5073577860751153, + "grad_norm": 2.205772638320923, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7242825627326965, + "num_tokens": 351829325.0, + "step": 13726 + }, + { + "epoch": 1.5074676037777288, + "grad_norm": 2.294729709625244, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7129208445549011, + "num_tokens": 351856034.0, + "step": 13727 + }, + { + "epoch": 1.5075774214803426, + "grad_norm": 2.109332799911499, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7144669890403748, + "num_tokens": 351885085.0, + "step": 13728 + }, + { + "epoch": 1.5076872391829563, + "grad_norm": 2.10811185836792, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7127773761749268, + "num_tokens": 351912438.0, + "step": 13729 + }, + { + "epoch": 1.50779705688557, + "grad_norm": 2.136178731918335, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7181851267814636, + "num_tokens": 351941237.0, + "step": 13730 + }, + { + "epoch": 1.5079068745881836, + "grad_norm": 2.0341010093688965, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7449512481689453, + "num_tokens": 351971741.0, + "step": 13731 + }, + { + "epoch": 1.5080166922907972, + "grad_norm": 2.0507614612579346, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7124722003936768, + "num_tokens": 352001492.0, + "step": 13732 + }, + { + "epoch": 1.508126509993411, + "grad_norm": 2.6046245098114014, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.728948712348938, + "num_tokens": 352022184.0, + "step": 13733 + }, + { + "epoch": 1.5082363276960247, + "grad_norm": 2.1150548458099365, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7091161012649536, + "num_tokens": 352049836.0, + "step": 13734 + }, + { + "epoch": 1.5083461453986382, + "grad_norm": 2.2536070346832275, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7162036895751953, + "num_tokens": 352074043.0, + "step": 13735 + }, + { + "epoch": 1.508455963101252, + "grad_norm": 2.2693212032318115, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7114851474761963, + "num_tokens": 352101657.0, + "step": 13736 + }, + { + "epoch": 1.5085657808038655, + "grad_norm": 2.3127219676971436, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6965289115905762, + "num_tokens": 352128877.0, + "step": 13737 + }, + { + "epoch": 1.5086755985064793, + "grad_norm": 2.2195396423339844, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7279796600341797, + "num_tokens": 352155468.0, + "step": 13738 + }, + { + "epoch": 1.508785416209093, + "grad_norm": 2.1368277072906494, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7161777019500732, + "num_tokens": 352185388.0, + "step": 13739 + }, + { + "epoch": 1.5088952339117065, + "grad_norm": 1.8972396850585938, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6955140829086304, + "num_tokens": 352223370.0, + "step": 13740 + }, + { + "epoch": 1.50900505161432, + "grad_norm": 2.4597244262695312, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7275570631027222, + "num_tokens": 352245603.0, + "step": 13741 + }, + { + "epoch": 1.5091148693169338, + "grad_norm": 2.3530962467193604, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7104781270027161, + "num_tokens": 352270219.0, + "step": 13742 + }, + { + "epoch": 1.5092246870195476, + "grad_norm": 2.1177985668182373, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7013654708862305, + "num_tokens": 352299741.0, + "step": 13743 + }, + { + "epoch": 1.5093345047221614, + "grad_norm": 2.3651082515716553, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.697899580001831, + "num_tokens": 352324841.0, + "step": 13744 + }, + { + "epoch": 1.5094443224247749, + "grad_norm": 2.2582902908325195, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7165349721908569, + "num_tokens": 352350647.0, + "step": 13745 + }, + { + "epoch": 1.5095541401273884, + "grad_norm": 2.2252557277679443, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7242012023925781, + "num_tokens": 352375839.0, + "step": 13746 + }, + { + "epoch": 1.5096639578300022, + "grad_norm": 2.1125168800354004, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7287193536758423, + "num_tokens": 352402386.0, + "step": 13747 + }, + { + "epoch": 1.509773775532616, + "grad_norm": 2.1210267543792725, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7203913331031799, + "num_tokens": 352432292.0, + "step": 13748 + }, + { + "epoch": 1.5098835932352295, + "grad_norm": 2.2125720977783203, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7201957702636719, + "num_tokens": 352458308.0, + "step": 13749 + }, + { + "epoch": 1.5099934109378432, + "grad_norm": 2.289790630340576, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7378057241439819, + "num_tokens": 352482409.0, + "step": 13750 + }, + { + "epoch": 1.5101032286404568, + "grad_norm": 2.1902966499328613, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7180642485618591, + "num_tokens": 352510208.0, + "step": 13751 + }, + { + "epoch": 1.5102130463430705, + "grad_norm": 2.0114188194274902, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7237188220024109, + "num_tokens": 352541538.0, + "step": 13752 + }, + { + "epoch": 1.5103228640456843, + "grad_norm": 2.313135862350464, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7462475299835205, + "num_tokens": 352564867.0, + "step": 13753 + }, + { + "epoch": 1.5104326817482978, + "grad_norm": 2.056582450866699, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7043540477752686, + "num_tokens": 352595539.0, + "step": 13754 + }, + { + "epoch": 1.5105424994509113, + "grad_norm": 2.0565099716186523, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7211096882820129, + "num_tokens": 352622954.0, + "step": 13755 + }, + { + "epoch": 1.510652317153525, + "grad_norm": 2.522855758666992, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7384021878242493, + "num_tokens": 352644350.0, + "step": 13756 + }, + { + "epoch": 1.5107621348561389, + "grad_norm": 2.2555854320526123, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7178898453712463, + "num_tokens": 352668000.0, + "step": 13757 + }, + { + "epoch": 1.5108719525587526, + "grad_norm": 2.0517168045043945, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7029263973236084, + "num_tokens": 352696200.0, + "step": 13758 + }, + { + "epoch": 1.5109817702613662, + "grad_norm": 2.5128607749938965, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7591688632965088, + "num_tokens": 352717480.0, + "step": 13759 + }, + { + "epoch": 1.5110915879639797, + "grad_norm": 2.240163803100586, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6775466203689575, + "num_tokens": 352744961.0, + "step": 13760 + }, + { + "epoch": 1.5112014056665934, + "grad_norm": 2.358764410018921, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7408630847930908, + "num_tokens": 352767444.0, + "step": 13761 + }, + { + "epoch": 1.5113112233692072, + "grad_norm": 2.4380648136138916, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7361060976982117, + "num_tokens": 352790926.0, + "step": 13762 + }, + { + "epoch": 1.5114210410718207, + "grad_norm": 2.0897305011749268, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6909691691398621, + "num_tokens": 352821014.0, + "step": 13763 + }, + { + "epoch": 1.5115308587744345, + "grad_norm": 2.282961368560791, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7217684984207153, + "num_tokens": 352846695.0, + "step": 13764 + }, + { + "epoch": 1.511640676477048, + "grad_norm": 2.17287015914917, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7230715751647949, + "num_tokens": 352873931.0, + "step": 13765 + }, + { + "epoch": 1.5117504941796618, + "grad_norm": 2.1880338191986084, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7377294301986694, + "num_tokens": 352898850.0, + "step": 13766 + }, + { + "epoch": 1.5118603118822755, + "grad_norm": 2.1634812355041504, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7182519435882568, + "num_tokens": 352926004.0, + "step": 13767 + }, + { + "epoch": 1.511970129584889, + "grad_norm": 1.9875377416610718, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.725765585899353, + "num_tokens": 352956241.0, + "step": 13768 + }, + { + "epoch": 1.5120799472875026, + "grad_norm": 2.4109461307525635, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7086488008499146, + "num_tokens": 352979944.0, + "step": 13769 + }, + { + "epoch": 1.5121897649901164, + "grad_norm": 2.351440668106079, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7159132361412048, + "num_tokens": 353003982.0, + "step": 13770 + }, + { + "epoch": 1.5122995826927301, + "grad_norm": 1.968056321144104, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7172542214393616, + "num_tokens": 353037183.0, + "step": 13771 + }, + { + "epoch": 1.5124094003953439, + "grad_norm": 2.395228385925293, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7202747464179993, + "num_tokens": 353059457.0, + "step": 13772 + }, + { + "epoch": 1.5125192180979574, + "grad_norm": 2.102163553237915, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7038236856460571, + "num_tokens": 353087208.0, + "step": 13773 + }, + { + "epoch": 1.512629035800571, + "grad_norm": 2.1584606170654297, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7130767107009888, + "num_tokens": 353115777.0, + "step": 13774 + }, + { + "epoch": 1.5127388535031847, + "grad_norm": 2.5823347568511963, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7306824326515198, + "num_tokens": 353135727.0, + "step": 13775 + }, + { + "epoch": 1.5128486712057985, + "grad_norm": 1.8752866983413696, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7199105620384216, + "num_tokens": 353168013.0, + "step": 13776 + }, + { + "epoch": 1.512958488908412, + "grad_norm": 2.019973039627075, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7253714799880981, + "num_tokens": 353197635.0, + "step": 13777 + }, + { + "epoch": 1.5130683066110255, + "grad_norm": 2.405683755874634, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7123929858207703, + "num_tokens": 353221680.0, + "step": 13778 + }, + { + "epoch": 1.5131781243136393, + "grad_norm": 2.484161853790283, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7104011178016663, + "num_tokens": 353244325.0, + "step": 13779 + }, + { + "epoch": 1.513287942016253, + "grad_norm": 2.45588755607605, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7328412532806396, + "num_tokens": 353266866.0, + "step": 13780 + }, + { + "epoch": 1.5133977597188668, + "grad_norm": 2.3674330711364746, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.741754412651062, + "num_tokens": 353290461.0, + "step": 13781 + }, + { + "epoch": 1.5135075774214803, + "grad_norm": 2.35516619682312, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7233055830001831, + "num_tokens": 353314281.0, + "step": 13782 + }, + { + "epoch": 1.5136173951240939, + "grad_norm": 2.131303071975708, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7306178212165833, + "num_tokens": 353343538.0, + "step": 13783 + }, + { + "epoch": 1.5137272128267076, + "grad_norm": 2.5894601345062256, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.727745532989502, + "num_tokens": 353364143.0, + "step": 13784 + }, + { + "epoch": 1.5138370305293214, + "grad_norm": 1.9665952920913696, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6948106288909912, + "num_tokens": 353394485.0, + "step": 13785 + }, + { + "epoch": 1.5139468482319351, + "grad_norm": 2.325864791870117, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7413911819458008, + "num_tokens": 353419622.0, + "step": 13786 + }, + { + "epoch": 1.5140566659345487, + "grad_norm": 2.4809257984161377, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7308122515678406, + "num_tokens": 353440496.0, + "step": 13787 + }, + { + "epoch": 1.5141664836371622, + "grad_norm": 2.3795242309570312, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.711140513420105, + "num_tokens": 353466271.0, + "step": 13788 + }, + { + "epoch": 1.514276301339776, + "grad_norm": 2.25917387008667, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7244151830673218, + "num_tokens": 353492859.0, + "step": 13789 + }, + { + "epoch": 1.5143861190423897, + "grad_norm": 1.9874060153961182, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7273815274238586, + "num_tokens": 353522669.0, + "step": 13790 + }, + { + "epoch": 1.5144959367450033, + "grad_norm": 2.276921510696411, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7028118968009949, + "num_tokens": 353547437.0, + "step": 13791 + }, + { + "epoch": 1.5146057544476168, + "grad_norm": 2.0948312282562256, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7011361718177795, + "num_tokens": 353575285.0, + "step": 13792 + }, + { + "epoch": 1.5147155721502306, + "grad_norm": 2.2890186309814453, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7117785215377808, + "num_tokens": 353600828.0, + "step": 13793 + }, + { + "epoch": 1.5148253898528443, + "grad_norm": 1.9994685649871826, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7053858041763306, + "num_tokens": 353634949.0, + "step": 13794 + }, + { + "epoch": 1.514935207555458, + "grad_norm": 2.332388162612915, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7277078628540039, + "num_tokens": 353658321.0, + "step": 13795 + }, + { + "epoch": 1.5150450252580716, + "grad_norm": 2.3649487495422363, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7423601150512695, + "num_tokens": 353682110.0, + "step": 13796 + }, + { + "epoch": 1.5151548429606851, + "grad_norm": 2.2049386501312256, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7040156722068787, + "num_tokens": 353711692.0, + "step": 13797 + }, + { + "epoch": 1.515264660663299, + "grad_norm": 2.1971869468688965, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.722258985042572, + "num_tokens": 353736147.0, + "step": 13798 + }, + { + "epoch": 1.5153744783659127, + "grad_norm": 2.589571714401245, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7390260696411133, + "num_tokens": 353755387.0, + "step": 13799 + }, + { + "epoch": 1.5154842960685262, + "grad_norm": 2.207350730895996, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7238472104072571, + "num_tokens": 353780449.0, + "step": 13800 + }, + { + "epoch": 1.51559411377114, + "grad_norm": 2.231099843978882, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7169778347015381, + "num_tokens": 353805173.0, + "step": 13801 + }, + { + "epoch": 1.5157039314737535, + "grad_norm": 2.2606680393218994, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7128657102584839, + "num_tokens": 353832308.0, + "step": 13802 + }, + { + "epoch": 1.5158137491763672, + "grad_norm": 2.1350862979888916, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7245895862579346, + "num_tokens": 353859644.0, + "step": 13803 + }, + { + "epoch": 1.515923566878981, + "grad_norm": 2.0936620235443115, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7090185880661011, + "num_tokens": 353888372.0, + "step": 13804 + }, + { + "epoch": 1.5160333845815945, + "grad_norm": 1.9636415243148804, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7115374803543091, + "num_tokens": 353920115.0, + "step": 13805 + }, + { + "epoch": 1.516143202284208, + "grad_norm": 2.4424855709075928, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7368791103363037, + "num_tokens": 353942605.0, + "step": 13806 + }, + { + "epoch": 1.5162530199868218, + "grad_norm": 2.2109436988830566, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7165476083755493, + "num_tokens": 353968867.0, + "step": 13807 + }, + { + "epoch": 1.5163628376894356, + "grad_norm": 2.33852219581604, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7245392799377441, + "num_tokens": 353991812.0, + "step": 13808 + }, + { + "epoch": 1.5164726553920493, + "grad_norm": 2.3718135356903076, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7196561098098755, + "num_tokens": 354015805.0, + "step": 13809 + }, + { + "epoch": 1.5165824730946629, + "grad_norm": 2.921016216278076, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7431076765060425, + "num_tokens": 354033888.0, + "step": 13810 + }, + { + "epoch": 1.5166922907972764, + "grad_norm": 2.338099956512451, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7205801010131836, + "num_tokens": 354057260.0, + "step": 13811 + }, + { + "epoch": 1.5168021084998902, + "grad_norm": 2.2636220455169678, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7353061437606812, + "num_tokens": 354081181.0, + "step": 13812 + }, + { + "epoch": 1.516911926202504, + "grad_norm": 2.071129560470581, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7064860463142395, + "num_tokens": 354110240.0, + "step": 13813 + }, + { + "epoch": 1.5170217439051175, + "grad_norm": 2.1751725673675537, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7329380512237549, + "num_tokens": 354137560.0, + "step": 13814 + }, + { + "epoch": 1.5171315616077312, + "grad_norm": 2.2565956115722656, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7075921297073364, + "num_tokens": 354162739.0, + "step": 13815 + }, + { + "epoch": 1.5172413793103448, + "grad_norm": 2.263934850692749, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7268959283828735, + "num_tokens": 354187727.0, + "step": 13816 + }, + { + "epoch": 1.5173511970129585, + "grad_norm": 2.118039608001709, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7141848802566528, + "num_tokens": 354219291.0, + "step": 13817 + }, + { + "epoch": 1.5174610147155723, + "grad_norm": 2.2463507652282715, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7139745950698853, + "num_tokens": 354246280.0, + "step": 13818 + }, + { + "epoch": 1.5175708324181858, + "grad_norm": 2.303450107574463, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.721089780330658, + "num_tokens": 354270485.0, + "step": 13819 + }, + { + "epoch": 1.5176806501207993, + "grad_norm": 2.3086094856262207, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7224618196487427, + "num_tokens": 354294577.0, + "step": 13820 + }, + { + "epoch": 1.517790467823413, + "grad_norm": 2.2060959339141846, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7356218099594116, + "num_tokens": 354320632.0, + "step": 13821 + }, + { + "epoch": 1.5179002855260268, + "grad_norm": 1.9671183824539185, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7282590866088867, + "num_tokens": 354351255.0, + "step": 13822 + }, + { + "epoch": 1.5180101032286406, + "grad_norm": 2.0698115825653076, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7242988348007202, + "num_tokens": 354380711.0, + "step": 13823 + }, + { + "epoch": 1.5181199209312541, + "grad_norm": 2.5832343101501465, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7331089973449707, + "num_tokens": 354401318.0, + "step": 13824 + }, + { + "epoch": 1.5182297386338677, + "grad_norm": 2.7079737186431885, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7165069580078125, + "num_tokens": 354420781.0, + "step": 13825 + }, + { + "epoch": 1.5183395563364814, + "grad_norm": 2.240448474884033, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7237561941146851, + "num_tokens": 354447339.0, + "step": 13826 + }, + { + "epoch": 1.5184493740390952, + "grad_norm": 2.3923728466033936, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7478930354118347, + "num_tokens": 354469654.0, + "step": 13827 + }, + { + "epoch": 1.5185591917417087, + "grad_norm": 2.142359495162964, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.724077582359314, + "num_tokens": 354496262.0, + "step": 13828 + }, + { + "epoch": 1.5186690094443223, + "grad_norm": 2.465151786804199, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7028186321258545, + "num_tokens": 354518409.0, + "step": 13829 + }, + { + "epoch": 1.518778827146936, + "grad_norm": 2.551886558532715, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7471367716789246, + "num_tokens": 354538554.0, + "step": 13830 + }, + { + "epoch": 1.5188886448495498, + "grad_norm": 2.278353452682495, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7362302541732788, + "num_tokens": 354561564.0, + "step": 13831 + }, + { + "epoch": 1.5189984625521635, + "grad_norm": 2.1705586910247803, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7409136295318604, + "num_tokens": 354587420.0, + "step": 13832 + }, + { + "epoch": 1.519108280254777, + "grad_norm": 2.6176257133483887, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7217351198196411, + "num_tokens": 354608761.0, + "step": 13833 + }, + { + "epoch": 1.5192180979573906, + "grad_norm": 2.379211187362671, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7122936248779297, + "num_tokens": 354633993.0, + "step": 13834 + }, + { + "epoch": 1.5193279156600044, + "grad_norm": 2.195120334625244, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7115681767463684, + "num_tokens": 354661175.0, + "step": 13835 + }, + { + "epoch": 1.5194377333626181, + "grad_norm": 2.2674121856689453, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.734236478805542, + "num_tokens": 354684036.0, + "step": 13836 + }, + { + "epoch": 1.5195475510652319, + "grad_norm": 2.3133456707000732, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7197136878967285, + "num_tokens": 354707193.0, + "step": 13837 + }, + { + "epoch": 1.5196573687678454, + "grad_norm": 2.3775768280029297, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.741152286529541, + "num_tokens": 354728044.0, + "step": 13838 + }, + { + "epoch": 1.519767186470459, + "grad_norm": 2.2902672290802, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7160205245018005, + "num_tokens": 354754388.0, + "step": 13839 + }, + { + "epoch": 1.5198770041730727, + "grad_norm": 2.316985845565796, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7171081304550171, + "num_tokens": 354777045.0, + "step": 13840 + }, + { + "epoch": 1.5199868218756865, + "grad_norm": 2.055832624435425, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7232968807220459, + "num_tokens": 354806806.0, + "step": 13841 + }, + { + "epoch": 1.5200966395783, + "grad_norm": 2.1725988388061523, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6911610960960388, + "num_tokens": 354834360.0, + "step": 13842 + }, + { + "epoch": 1.5202064572809135, + "grad_norm": 2.163742780685425, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7058669328689575, + "num_tokens": 354862181.0, + "step": 13843 + }, + { + "epoch": 1.5203162749835273, + "grad_norm": 2.161459445953369, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7109805941581726, + "num_tokens": 354890227.0, + "step": 13844 + }, + { + "epoch": 1.520426092686141, + "grad_norm": 2.3844501972198486, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7442182302474976, + "num_tokens": 354912391.0, + "step": 13845 + }, + { + "epoch": 1.5205359103887548, + "grad_norm": 2.378101110458374, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7072352170944214, + "num_tokens": 354940038.0, + "step": 13846 + }, + { + "epoch": 1.5206457280913683, + "grad_norm": 2.2143800258636475, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7162175178527832, + "num_tokens": 354965920.0, + "step": 13847 + }, + { + "epoch": 1.5207555457939819, + "grad_norm": 2.0665957927703857, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7138084173202515, + "num_tokens": 354995539.0, + "step": 13848 + }, + { + "epoch": 1.5208653634965956, + "grad_norm": 2.335313320159912, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7002571821212769, + "num_tokens": 355021854.0, + "step": 13849 + }, + { + "epoch": 1.5209751811992094, + "grad_norm": 2.3775107860565186, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7254531383514404, + "num_tokens": 355044256.0, + "step": 13850 + }, + { + "epoch": 1.5210849989018231, + "grad_norm": 2.0031211376190186, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6953202486038208, + "num_tokens": 355077310.0, + "step": 13851 + }, + { + "epoch": 1.5211948166044367, + "grad_norm": 2.3997433185577393, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7185649275779724, + "num_tokens": 355101443.0, + "step": 13852 + }, + { + "epoch": 1.5213046343070502, + "grad_norm": 2.3253040313720703, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7283567190170288, + "num_tokens": 355124682.0, + "step": 13853 + }, + { + "epoch": 1.521414452009664, + "grad_norm": 2.463301420211792, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6943788528442383, + "num_tokens": 355147519.0, + "step": 13854 + }, + { + "epoch": 1.5215242697122777, + "grad_norm": 2.046414852142334, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7231466770172119, + "num_tokens": 355178182.0, + "step": 13855 + }, + { + "epoch": 1.5216340874148913, + "grad_norm": 2.7353529930114746, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7229032516479492, + "num_tokens": 355196361.0, + "step": 13856 + }, + { + "epoch": 1.5217439051175048, + "grad_norm": 2.632237195968628, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7319303750991821, + "num_tokens": 355216201.0, + "step": 13857 + }, + { + "epoch": 1.5218537228201185, + "grad_norm": 2.391632556915283, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7231788039207458, + "num_tokens": 355239200.0, + "step": 13858 + }, + { + "epoch": 1.5219635405227323, + "grad_norm": 2.2290873527526855, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.716254472732544, + "num_tokens": 355264712.0, + "step": 13859 + }, + { + "epoch": 1.522073358225346, + "grad_norm": 2.2033028602600098, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.705916166305542, + "num_tokens": 355291158.0, + "step": 13860 + }, + { + "epoch": 1.5221831759279596, + "grad_norm": 2.3217625617980957, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7052531242370605, + "num_tokens": 355316127.0, + "step": 13861 + }, + { + "epoch": 1.5222929936305731, + "grad_norm": 2.31707501411438, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7481536865234375, + "num_tokens": 355337409.0, + "step": 13862 + }, + { + "epoch": 1.522402811333187, + "grad_norm": 2.4637527465820312, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7126536965370178, + "num_tokens": 355359014.0, + "step": 13863 + }, + { + "epoch": 1.5225126290358006, + "grad_norm": 1.9799667596817017, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7123634219169617, + "num_tokens": 355389860.0, + "step": 13864 + }, + { + "epoch": 1.5226224467384142, + "grad_norm": 1.9912968873977661, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7014508247375488, + "num_tokens": 355420360.0, + "step": 13865 + }, + { + "epoch": 1.522732264441028, + "grad_norm": 2.2652103900909424, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7108822464942932, + "num_tokens": 355444688.0, + "step": 13866 + }, + { + "epoch": 1.5228420821436415, + "grad_norm": 2.1512222290039062, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.698567807674408, + "num_tokens": 355472359.0, + "step": 13867 + }, + { + "epoch": 1.5229518998462552, + "grad_norm": 2.1576154232025146, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7109982371330261, + "num_tokens": 355498576.0, + "step": 13868 + }, + { + "epoch": 1.523061717548869, + "grad_norm": 2.3163692951202393, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7371310591697693, + "num_tokens": 355520782.0, + "step": 13869 + }, + { + "epoch": 1.5231715352514825, + "grad_norm": 2.5938377380371094, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7343243360519409, + "num_tokens": 355541888.0, + "step": 13870 + }, + { + "epoch": 1.523281352954096, + "grad_norm": 2.563537120819092, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7298007011413574, + "num_tokens": 355563087.0, + "step": 13871 + }, + { + "epoch": 1.5233911706567098, + "grad_norm": 2.404524326324463, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7303336262702942, + "num_tokens": 355584369.0, + "step": 13872 + }, + { + "epoch": 1.5235009883593236, + "grad_norm": 2.33413028717041, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7044128775596619, + "num_tokens": 355611271.0, + "step": 13873 + }, + { + "epoch": 1.5236108060619373, + "grad_norm": 2.356496572494507, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7361932396888733, + "num_tokens": 355633460.0, + "step": 13874 + }, + { + "epoch": 1.5237206237645509, + "grad_norm": 2.302673816680908, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7100999355316162, + "num_tokens": 355659180.0, + "step": 13875 + }, + { + "epoch": 1.5238304414671644, + "grad_norm": 2.311840772628784, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7033261656761169, + "num_tokens": 355689035.0, + "step": 13876 + }, + { + "epoch": 1.5239402591697782, + "grad_norm": 2.6187500953674316, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7245435118675232, + "num_tokens": 355708706.0, + "step": 13877 + }, + { + "epoch": 1.524050076872392, + "grad_norm": 2.287292003631592, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7109172344207764, + "num_tokens": 355733498.0, + "step": 13878 + }, + { + "epoch": 1.5241598945750054, + "grad_norm": 2.0661933422088623, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.6984647512435913, + "num_tokens": 355764349.0, + "step": 13879 + }, + { + "epoch": 1.5242697122776192, + "grad_norm": 2.300726890563965, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7015531063079834, + "num_tokens": 355787263.0, + "step": 13880 + }, + { + "epoch": 1.5243795299802327, + "grad_norm": 2.292853832244873, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7589403390884399, + "num_tokens": 355808649.0, + "step": 13881 + }, + { + "epoch": 1.5244893476828465, + "grad_norm": 2.3566267490386963, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7159472703933716, + "num_tokens": 355832381.0, + "step": 13882 + }, + { + "epoch": 1.5245991653854603, + "grad_norm": 2.4360713958740234, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7054071426391602, + "num_tokens": 355854471.0, + "step": 13883 + }, + { + "epoch": 1.5247089830880738, + "grad_norm": 2.3799800872802734, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7247934341430664, + "num_tokens": 355875879.0, + "step": 13884 + }, + { + "epoch": 1.5248188007906873, + "grad_norm": 2.2864253520965576, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7281872034072876, + "num_tokens": 355900532.0, + "step": 13885 + }, + { + "epoch": 1.524928618493301, + "grad_norm": 2.2859654426574707, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.748481035232544, + "num_tokens": 355923691.0, + "step": 13886 + }, + { + "epoch": 1.5250384361959148, + "grad_norm": 2.415065050125122, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7237228751182556, + "num_tokens": 355944724.0, + "step": 13887 + }, + { + "epoch": 1.5251482538985286, + "grad_norm": 2.296286106109619, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7123069763183594, + "num_tokens": 355968688.0, + "step": 13888 + }, + { + "epoch": 1.5252580716011421, + "grad_norm": 2.25671124458313, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7214556932449341, + "num_tokens": 355993600.0, + "step": 13889 + }, + { + "epoch": 1.5253678893037557, + "grad_norm": 2.217843532562256, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7140387892723083, + "num_tokens": 356019962.0, + "step": 13890 + }, + { + "epoch": 1.5254777070063694, + "grad_norm": 2.2754201889038086, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.700187087059021, + "num_tokens": 356044631.0, + "step": 13891 + }, + { + "epoch": 1.5255875247089832, + "grad_norm": 2.2814736366271973, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7124868631362915, + "num_tokens": 356070154.0, + "step": 13892 + }, + { + "epoch": 1.5256973424115967, + "grad_norm": 2.4373209476470947, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7140570282936096, + "num_tokens": 356093989.0, + "step": 13893 + }, + { + "epoch": 1.5258071601142102, + "grad_norm": 2.235541343688965, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7004470229148865, + "num_tokens": 356119788.0, + "step": 13894 + }, + { + "epoch": 1.525916977816824, + "grad_norm": 2.149460554122925, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7127326726913452, + "num_tokens": 356147071.0, + "step": 13895 + }, + { + "epoch": 1.5260267955194378, + "grad_norm": 2.0761654376983643, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7075883150100708, + "num_tokens": 356175455.0, + "step": 13896 + }, + { + "epoch": 1.5261366132220515, + "grad_norm": 2.446580410003662, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7319391965866089, + "num_tokens": 356196802.0, + "step": 13897 + }, + { + "epoch": 1.526246430924665, + "grad_norm": 2.434901714324951, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7296864986419678, + "num_tokens": 356218995.0, + "step": 13898 + }, + { + "epoch": 1.5263562486272786, + "grad_norm": 2.3020710945129395, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7161585092544556, + "num_tokens": 356242587.0, + "step": 13899 + }, + { + "epoch": 1.5264660663298923, + "grad_norm": 2.5168612003326416, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7329370975494385, + "num_tokens": 356263189.0, + "step": 13900 + }, + { + "epoch": 1.526575884032506, + "grad_norm": 2.3074722290039062, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7126154899597168, + "num_tokens": 356288023.0, + "step": 13901 + }, + { + "epoch": 1.5266857017351199, + "grad_norm": 2.359379768371582, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7452960014343262, + "num_tokens": 356310521.0, + "step": 13902 + }, + { + "epoch": 1.5267955194377334, + "grad_norm": 2.316079616546631, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7141245603561401, + "num_tokens": 356335194.0, + "step": 13903 + }, + { + "epoch": 1.526905337140347, + "grad_norm": 2.1766130924224854, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7143312096595764, + "num_tokens": 356361271.0, + "step": 13904 + }, + { + "epoch": 1.5270151548429607, + "grad_norm": 2.5741145610809326, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7154523134231567, + "num_tokens": 356382114.0, + "step": 13905 + }, + { + "epoch": 1.5271249725455744, + "grad_norm": 2.395883083343506, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7183341979980469, + "num_tokens": 356404721.0, + "step": 13906 + }, + { + "epoch": 1.527234790248188, + "grad_norm": 2.031209945678711, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7302176356315613, + "num_tokens": 356434734.0, + "step": 13907 + }, + { + "epoch": 1.5273446079508015, + "grad_norm": 2.4984655380249023, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7276049852371216, + "num_tokens": 356457041.0, + "step": 13908 + }, + { + "epoch": 1.5274544256534153, + "grad_norm": 2.3318734169006348, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7433725595474243, + "num_tokens": 356480699.0, + "step": 13909 + }, + { + "epoch": 1.527564243356029, + "grad_norm": 2.740630626678467, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7371536493301392, + "num_tokens": 356499593.0, + "step": 13910 + }, + { + "epoch": 1.5276740610586428, + "grad_norm": 2.42746639251709, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7359553575515747, + "num_tokens": 356521771.0, + "step": 13911 + }, + { + "epoch": 1.5277838787612563, + "grad_norm": 2.1302337646484375, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.725051760673523, + "num_tokens": 356548736.0, + "step": 13912 + }, + { + "epoch": 1.5278936964638699, + "grad_norm": 2.212754487991333, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7143407464027405, + "num_tokens": 356575290.0, + "step": 13913 + }, + { + "epoch": 1.5280035141664836, + "grad_norm": 2.4633889198303223, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7201004028320312, + "num_tokens": 356598853.0, + "step": 13914 + }, + { + "epoch": 1.5281133318690974, + "grad_norm": 2.203416347503662, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7254801988601685, + "num_tokens": 356625093.0, + "step": 13915 + }, + { + "epoch": 1.5282231495717111, + "grad_norm": 2.307826519012451, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7087434530258179, + "num_tokens": 356650967.0, + "step": 13916 + }, + { + "epoch": 1.5283329672743247, + "grad_norm": 2.0862224102020264, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7130628228187561, + "num_tokens": 356680902.0, + "step": 13917 + }, + { + "epoch": 1.5284427849769382, + "grad_norm": 2.348992347717285, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7238600850105286, + "num_tokens": 356705008.0, + "step": 13918 + }, + { + "epoch": 1.528552602679552, + "grad_norm": 2.4765329360961914, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7402230501174927, + "num_tokens": 356727189.0, + "step": 13919 + }, + { + "epoch": 1.5286624203821657, + "grad_norm": 2.4157400131225586, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.726070761680603, + "num_tokens": 356749895.0, + "step": 13920 + }, + { + "epoch": 1.5287722380847792, + "grad_norm": 1.8293873071670532, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7053387761116028, + "num_tokens": 356787693.0, + "step": 13921 + }, + { + "epoch": 1.5288820557873928, + "grad_norm": 2.4507458209991455, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.729366660118103, + "num_tokens": 356809048.0, + "step": 13922 + }, + { + "epoch": 1.5289918734900065, + "grad_norm": 2.369346857070923, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7534921169281006, + "num_tokens": 356832419.0, + "step": 13923 + }, + { + "epoch": 1.5291016911926203, + "grad_norm": 2.193674087524414, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7019264101982117, + "num_tokens": 356861274.0, + "step": 13924 + }, + { + "epoch": 1.529211508895234, + "grad_norm": 2.502577781677246, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7285080552101135, + "num_tokens": 356881598.0, + "step": 13925 + }, + { + "epoch": 1.5293213265978476, + "grad_norm": 2.160102605819702, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7497680187225342, + "num_tokens": 356905934.0, + "step": 13926 + }, + { + "epoch": 1.5294311443004611, + "grad_norm": 2.2432303428649902, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.720038652420044, + "num_tokens": 356930859.0, + "step": 13927 + }, + { + "epoch": 1.5295409620030749, + "grad_norm": 2.096674680709839, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7148175835609436, + "num_tokens": 356959425.0, + "step": 13928 + }, + { + "epoch": 1.5296507797056886, + "grad_norm": 2.401789665222168, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.733476996421814, + "num_tokens": 356983036.0, + "step": 13929 + }, + { + "epoch": 1.5297605974083022, + "grad_norm": 2.1196370124816895, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7110230922698975, + "num_tokens": 357011504.0, + "step": 13930 + }, + { + "epoch": 1.529870415110916, + "grad_norm": 2.1480071544647217, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7351983785629272, + "num_tokens": 357038473.0, + "step": 13931 + }, + { + "epoch": 1.5299802328135295, + "grad_norm": 2.3423566818237305, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7424907684326172, + "num_tokens": 357060106.0, + "step": 13932 + }, + { + "epoch": 1.5300900505161432, + "grad_norm": 2.4508140087127686, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7323589324951172, + "num_tokens": 357080478.0, + "step": 13933 + }, + { + "epoch": 1.530199868218757, + "grad_norm": 2.488614559173584, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7378222942352295, + "num_tokens": 357103025.0, + "step": 13934 + }, + { + "epoch": 1.5303096859213705, + "grad_norm": 2.654005527496338, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7294235229492188, + "num_tokens": 357124749.0, + "step": 13935 + }, + { + "epoch": 1.530419503623984, + "grad_norm": 2.1957192420959473, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7106562256813049, + "num_tokens": 357153593.0, + "step": 13936 + }, + { + "epoch": 1.5305293213265978, + "grad_norm": 2.212846517562866, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7108772993087769, + "num_tokens": 357178952.0, + "step": 13937 + }, + { + "epoch": 1.5306391390292116, + "grad_norm": 2.14681077003479, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7392043471336365, + "num_tokens": 357204720.0, + "step": 13938 + }, + { + "epoch": 1.5307489567318253, + "grad_norm": 2.2543587684631348, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7194246053695679, + "num_tokens": 357229291.0, + "step": 13939 + }, + { + "epoch": 1.5308587744344389, + "grad_norm": 1.945663571357727, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7146287560462952, + "num_tokens": 357261184.0, + "step": 13940 + }, + { + "epoch": 1.5309685921370524, + "grad_norm": 2.024827241897583, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.6757545471191406, + "num_tokens": 357297061.0, + "step": 13941 + }, + { + "epoch": 1.5310784098396661, + "grad_norm": 2.3718490600585938, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7094553709030151, + "num_tokens": 357319865.0, + "step": 13942 + }, + { + "epoch": 1.53118822754228, + "grad_norm": 2.179539442062378, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.730229914188385, + "num_tokens": 357346187.0, + "step": 13943 + }, + { + "epoch": 1.5312980452448934, + "grad_norm": 2.0905091762542725, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7052695751190186, + "num_tokens": 357375691.0, + "step": 13944 + }, + { + "epoch": 1.5314078629475072, + "grad_norm": 2.147367238998413, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7153542041778564, + "num_tokens": 357404393.0, + "step": 13945 + }, + { + "epoch": 1.5315176806501207, + "grad_norm": 2.401181697845459, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.715350329875946, + "num_tokens": 357426213.0, + "step": 13946 + }, + { + "epoch": 1.5316274983527345, + "grad_norm": 2.019479513168335, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7295541167259216, + "num_tokens": 357456220.0, + "step": 13947 + }, + { + "epoch": 1.5317373160553482, + "grad_norm": 2.0252132415771484, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.721908688545227, + "num_tokens": 357489492.0, + "step": 13948 + }, + { + "epoch": 1.5318471337579618, + "grad_norm": 2.1768643856048584, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.6999526023864746, + "num_tokens": 357516403.0, + "step": 13949 + }, + { + "epoch": 1.5319569514605753, + "grad_norm": 2.3081471920013428, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7160835266113281, + "num_tokens": 357541413.0, + "step": 13950 + }, + { + "epoch": 1.532066769163189, + "grad_norm": 2.353027820587158, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7193043828010559, + "num_tokens": 357564992.0, + "step": 13951 + }, + { + "epoch": 1.5321765868658028, + "grad_norm": 1.8962701559066772, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6781438589096069, + "num_tokens": 357603176.0, + "step": 13952 + }, + { + "epoch": 1.5322864045684166, + "grad_norm": 2.3043854236602783, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7102752923965454, + "num_tokens": 357630193.0, + "step": 13953 + }, + { + "epoch": 1.5323962222710301, + "grad_norm": 2.2473838329315186, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7387347221374512, + "num_tokens": 357655159.0, + "step": 13954 + }, + { + "epoch": 1.5325060399736437, + "grad_norm": 2.5348691940307617, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7095662355422974, + "num_tokens": 357678727.0, + "step": 13955 + }, + { + "epoch": 1.5326158576762574, + "grad_norm": 2.2220029830932617, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7159320712089539, + "num_tokens": 357704448.0, + "step": 13956 + }, + { + "epoch": 1.5327256753788712, + "grad_norm": 2.179685115814209, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7132344245910645, + "num_tokens": 357732353.0, + "step": 13957 + }, + { + "epoch": 1.5328354930814847, + "grad_norm": 2.044344186782837, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7103210687637329, + "num_tokens": 357762158.0, + "step": 13958 + }, + { + "epoch": 1.5329453107840982, + "grad_norm": 2.3322837352752686, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7239886522293091, + "num_tokens": 357785704.0, + "step": 13959 + }, + { + "epoch": 1.533055128486712, + "grad_norm": 2.14797043800354, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7262160778045654, + "num_tokens": 357813573.0, + "step": 13960 + }, + { + "epoch": 1.5331649461893258, + "grad_norm": 2.1635429859161377, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7236636877059937, + "num_tokens": 357841193.0, + "step": 13961 + }, + { + "epoch": 1.5332747638919395, + "grad_norm": 2.1073555946350098, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6969868540763855, + "num_tokens": 357869002.0, + "step": 13962 + }, + { + "epoch": 1.533384581594553, + "grad_norm": 2.441551446914673, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7319225072860718, + "num_tokens": 357891581.0, + "step": 13963 + }, + { + "epoch": 1.5334943992971666, + "grad_norm": 2.135571002960205, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6976000070571899, + "num_tokens": 357923146.0, + "step": 13964 + }, + { + "epoch": 1.5336042169997803, + "grad_norm": 2.569255828857422, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7109308242797852, + "num_tokens": 357942909.0, + "step": 13965 + }, + { + "epoch": 1.533714034702394, + "grad_norm": 2.446491241455078, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7160471081733704, + "num_tokens": 357966605.0, + "step": 13966 + }, + { + "epoch": 1.5338238524050078, + "grad_norm": 2.3835630416870117, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6904159188270569, + "num_tokens": 357992414.0, + "step": 13967 + }, + { + "epoch": 1.5339336701076214, + "grad_norm": 2.372816562652588, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.715043842792511, + "num_tokens": 358015654.0, + "step": 13968 + }, + { + "epoch": 1.534043487810235, + "grad_norm": 2.4164254665374756, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7252035140991211, + "num_tokens": 358036373.0, + "step": 13969 + }, + { + "epoch": 1.5341533055128487, + "grad_norm": 2.0706706047058105, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7023578882217407, + "num_tokens": 358066263.0, + "step": 13970 + }, + { + "epoch": 1.5342631232154624, + "grad_norm": 2.8214194774627686, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7247439622879028, + "num_tokens": 358084976.0, + "step": 13971 + }, + { + "epoch": 1.534372940918076, + "grad_norm": 2.028808355331421, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6971805095672607, + "num_tokens": 358117073.0, + "step": 13972 + }, + { + "epoch": 1.5344827586206895, + "grad_norm": 2.5549869537353516, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7520581483840942, + "num_tokens": 358137250.0, + "step": 13973 + }, + { + "epoch": 1.5345925763233033, + "grad_norm": 1.9936054944992065, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7020938396453857, + "num_tokens": 358167762.0, + "step": 13974 + }, + { + "epoch": 1.534702394025917, + "grad_norm": 2.533474922180176, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7320080399513245, + "num_tokens": 358187876.0, + "step": 13975 + }, + { + "epoch": 1.5348122117285308, + "grad_norm": 2.2298636436462402, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7214428186416626, + "num_tokens": 358212330.0, + "step": 13976 + }, + { + "epoch": 1.5349220294311443, + "grad_norm": 1.9783819913864136, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6922867298126221, + "num_tokens": 358243992.0, + "step": 13977 + }, + { + "epoch": 1.5350318471337578, + "grad_norm": 2.303490161895752, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.72789466381073, + "num_tokens": 358267490.0, + "step": 13978 + }, + { + "epoch": 1.5351416648363716, + "grad_norm": 2.355867862701416, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7112525701522827, + "num_tokens": 358293340.0, + "step": 13979 + }, + { + "epoch": 1.5352514825389854, + "grad_norm": 2.4384422302246094, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7214445471763611, + "num_tokens": 358316600.0, + "step": 13980 + }, + { + "epoch": 1.535361300241599, + "grad_norm": 2.703409433364868, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.727257251739502, + "num_tokens": 358336524.0, + "step": 13981 + }, + { + "epoch": 1.5354711179442126, + "grad_norm": 2.2872471809387207, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7207602858543396, + "num_tokens": 358361761.0, + "step": 13982 + }, + { + "epoch": 1.5355809356468262, + "grad_norm": 2.301265001296997, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7087291479110718, + "num_tokens": 358388581.0, + "step": 13983 + }, + { + "epoch": 1.53569075334944, + "grad_norm": 2.1442224979400635, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7080858945846558, + "num_tokens": 358417188.0, + "step": 13984 + }, + { + "epoch": 1.5358005710520537, + "grad_norm": 2.3878884315490723, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7231930494308472, + "num_tokens": 358440712.0, + "step": 13985 + }, + { + "epoch": 1.5359103887546672, + "grad_norm": 2.352968692779541, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7007192373275757, + "num_tokens": 358465306.0, + "step": 13986 + }, + { + "epoch": 1.5360202064572808, + "grad_norm": 2.2074811458587646, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7117782831192017, + "num_tokens": 358490258.0, + "step": 13987 + }, + { + "epoch": 1.5361300241598945, + "grad_norm": 2.419640302658081, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7294853925704956, + "num_tokens": 358512500.0, + "step": 13988 + }, + { + "epoch": 1.5362398418625083, + "grad_norm": 2.0090088844299316, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7300843000411987, + "num_tokens": 358541458.0, + "step": 13989 + }, + { + "epoch": 1.536349659565122, + "grad_norm": 2.2096540927886963, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7055262923240662, + "num_tokens": 358568387.0, + "step": 13990 + }, + { + "epoch": 1.5364594772677356, + "grad_norm": 2.1759603023529053, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7149773836135864, + "num_tokens": 358596228.0, + "step": 13991 + }, + { + "epoch": 1.536569294970349, + "grad_norm": 2.233482599258423, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6956385374069214, + "num_tokens": 358621887.0, + "step": 13992 + }, + { + "epoch": 1.5366791126729629, + "grad_norm": 2.3991000652313232, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7134792804718018, + "num_tokens": 358645619.0, + "step": 13993 + }, + { + "epoch": 1.5367889303755766, + "grad_norm": 2.453047037124634, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7296413779258728, + "num_tokens": 358667494.0, + "step": 13994 + }, + { + "epoch": 1.5368987480781902, + "grad_norm": 2.2564899921417236, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6968297958374023, + "num_tokens": 358693243.0, + "step": 13995 + }, + { + "epoch": 1.537008565780804, + "grad_norm": 2.2107601165771484, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7235139012336731, + "num_tokens": 358718683.0, + "step": 13996 + }, + { + "epoch": 1.5371183834834174, + "grad_norm": 2.3579680919647217, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7524387836456299, + "num_tokens": 358740488.0, + "step": 13997 + }, + { + "epoch": 1.5372282011860312, + "grad_norm": 1.945420503616333, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7431900501251221, + "num_tokens": 358772812.0, + "step": 13998 + }, + { + "epoch": 1.537338018888645, + "grad_norm": 2.257337808609009, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.725479006767273, + "num_tokens": 358797943.0, + "step": 13999 + }, + { + "epoch": 1.5374478365912585, + "grad_norm": 2.256361961364746, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7002611756324768, + "num_tokens": 358821762.0, + "step": 14000 + }, + { + "epoch": 1.537557654293872, + "grad_norm": 2.6389999389648438, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7211213111877441, + "num_tokens": 358841432.0, + "step": 14001 + }, + { + "epoch": 1.5376674719964858, + "grad_norm": 1.9594330787658691, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7156075239181519, + "num_tokens": 358871938.0, + "step": 14002 + }, + { + "epoch": 1.5377772896990995, + "grad_norm": 2.2455666065216064, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7199698686599731, + "num_tokens": 358897871.0, + "step": 14003 + }, + { + "epoch": 1.5378871074017133, + "grad_norm": 2.4381747245788574, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7130016088485718, + "num_tokens": 358920636.0, + "step": 14004 + }, + { + "epoch": 1.5379969251043268, + "grad_norm": 2.2818334102630615, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7251436710357666, + "num_tokens": 358946082.0, + "step": 14005 + }, + { + "epoch": 1.5381067428069404, + "grad_norm": 2.205148696899414, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7023038864135742, + "num_tokens": 358974067.0, + "step": 14006 + }, + { + "epoch": 1.5382165605095541, + "grad_norm": 2.1419811248779297, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7476912140846252, + "num_tokens": 358999581.0, + "step": 14007 + }, + { + "epoch": 1.5383263782121679, + "grad_norm": 2.262411117553711, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7219259738922119, + "num_tokens": 359024302.0, + "step": 14008 + }, + { + "epoch": 1.5384361959147814, + "grad_norm": 2.132958173751831, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.739332914352417, + "num_tokens": 359052560.0, + "step": 14009 + }, + { + "epoch": 1.538546013617395, + "grad_norm": 1.9564564228057861, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6883008480072021, + "num_tokens": 359085512.0, + "step": 14010 + }, + { + "epoch": 1.5386558313200087, + "grad_norm": 2.212200880050659, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7294704914093018, + "num_tokens": 359110099.0, + "step": 14011 + }, + { + "epoch": 1.5387656490226225, + "grad_norm": 2.3847498893737793, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7369778752326965, + "num_tokens": 359132050.0, + "step": 14012 + }, + { + "epoch": 1.5388754667252362, + "grad_norm": 2.3377113342285156, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7055370807647705, + "num_tokens": 359156220.0, + "step": 14013 + }, + { + "epoch": 1.5389852844278498, + "grad_norm": 2.0815155506134033, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7133817672729492, + "num_tokens": 359185899.0, + "step": 14014 + }, + { + "epoch": 1.5390951021304633, + "grad_norm": 2.473942279815674, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7530505657196045, + "num_tokens": 359208435.0, + "step": 14015 + }, + { + "epoch": 1.539204919833077, + "grad_norm": 2.5252768993377686, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7472040057182312, + "num_tokens": 359230309.0, + "step": 14016 + }, + { + "epoch": 1.5393147375356908, + "grad_norm": 2.337218999862671, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7257778644561768, + "num_tokens": 359253607.0, + "step": 14017 + }, + { + "epoch": 1.5394245552383046, + "grad_norm": 2.1834583282470703, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7157750129699707, + "num_tokens": 359280369.0, + "step": 14018 + }, + { + "epoch": 1.539534372940918, + "grad_norm": 2.603064775466919, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7409465312957764, + "num_tokens": 359301331.0, + "step": 14019 + }, + { + "epoch": 1.5396441906435316, + "grad_norm": 2.210880756378174, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7183560132980347, + "num_tokens": 359328386.0, + "step": 14020 + }, + { + "epoch": 1.5397540083461454, + "grad_norm": 2.280604124069214, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7020442485809326, + "num_tokens": 359354262.0, + "step": 14021 + }, + { + "epoch": 1.5398638260487592, + "grad_norm": 2.1530115604400635, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7143685817718506, + "num_tokens": 359380485.0, + "step": 14022 + }, + { + "epoch": 1.5399736437513727, + "grad_norm": 2.1407177448272705, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7218722105026245, + "num_tokens": 359409113.0, + "step": 14023 + }, + { + "epoch": 1.5400834614539862, + "grad_norm": 2.7151987552642822, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7442039251327515, + "num_tokens": 359427640.0, + "step": 14024 + }, + { + "epoch": 1.5401932791566, + "grad_norm": 2.42966365814209, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7309262156486511, + "num_tokens": 359449709.0, + "step": 14025 + }, + { + "epoch": 1.5403030968592137, + "grad_norm": 1.8194334506988525, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6846381425857544, + "num_tokens": 359485250.0, + "step": 14026 + }, + { + "epoch": 1.5404129145618275, + "grad_norm": 2.0639326572418213, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7117102742195129, + "num_tokens": 359515474.0, + "step": 14027 + }, + { + "epoch": 1.540522732264441, + "grad_norm": 2.4681220054626465, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.723189115524292, + "num_tokens": 359536876.0, + "step": 14028 + }, + { + "epoch": 1.5406325499670546, + "grad_norm": 1.8941787481307983, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7165958881378174, + "num_tokens": 359570862.0, + "step": 14029 + }, + { + "epoch": 1.5407423676696683, + "grad_norm": 2.300990343093872, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.6988098621368408, + "num_tokens": 359594598.0, + "step": 14030 + }, + { + "epoch": 1.540852185372282, + "grad_norm": 2.037407875061035, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7077627182006836, + "num_tokens": 359624658.0, + "step": 14031 + }, + { + "epoch": 1.5409620030748958, + "grad_norm": 2.2110769748687744, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7345736026763916, + "num_tokens": 359649267.0, + "step": 14032 + }, + { + "epoch": 1.5410718207775094, + "grad_norm": 2.398465871810913, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7292959094047546, + "num_tokens": 359671816.0, + "step": 14033 + }, + { + "epoch": 1.541181638480123, + "grad_norm": 2.3939476013183594, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.715685248374939, + "num_tokens": 359695301.0, + "step": 14034 + }, + { + "epoch": 1.5412914561827367, + "grad_norm": 2.321835994720459, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7068068981170654, + "num_tokens": 359720555.0, + "step": 14035 + }, + { + "epoch": 1.5414012738853504, + "grad_norm": 2.3720381259918213, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7270249128341675, + "num_tokens": 359744450.0, + "step": 14036 + }, + { + "epoch": 1.541511091587964, + "grad_norm": 1.9687540531158447, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7045093774795532, + "num_tokens": 359777126.0, + "step": 14037 + }, + { + "epoch": 1.5416209092905775, + "grad_norm": 2.5368852615356445, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7427101135253906, + "num_tokens": 359797083.0, + "step": 14038 + }, + { + "epoch": 1.5417307269931912, + "grad_norm": 2.5049655437469482, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7074321508407593, + "num_tokens": 359818578.0, + "step": 14039 + }, + { + "epoch": 1.541840544695805, + "grad_norm": 2.7198054790496826, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7200304865837097, + "num_tokens": 359837963.0, + "step": 14040 + }, + { + "epoch": 1.5419503623984188, + "grad_norm": 2.0599498748779297, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7271483540534973, + "num_tokens": 359867555.0, + "step": 14041 + }, + { + "epoch": 1.5420601801010323, + "grad_norm": 2.631565570831299, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7140031456947327, + "num_tokens": 359889675.0, + "step": 14042 + }, + { + "epoch": 1.5421699978036458, + "grad_norm": 2.3315396308898926, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.73212069272995, + "num_tokens": 359912680.0, + "step": 14043 + }, + { + "epoch": 1.5422798155062596, + "grad_norm": 2.0566070079803467, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7052604556083679, + "num_tokens": 359942354.0, + "step": 14044 + }, + { + "epoch": 1.5423896332088733, + "grad_norm": 2.521357774734497, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7073487043380737, + "num_tokens": 359963969.0, + "step": 14045 + }, + { + "epoch": 1.5424994509114869, + "grad_norm": 2.31982159614563, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7006381750106812, + "num_tokens": 359989410.0, + "step": 14046 + }, + { + "epoch": 1.5426092686141006, + "grad_norm": 2.0985119342803955, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7142082452774048, + "num_tokens": 360017117.0, + "step": 14047 + }, + { + "epoch": 1.5427190863167142, + "grad_norm": 2.2689425945281982, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7155759334564209, + "num_tokens": 360043070.0, + "step": 14048 + }, + { + "epoch": 1.542828904019328, + "grad_norm": 2.252810001373291, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7056831121444702, + "num_tokens": 360069654.0, + "step": 14049 + }, + { + "epoch": 1.5429387217219417, + "grad_norm": 2.2615280151367188, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7074074745178223, + "num_tokens": 360096893.0, + "step": 14050 + }, + { + "epoch": 1.5430485394245552, + "grad_norm": 2.0301010608673096, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6845777034759521, + "num_tokens": 360130114.0, + "step": 14051 + }, + { + "epoch": 1.5431583571271688, + "grad_norm": 2.2799558639526367, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7024948000907898, + "num_tokens": 360156312.0, + "step": 14052 + }, + { + "epoch": 1.5432681748297825, + "grad_norm": 2.2515175342559814, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6932139992713928, + "num_tokens": 360184086.0, + "step": 14053 + }, + { + "epoch": 1.5433779925323963, + "grad_norm": 2.2265899181365967, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7053127288818359, + "num_tokens": 360209690.0, + "step": 14054 + }, + { + "epoch": 1.54348781023501, + "grad_norm": 2.420649528503418, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7419977784156799, + "num_tokens": 360230352.0, + "step": 14055 + }, + { + "epoch": 1.5435976279376236, + "grad_norm": 2.1748600006103516, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7039511799812317, + "num_tokens": 360257105.0, + "step": 14056 + }, + { + "epoch": 1.543707445640237, + "grad_norm": 2.2197186946868896, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7147611379623413, + "num_tokens": 360285737.0, + "step": 14057 + }, + { + "epoch": 1.5438172633428509, + "grad_norm": 1.8980708122253418, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7013537883758545, + "num_tokens": 360319651.0, + "step": 14058 + }, + { + "epoch": 1.5439270810454646, + "grad_norm": 2.4018499851226807, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6946886777877808, + "num_tokens": 360344444.0, + "step": 14059 + }, + { + "epoch": 1.5440368987480781, + "grad_norm": 2.2164976596832275, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7086283564567566, + "num_tokens": 360370428.0, + "step": 14060 + }, + { + "epoch": 1.544146716450692, + "grad_norm": 2.1585540771484375, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7077058553695679, + "num_tokens": 360397427.0, + "step": 14061 + }, + { + "epoch": 1.5442565341533054, + "grad_norm": 2.2981576919555664, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7720729112625122, + "num_tokens": 360419852.0, + "step": 14062 + }, + { + "epoch": 1.5443663518559192, + "grad_norm": 2.322556495666504, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7416073083877563, + "num_tokens": 360443535.0, + "step": 14063 + }, + { + "epoch": 1.544476169558533, + "grad_norm": 2.2253963947296143, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7184445261955261, + "num_tokens": 360470427.0, + "step": 14064 + }, + { + "epoch": 1.5445859872611465, + "grad_norm": 2.3135080337524414, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7417515516281128, + "num_tokens": 360493492.0, + "step": 14065 + }, + { + "epoch": 1.54469580496376, + "grad_norm": 2.49666428565979, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7249022722244263, + "num_tokens": 360513754.0, + "step": 14066 + }, + { + "epoch": 1.5448056226663738, + "grad_norm": 2.1569461822509766, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7512598633766174, + "num_tokens": 360538017.0, + "step": 14067 + }, + { + "epoch": 1.5449154403689875, + "grad_norm": 2.324950695037842, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7270052433013916, + "num_tokens": 360562406.0, + "step": 14068 + }, + { + "epoch": 1.5450252580716013, + "grad_norm": 2.3810060024261475, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7011007070541382, + "num_tokens": 360586023.0, + "step": 14069 + }, + { + "epoch": 1.5451350757742148, + "grad_norm": 2.002262592315674, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6933488845825195, + "num_tokens": 360616760.0, + "step": 14070 + }, + { + "epoch": 1.5452448934768284, + "grad_norm": 2.421435832977295, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7272353172302246, + "num_tokens": 360638538.0, + "step": 14071 + }, + { + "epoch": 1.5453547111794421, + "grad_norm": 1.9945604801177979, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7114968299865723, + "num_tokens": 360671731.0, + "step": 14072 + }, + { + "epoch": 1.5454645288820559, + "grad_norm": 2.218909502029419, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6997873187065125, + "num_tokens": 360700316.0, + "step": 14073 + }, + { + "epoch": 1.5455743465846694, + "grad_norm": 2.394308090209961, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7317695021629333, + "num_tokens": 360721827.0, + "step": 14074 + }, + { + "epoch": 1.545684164287283, + "grad_norm": 2.4603846073150635, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.733980655670166, + "num_tokens": 360742822.0, + "step": 14075 + }, + { + "epoch": 1.5457939819898967, + "grad_norm": 2.129800796508789, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7074723243713379, + "num_tokens": 360769240.0, + "step": 14076 + }, + { + "epoch": 1.5459037996925105, + "grad_norm": 2.4053571224212646, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7063167095184326, + "num_tokens": 360793452.0, + "step": 14077 + }, + { + "epoch": 1.5460136173951242, + "grad_norm": 2.3971121311187744, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7106374502182007, + "num_tokens": 360816477.0, + "step": 14078 + }, + { + "epoch": 1.5461234350977378, + "grad_norm": 2.170508861541748, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6952978372573853, + "num_tokens": 360847429.0, + "step": 14079 + }, + { + "epoch": 1.5462332528003513, + "grad_norm": 2.4354918003082275, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7339065670967102, + "num_tokens": 360869983.0, + "step": 14080 + }, + { + "epoch": 1.546343070502965, + "grad_norm": 2.1538450717926025, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7232247591018677, + "num_tokens": 360899101.0, + "step": 14081 + }, + { + "epoch": 1.5464528882055788, + "grad_norm": 2.447657823562622, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.714116632938385, + "num_tokens": 360921136.0, + "step": 14082 + }, + { + "epoch": 1.5465627059081926, + "grad_norm": 2.2152953147888184, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7050541639328003, + "num_tokens": 360949856.0, + "step": 14083 + }, + { + "epoch": 1.546672523610806, + "grad_norm": 2.5055789947509766, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7178192138671875, + "num_tokens": 360972615.0, + "step": 14084 + }, + { + "epoch": 1.5467823413134196, + "grad_norm": 2.0856711864471436, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.720384955406189, + "num_tokens": 360999909.0, + "step": 14085 + }, + { + "epoch": 1.5468921590160334, + "grad_norm": 2.2288639545440674, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7142065763473511, + "num_tokens": 361024492.0, + "step": 14086 + }, + { + "epoch": 1.5470019767186471, + "grad_norm": 2.2065441608428955, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.726172924041748, + "num_tokens": 361050746.0, + "step": 14087 + }, + { + "epoch": 1.5471117944212607, + "grad_norm": 2.413118362426758, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7165116667747498, + "num_tokens": 361075211.0, + "step": 14088 + }, + { + "epoch": 1.5472216121238742, + "grad_norm": 2.7458057403564453, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7301993370056152, + "num_tokens": 361094508.0, + "step": 14089 + }, + { + "epoch": 1.547331429826488, + "grad_norm": 2.472637414932251, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.733981192111969, + "num_tokens": 361115551.0, + "step": 14090 + }, + { + "epoch": 1.5474412475291017, + "grad_norm": 2.0931203365325928, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6985644102096558, + "num_tokens": 361145364.0, + "step": 14091 + }, + { + "epoch": 1.5475510652317155, + "grad_norm": 2.5140597820281982, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7282295227050781, + "num_tokens": 361166237.0, + "step": 14092 + }, + { + "epoch": 1.547660882934329, + "grad_norm": 2.4291348457336426, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7344928979873657, + "num_tokens": 361188553.0, + "step": 14093 + }, + { + "epoch": 1.5477707006369426, + "grad_norm": 2.3764195442199707, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.723432183265686, + "num_tokens": 361210876.0, + "step": 14094 + }, + { + "epoch": 1.5478805183395563, + "grad_norm": 2.291964054107666, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.724432110786438, + "num_tokens": 361234115.0, + "step": 14095 + }, + { + "epoch": 1.54799033604217, + "grad_norm": 2.326904296875, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7244352698326111, + "num_tokens": 361259909.0, + "step": 14096 + }, + { + "epoch": 1.5481001537447838, + "grad_norm": 2.0817372798919678, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7312653660774231, + "num_tokens": 361287646.0, + "step": 14097 + }, + { + "epoch": 1.5482099714473974, + "grad_norm": 2.4610016345977783, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7145551443099976, + "num_tokens": 361310106.0, + "step": 14098 + }, + { + "epoch": 1.548319789150011, + "grad_norm": 2.1008033752441406, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7247188091278076, + "num_tokens": 361338540.0, + "step": 14099 + }, + { + "epoch": 1.5484296068526247, + "grad_norm": 1.9768190383911133, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7050313353538513, + "num_tokens": 361370812.0, + "step": 14100 + }, + { + "epoch": 1.5485394245552384, + "grad_norm": 2.365183115005493, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7245354652404785, + "num_tokens": 361393603.0, + "step": 14101 + }, + { + "epoch": 1.548649242257852, + "grad_norm": 2.164581537246704, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6965423822402954, + "num_tokens": 361422292.0, + "step": 14102 + }, + { + "epoch": 1.5487590599604655, + "grad_norm": 2.1213624477386475, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7325928211212158, + "num_tokens": 361449118.0, + "step": 14103 + }, + { + "epoch": 1.5488688776630792, + "grad_norm": 2.145752429962158, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7177367210388184, + "num_tokens": 361476302.0, + "step": 14104 + }, + { + "epoch": 1.548978695365693, + "grad_norm": 2.0441877841949463, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6987354159355164, + "num_tokens": 361506787.0, + "step": 14105 + }, + { + "epoch": 1.5490885130683067, + "grad_norm": 2.175368070602417, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6971396207809448, + "num_tokens": 361534556.0, + "step": 14106 + }, + { + "epoch": 1.5491983307709203, + "grad_norm": 2.275733709335327, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7216185331344604, + "num_tokens": 361558659.0, + "step": 14107 + }, + { + "epoch": 1.5493081484735338, + "grad_norm": 2.437833786010742, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7210347652435303, + "num_tokens": 361581946.0, + "step": 14108 + }, + { + "epoch": 1.5494179661761476, + "grad_norm": 2.372124433517456, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7192897796630859, + "num_tokens": 361603056.0, + "step": 14109 + }, + { + "epoch": 1.5495277838787613, + "grad_norm": 2.0216825008392334, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7099525332450867, + "num_tokens": 361635431.0, + "step": 14110 + }, + { + "epoch": 1.5496376015813749, + "grad_norm": 2.6640372276306152, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7284359931945801, + "num_tokens": 361655859.0, + "step": 14111 + }, + { + "epoch": 1.5497474192839886, + "grad_norm": 2.20521879196167, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6944455504417419, + "num_tokens": 361684136.0, + "step": 14112 + }, + { + "epoch": 1.5498572369866022, + "grad_norm": 1.978603482246399, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7246754169464111, + "num_tokens": 361716876.0, + "step": 14113 + }, + { + "epoch": 1.549967054689216, + "grad_norm": 2.536665678024292, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.6995570659637451, + "num_tokens": 361739052.0, + "step": 14114 + }, + { + "epoch": 1.5500768723918297, + "grad_norm": 2.394531726837158, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7307665944099426, + "num_tokens": 361760777.0, + "step": 14115 + }, + { + "epoch": 1.5501866900944432, + "grad_norm": 2.2592618465423584, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7247772812843323, + "num_tokens": 361786636.0, + "step": 14116 + }, + { + "epoch": 1.5502965077970567, + "grad_norm": 2.677407741546631, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7222185730934143, + "num_tokens": 361807184.0, + "step": 14117 + }, + { + "epoch": 1.5504063254996705, + "grad_norm": 2.2256357669830322, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7443467378616333, + "num_tokens": 361830929.0, + "step": 14118 + }, + { + "epoch": 1.5505161432022843, + "grad_norm": 2.4732437133789062, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7351442575454712, + "num_tokens": 361850909.0, + "step": 14119 + }, + { + "epoch": 1.550625960904898, + "grad_norm": 2.2790169715881348, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7348824739456177, + "num_tokens": 361875006.0, + "step": 14120 + }, + { + "epoch": 1.5507357786075116, + "grad_norm": 2.4414010047912598, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7216352224349976, + "num_tokens": 361897745.0, + "step": 14121 + }, + { + "epoch": 1.550845596310125, + "grad_norm": 2.008976697921753, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7235936522483826, + "num_tokens": 361929016.0, + "step": 14122 + }, + { + "epoch": 1.5509554140127388, + "grad_norm": 2.1803197860717773, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7067267894744873, + "num_tokens": 361956413.0, + "step": 14123 + }, + { + "epoch": 1.5510652317153526, + "grad_norm": 2.3727223873138428, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7212284803390503, + "num_tokens": 361981100.0, + "step": 14124 + }, + { + "epoch": 1.5511750494179661, + "grad_norm": 2.098679542541504, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7252607941627502, + "num_tokens": 362009691.0, + "step": 14125 + }, + { + "epoch": 1.55128486712058, + "grad_norm": 2.204298257827759, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7071975469589233, + "num_tokens": 362036884.0, + "step": 14126 + }, + { + "epoch": 1.5513946848231934, + "grad_norm": 2.380099296569824, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7469468712806702, + "num_tokens": 362058074.0, + "step": 14127 + }, + { + "epoch": 1.5515045025258072, + "grad_norm": 2.2171790599823, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7380932569503784, + "num_tokens": 362082396.0, + "step": 14128 + }, + { + "epoch": 1.551614320228421, + "grad_norm": 2.2938315868377686, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7139611840248108, + "num_tokens": 362107300.0, + "step": 14129 + }, + { + "epoch": 1.5517241379310345, + "grad_norm": 2.4223015308380127, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7398091554641724, + "num_tokens": 362130342.0, + "step": 14130 + }, + { + "epoch": 1.551833955633648, + "grad_norm": 2.077798843383789, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7253191471099854, + "num_tokens": 362156524.0, + "step": 14131 + }, + { + "epoch": 1.5519437733362618, + "grad_norm": 2.305745840072632, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7223047018051147, + "num_tokens": 362183730.0, + "step": 14132 + }, + { + "epoch": 1.5520535910388755, + "grad_norm": 2.2957892417907715, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7288697957992554, + "num_tokens": 362208332.0, + "step": 14133 + }, + { + "epoch": 1.5521634087414893, + "grad_norm": 2.705379009246826, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.745330810546875, + "num_tokens": 362227674.0, + "step": 14134 + }, + { + "epoch": 1.5522732264441028, + "grad_norm": 2.014037847518921, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7297118306159973, + "num_tokens": 362258809.0, + "step": 14135 + }, + { + "epoch": 1.5523830441467164, + "grad_norm": 2.201143741607666, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7186384797096252, + "num_tokens": 362284938.0, + "step": 14136 + }, + { + "epoch": 1.55249286184933, + "grad_norm": 1.9709312915802002, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7376207113265991, + "num_tokens": 362314581.0, + "step": 14137 + }, + { + "epoch": 1.5526026795519439, + "grad_norm": 2.379455804824829, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6971018314361572, + "num_tokens": 362338732.0, + "step": 14138 + }, + { + "epoch": 1.5527124972545574, + "grad_norm": 2.329622983932495, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7106325626373291, + "num_tokens": 362363253.0, + "step": 14139 + }, + { + "epoch": 1.552822314957171, + "grad_norm": 2.4646801948547363, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7196073532104492, + "num_tokens": 362387027.0, + "step": 14140 + }, + { + "epoch": 1.5529321326597847, + "grad_norm": 2.3012049198150635, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.715978741645813, + "num_tokens": 362410307.0, + "step": 14141 + }, + { + "epoch": 1.5530419503623984, + "grad_norm": 1.9878170490264893, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7065226435661316, + "num_tokens": 362444370.0, + "step": 14142 + }, + { + "epoch": 1.5531517680650122, + "grad_norm": 2.3939895629882812, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.722290575504303, + "num_tokens": 362466750.0, + "step": 14143 + }, + { + "epoch": 1.5532615857676257, + "grad_norm": 2.5556068420410156, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7188728451728821, + "num_tokens": 362486720.0, + "step": 14144 + }, + { + "epoch": 1.5533714034702393, + "grad_norm": 2.10033917427063, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7115471363067627, + "num_tokens": 362513275.0, + "step": 14145 + }, + { + "epoch": 1.553481221172853, + "grad_norm": 2.152895212173462, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.731406569480896, + "num_tokens": 362539346.0, + "step": 14146 + }, + { + "epoch": 1.5535910388754668, + "grad_norm": 2.320801258087158, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7114222049713135, + "num_tokens": 362564442.0, + "step": 14147 + }, + { + "epoch": 1.5537008565780805, + "grad_norm": 2.42476749420166, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7369022369384766, + "num_tokens": 362587712.0, + "step": 14148 + }, + { + "epoch": 1.553810674280694, + "grad_norm": 2.3168845176696777, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7255532741546631, + "num_tokens": 362610684.0, + "step": 14149 + }, + { + "epoch": 1.5539204919833076, + "grad_norm": 2.2921903133392334, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7031357288360596, + "num_tokens": 362633807.0, + "step": 14150 + }, + { + "epoch": 1.5540303096859214, + "grad_norm": 2.4235808849334717, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7485288381576538, + "num_tokens": 362655208.0, + "step": 14151 + }, + { + "epoch": 1.5541401273885351, + "grad_norm": 2.2551047801971436, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7215933203697205, + "num_tokens": 362680867.0, + "step": 14152 + }, + { + "epoch": 1.5542499450911487, + "grad_norm": 2.07796311378479, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7196916937828064, + "num_tokens": 362709486.0, + "step": 14153 + }, + { + "epoch": 1.5543597627937622, + "grad_norm": 2.436004161834717, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.72139573097229, + "num_tokens": 362732356.0, + "step": 14154 + }, + { + "epoch": 1.554469580496376, + "grad_norm": 2.2383034229278564, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7323213815689087, + "num_tokens": 362759641.0, + "step": 14155 + }, + { + "epoch": 1.5545793981989897, + "grad_norm": 2.0599472522735596, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6933771371841431, + "num_tokens": 362791849.0, + "step": 14156 + }, + { + "epoch": 1.5546892159016035, + "grad_norm": 2.22521710395813, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7296702265739441, + "num_tokens": 362820163.0, + "step": 14157 + }, + { + "epoch": 1.554799033604217, + "grad_norm": 2.063058614730835, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7081363797187805, + "num_tokens": 362850796.0, + "step": 14158 + }, + { + "epoch": 1.5549088513068305, + "grad_norm": 2.0134339332580566, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7064812779426575, + "num_tokens": 362880698.0, + "step": 14159 + }, + { + "epoch": 1.5550186690094443, + "grad_norm": 2.1960010528564453, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7011468410491943, + "num_tokens": 362910730.0, + "step": 14160 + }, + { + "epoch": 1.555128486712058, + "grad_norm": 2.1744604110717773, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7189249992370605, + "num_tokens": 362939002.0, + "step": 14161 + }, + { + "epoch": 1.5552383044146716, + "grad_norm": 2.466724157333374, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.697790265083313, + "num_tokens": 362962548.0, + "step": 14162 + }, + { + "epoch": 1.5553481221172853, + "grad_norm": 2.27998948097229, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7005681395530701, + "num_tokens": 362985693.0, + "step": 14163 + }, + { + "epoch": 1.5554579398198989, + "grad_norm": 2.554020404815674, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7323049306869507, + "num_tokens": 363006058.0, + "step": 14164 + }, + { + "epoch": 1.5555677575225126, + "grad_norm": 2.266531229019165, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7177940011024475, + "num_tokens": 363031222.0, + "step": 14165 + }, + { + "epoch": 1.5556775752251264, + "grad_norm": 2.301663875579834, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7060670852661133, + "num_tokens": 363056559.0, + "step": 14166 + }, + { + "epoch": 1.55578739292774, + "grad_norm": 2.067861557006836, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7147948741912842, + "num_tokens": 363086176.0, + "step": 14167 + }, + { + "epoch": 1.5558972106303535, + "grad_norm": 2.205991268157959, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7062307596206665, + "num_tokens": 363112693.0, + "step": 14168 + }, + { + "epoch": 1.5560070283329672, + "grad_norm": 2.271312713623047, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7208431959152222, + "num_tokens": 363137190.0, + "step": 14169 + }, + { + "epoch": 1.556116846035581, + "grad_norm": 2.1713459491729736, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7188932299613953, + "num_tokens": 363165217.0, + "step": 14170 + }, + { + "epoch": 1.5562266637381947, + "grad_norm": 2.326572895050049, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7159677743911743, + "num_tokens": 363190706.0, + "step": 14171 + }, + { + "epoch": 1.5563364814408083, + "grad_norm": 2.4131059646606445, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7269817590713501, + "num_tokens": 363213457.0, + "step": 14172 + }, + { + "epoch": 1.5564462991434218, + "grad_norm": 2.369368553161621, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7118827104568481, + "num_tokens": 363237856.0, + "step": 14173 + }, + { + "epoch": 1.5565561168460356, + "grad_norm": 2.4858365058898926, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7258076071739197, + "num_tokens": 363260120.0, + "step": 14174 + }, + { + "epoch": 1.5566659345486493, + "grad_norm": 2.1675689220428467, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7057644128799438, + "num_tokens": 363288311.0, + "step": 14175 + }, + { + "epoch": 1.5567757522512629, + "grad_norm": 2.5270743370056152, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6985911726951599, + "num_tokens": 363309944.0, + "step": 14176 + }, + { + "epoch": 1.5568855699538766, + "grad_norm": 2.5423202514648438, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7353300452232361, + "num_tokens": 363331248.0, + "step": 14177 + }, + { + "epoch": 1.5569953876564901, + "grad_norm": 2.501781702041626, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.722569465637207, + "num_tokens": 363351155.0, + "step": 14178 + }, + { + "epoch": 1.557105205359104, + "grad_norm": 2.260300397872925, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7214251160621643, + "num_tokens": 363376282.0, + "step": 14179 + }, + { + "epoch": 1.5572150230617177, + "grad_norm": 2.111514091491699, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7068217992782593, + "num_tokens": 363402446.0, + "step": 14180 + }, + { + "epoch": 1.5573248407643312, + "grad_norm": 2.280156373977661, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7359153032302856, + "num_tokens": 363425836.0, + "step": 14181 + }, + { + "epoch": 1.5574346584669447, + "grad_norm": 2.199586868286133, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7245385646820068, + "num_tokens": 363451072.0, + "step": 14182 + }, + { + "epoch": 1.5575444761695585, + "grad_norm": 2.335347890853882, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.750829815864563, + "num_tokens": 363471607.0, + "step": 14183 + }, + { + "epoch": 1.5576542938721722, + "grad_norm": 2.4837210178375244, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7214982509613037, + "num_tokens": 363493861.0, + "step": 14184 + }, + { + "epoch": 1.557764111574786, + "grad_norm": 2.2151904106140137, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7256051301956177, + "num_tokens": 363521194.0, + "step": 14185 + }, + { + "epoch": 1.5578739292773995, + "grad_norm": 2.2982139587402344, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.722638726234436, + "num_tokens": 363545031.0, + "step": 14186 + }, + { + "epoch": 1.557983746980013, + "grad_norm": 2.372164011001587, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7056831121444702, + "num_tokens": 363569648.0, + "step": 14187 + }, + { + "epoch": 1.5580935646826268, + "grad_norm": 2.1062798500061035, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.727306604385376, + "num_tokens": 363601973.0, + "step": 14188 + }, + { + "epoch": 1.5582033823852406, + "grad_norm": 1.931525707244873, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7047659158706665, + "num_tokens": 363634069.0, + "step": 14189 + }, + { + "epoch": 1.5583132000878541, + "grad_norm": 2.3228249549865723, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6983350515365601, + "num_tokens": 363660460.0, + "step": 14190 + }, + { + "epoch": 1.5584230177904677, + "grad_norm": 2.2982707023620605, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7202464938163757, + "num_tokens": 363687015.0, + "step": 14191 + }, + { + "epoch": 1.5585328354930814, + "grad_norm": 2.158975124359131, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7142577767372131, + "num_tokens": 363716130.0, + "step": 14192 + }, + { + "epoch": 1.5586426531956952, + "grad_norm": 2.332519054412842, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7367134094238281, + "num_tokens": 363738478.0, + "step": 14193 + }, + { + "epoch": 1.558752470898309, + "grad_norm": 2.163677453994751, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7335915565490723, + "num_tokens": 363763371.0, + "step": 14194 + }, + { + "epoch": 1.5588622886009225, + "grad_norm": 2.400717258453369, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.752307653427124, + "num_tokens": 363786658.0, + "step": 14195 + }, + { + "epoch": 1.558972106303536, + "grad_norm": 2.7920429706573486, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.731482982635498, + "num_tokens": 363805936.0, + "step": 14196 + }, + { + "epoch": 1.5590819240061498, + "grad_norm": 2.3719749450683594, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7114166021347046, + "num_tokens": 363831469.0, + "step": 14197 + }, + { + "epoch": 1.5591917417087635, + "grad_norm": 2.2074973583221436, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7367216348648071, + "num_tokens": 363858236.0, + "step": 14198 + }, + { + "epoch": 1.5593015594113773, + "grad_norm": 2.18108868598938, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7321818470954895, + "num_tokens": 363883346.0, + "step": 14199 + }, + { + "epoch": 1.5594113771139908, + "grad_norm": 2.039686441421509, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7191351652145386, + "num_tokens": 363912153.0, + "step": 14200 + }, + { + "epoch": 1.5595211948166043, + "grad_norm": 2.208785057067871, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6893264055252075, + "num_tokens": 363940636.0, + "step": 14201 + }, + { + "epoch": 1.559631012519218, + "grad_norm": 2.099247455596924, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7228630781173706, + "num_tokens": 363967521.0, + "step": 14202 + }, + { + "epoch": 1.5597408302218319, + "grad_norm": 2.5203537940979004, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7491359710693359, + "num_tokens": 363989027.0, + "step": 14203 + }, + { + "epoch": 1.5598506479244454, + "grad_norm": 2.355835199356079, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7109782695770264, + "num_tokens": 364013991.0, + "step": 14204 + }, + { + "epoch": 1.559960465627059, + "grad_norm": 2.134239912033081, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7256060838699341, + "num_tokens": 364042053.0, + "step": 14205 + }, + { + "epoch": 1.5600702833296727, + "grad_norm": 2.329732656478882, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6995911002159119, + "num_tokens": 364067123.0, + "step": 14206 + }, + { + "epoch": 1.5601801010322864, + "grad_norm": 2.3310353755950928, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7100867033004761, + "num_tokens": 364090908.0, + "step": 14207 + }, + { + "epoch": 1.5602899187349002, + "grad_norm": 2.2848174571990967, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7361829280853271, + "num_tokens": 364115350.0, + "step": 14208 + }, + { + "epoch": 1.5603997364375137, + "grad_norm": 2.0200304985046387, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7052567005157471, + "num_tokens": 364146561.0, + "step": 14209 + }, + { + "epoch": 1.5605095541401273, + "grad_norm": 2.2281417846679688, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7202740907669067, + "num_tokens": 364170680.0, + "step": 14210 + }, + { + "epoch": 1.560619371842741, + "grad_norm": 2.4937989711761475, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7300015091896057, + "num_tokens": 364190240.0, + "step": 14211 + }, + { + "epoch": 1.5607291895453548, + "grad_norm": 2.1239664554595947, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.725845992565155, + "num_tokens": 364218102.0, + "step": 14212 + }, + { + "epoch": 1.5608390072479685, + "grad_norm": 2.199333429336548, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7264159321784973, + "num_tokens": 364244961.0, + "step": 14213 + }, + { + "epoch": 1.560948824950582, + "grad_norm": 2.139737129211426, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7194182276725769, + "num_tokens": 364271638.0, + "step": 14214 + }, + { + "epoch": 1.5610586426531956, + "grad_norm": 2.065155267715454, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.715471088886261, + "num_tokens": 364302027.0, + "step": 14215 + }, + { + "epoch": 1.5611684603558094, + "grad_norm": 2.119166135787964, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7044222354888916, + "num_tokens": 364331995.0, + "step": 14216 + }, + { + "epoch": 1.5612782780584231, + "grad_norm": 2.390894889831543, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7195583581924438, + "num_tokens": 364354524.0, + "step": 14217 + }, + { + "epoch": 1.5613880957610367, + "grad_norm": 1.961533784866333, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7269084453582764, + "num_tokens": 364386490.0, + "step": 14218 + }, + { + "epoch": 1.5614979134636502, + "grad_norm": 2.4981114864349365, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7355309724807739, + "num_tokens": 364408392.0, + "step": 14219 + }, + { + "epoch": 1.561607731166264, + "grad_norm": 2.6538076400756836, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7251290082931519, + "num_tokens": 364428159.0, + "step": 14220 + }, + { + "epoch": 1.5617175488688777, + "grad_norm": 2.5902388095855713, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7310754060745239, + "num_tokens": 364449793.0, + "step": 14221 + }, + { + "epoch": 1.5618273665714915, + "grad_norm": 2.385187864303589, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7219219207763672, + "num_tokens": 364475003.0, + "step": 14222 + }, + { + "epoch": 1.561937184274105, + "grad_norm": 2.382081985473633, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7104883193969727, + "num_tokens": 364499787.0, + "step": 14223 + }, + { + "epoch": 1.5620470019767185, + "grad_norm": 2.160177230834961, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7309731245040894, + "num_tokens": 364527464.0, + "step": 14224 + }, + { + "epoch": 1.5621568196793323, + "grad_norm": 2.5099456310272217, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7113432288169861, + "num_tokens": 364547663.0, + "step": 14225 + }, + { + "epoch": 1.562266637381946, + "grad_norm": 2.2052409648895264, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6931100487709045, + "num_tokens": 364574887.0, + "step": 14226 + }, + { + "epoch": 1.5623764550845596, + "grad_norm": 2.216999053955078, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7245633602142334, + "num_tokens": 364602888.0, + "step": 14227 + }, + { + "epoch": 1.5624862727871733, + "grad_norm": 2.5415945053100586, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7232095003128052, + "num_tokens": 364622711.0, + "step": 14228 + }, + { + "epoch": 1.5625960904897869, + "grad_norm": 2.307223320007324, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7198360562324524, + "num_tokens": 364648522.0, + "step": 14229 + }, + { + "epoch": 1.5627059081924006, + "grad_norm": 2.2224819660186768, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6955589652061462, + "num_tokens": 364676212.0, + "step": 14230 + }, + { + "epoch": 1.5628157258950144, + "grad_norm": 2.329712390899658, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.720820426940918, + "num_tokens": 364701635.0, + "step": 14231 + }, + { + "epoch": 1.562925543597628, + "grad_norm": 2.342623233795166, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7288458347320557, + "num_tokens": 364724667.0, + "step": 14232 + }, + { + "epoch": 1.5630353613002415, + "grad_norm": 2.260347366333008, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7223556041717529, + "num_tokens": 364748738.0, + "step": 14233 + }, + { + "epoch": 1.5631451790028552, + "grad_norm": 2.205972671508789, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7030238509178162, + "num_tokens": 364774814.0, + "step": 14234 + }, + { + "epoch": 1.563254996705469, + "grad_norm": 2.656435966491699, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7444152235984802, + "num_tokens": 364793169.0, + "step": 14235 + }, + { + "epoch": 1.5633648144080827, + "grad_norm": 2.0743114948272705, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7121319770812988, + "num_tokens": 364822672.0, + "step": 14236 + }, + { + "epoch": 1.5634746321106963, + "grad_norm": 2.069575548171997, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7124679684638977, + "num_tokens": 364850317.0, + "step": 14237 + }, + { + "epoch": 1.5635844498133098, + "grad_norm": 2.191911220550537, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7247811555862427, + "num_tokens": 364878764.0, + "step": 14238 + }, + { + "epoch": 1.5636942675159236, + "grad_norm": 2.35459041595459, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7015296816825867, + "num_tokens": 364905854.0, + "step": 14239 + }, + { + "epoch": 1.5638040852185373, + "grad_norm": 2.448810338973999, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7353895902633667, + "num_tokens": 364926584.0, + "step": 14240 + }, + { + "epoch": 1.5639139029211508, + "grad_norm": 2.1312649250030518, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7244359254837036, + "num_tokens": 364954317.0, + "step": 14241 + }, + { + "epoch": 1.5640237206237646, + "grad_norm": 2.068375587463379, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7014191150665283, + "num_tokens": 364985289.0, + "step": 14242 + }, + { + "epoch": 1.5641335383263781, + "grad_norm": 2.4788453578948975, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.722254753112793, + "num_tokens": 365008321.0, + "step": 14243 + }, + { + "epoch": 1.564243356028992, + "grad_norm": 2.0383243560791016, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7011091709136963, + "num_tokens": 365038635.0, + "step": 14244 + }, + { + "epoch": 1.5643531737316057, + "grad_norm": 2.047292470932007, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7105417251586914, + "num_tokens": 365068993.0, + "step": 14245 + }, + { + "epoch": 1.5644629914342192, + "grad_norm": 2.05249285697937, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7179402709007263, + "num_tokens": 365097896.0, + "step": 14246 + }, + { + "epoch": 1.5645728091368327, + "grad_norm": 2.2227041721343994, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7433755397796631, + "num_tokens": 365123063.0, + "step": 14247 + }, + { + "epoch": 1.5646826268394465, + "grad_norm": 2.378843069076538, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7120640277862549, + "num_tokens": 365148227.0, + "step": 14248 + }, + { + "epoch": 1.5647924445420602, + "grad_norm": 2.248129367828369, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.728574275970459, + "num_tokens": 365173730.0, + "step": 14249 + }, + { + "epoch": 1.564902262244674, + "grad_norm": 2.11503005027771, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7179856896400452, + "num_tokens": 365199981.0, + "step": 14250 + }, + { + "epoch": 1.5650120799472875, + "grad_norm": 2.5141305923461914, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7198082804679871, + "num_tokens": 365220434.0, + "step": 14251 + }, + { + "epoch": 1.565121897649901, + "grad_norm": 2.3806560039520264, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7082677483558655, + "num_tokens": 365243756.0, + "step": 14252 + }, + { + "epoch": 1.5652317153525148, + "grad_norm": 2.1376373767852783, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7208021283149719, + "num_tokens": 365271571.0, + "step": 14253 + }, + { + "epoch": 1.5653415330551286, + "grad_norm": 2.1525800228118896, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.71177077293396, + "num_tokens": 365299415.0, + "step": 14254 + }, + { + "epoch": 1.565451350757742, + "grad_norm": 2.8429572582244873, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.745254397392273, + "num_tokens": 365316927.0, + "step": 14255 + }, + { + "epoch": 1.5655611684603556, + "grad_norm": 2.5966248512268066, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7010326385498047, + "num_tokens": 365338516.0, + "step": 14256 + }, + { + "epoch": 1.5656709861629694, + "grad_norm": 2.002951145172119, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7177214026451111, + "num_tokens": 365372905.0, + "step": 14257 + }, + { + "epoch": 1.5657808038655832, + "grad_norm": 2.4941372871398926, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7439693808555603, + "num_tokens": 365395150.0, + "step": 14258 + }, + { + "epoch": 1.565890621568197, + "grad_norm": 2.0931332111358643, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.713807225227356, + "num_tokens": 365426707.0, + "step": 14259 + }, + { + "epoch": 1.5660004392708105, + "grad_norm": 2.2302684783935547, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7358570098876953, + "num_tokens": 365449024.0, + "step": 14260 + }, + { + "epoch": 1.566110256973424, + "grad_norm": 2.109419345855713, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7103103399276733, + "num_tokens": 365478757.0, + "step": 14261 + }, + { + "epoch": 1.5662200746760377, + "grad_norm": 2.0837807655334473, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7036075592041016, + "num_tokens": 365508247.0, + "step": 14262 + }, + { + "epoch": 1.5663298923786515, + "grad_norm": 2.3138136863708496, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7102383971214294, + "num_tokens": 365532521.0, + "step": 14263 + }, + { + "epoch": 1.5664397100812653, + "grad_norm": 2.2278330326080322, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7105659246444702, + "num_tokens": 365557935.0, + "step": 14264 + }, + { + "epoch": 1.5665495277838788, + "grad_norm": 2.1644654273986816, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7133399248123169, + "num_tokens": 365586116.0, + "step": 14265 + }, + { + "epoch": 1.5666593454864923, + "grad_norm": 2.1556389331817627, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7095704078674316, + "num_tokens": 365614319.0, + "step": 14266 + }, + { + "epoch": 1.566769163189106, + "grad_norm": 1.9287279844284058, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6991133689880371, + "num_tokens": 365648518.0, + "step": 14267 + }, + { + "epoch": 1.5668789808917198, + "grad_norm": 2.199331760406494, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7215849757194519, + "num_tokens": 365674588.0, + "step": 14268 + }, + { + "epoch": 1.5669887985943334, + "grad_norm": 2.2480356693267822, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7085191011428833, + "num_tokens": 365699322.0, + "step": 14269 + }, + { + "epoch": 1.567098616296947, + "grad_norm": 2.268193244934082, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7294586896896362, + "num_tokens": 365724433.0, + "step": 14270 + }, + { + "epoch": 1.5672084339995607, + "grad_norm": 2.0308053493499756, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7247147560119629, + "num_tokens": 365755888.0, + "step": 14271 + }, + { + "epoch": 1.5673182517021744, + "grad_norm": 2.1959800720214844, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7151484489440918, + "num_tokens": 365780426.0, + "step": 14272 + }, + { + "epoch": 1.5674280694047882, + "grad_norm": 2.281466245651245, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.727783739566803, + "num_tokens": 365805724.0, + "step": 14273 + }, + { + "epoch": 1.5675378871074017, + "grad_norm": 2.2849502563476562, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7162325978279114, + "num_tokens": 365830592.0, + "step": 14274 + }, + { + "epoch": 1.5676477048100153, + "grad_norm": 2.3745689392089844, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7149666547775269, + "num_tokens": 365855765.0, + "step": 14275 + }, + { + "epoch": 1.567757522512629, + "grad_norm": 2.401484966278076, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.737067461013794, + "num_tokens": 365877743.0, + "step": 14276 + }, + { + "epoch": 1.5678673402152428, + "grad_norm": 2.2444088459014893, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.733977198600769, + "num_tokens": 365903016.0, + "step": 14277 + }, + { + "epoch": 1.5679771579178565, + "grad_norm": 2.039533853530884, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7125952243804932, + "num_tokens": 365933546.0, + "step": 14278 + }, + { + "epoch": 1.56808697562047, + "grad_norm": 2.353858232498169, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7217531204223633, + "num_tokens": 365957822.0, + "step": 14279 + }, + { + "epoch": 1.5681967933230836, + "grad_norm": 2.1654675006866455, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6838804483413696, + "num_tokens": 365988142.0, + "step": 14280 + }, + { + "epoch": 1.5683066110256974, + "grad_norm": 2.1348252296447754, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7244886159896851, + "num_tokens": 366015560.0, + "step": 14281 + }, + { + "epoch": 1.568416428728311, + "grad_norm": 2.3311331272125244, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7102221250534058, + "num_tokens": 366040980.0, + "step": 14282 + }, + { + "epoch": 1.5685262464309246, + "grad_norm": 1.9850748777389526, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7184849977493286, + "num_tokens": 366073073.0, + "step": 14283 + }, + { + "epoch": 1.5686360641335382, + "grad_norm": 2.0493667125701904, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.710107684135437, + "num_tokens": 366101398.0, + "step": 14284 + }, + { + "epoch": 1.568745881836152, + "grad_norm": 2.354825258255005, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7322136759757996, + "num_tokens": 366124316.0, + "step": 14285 + }, + { + "epoch": 1.5688556995387657, + "grad_norm": 2.2546839714050293, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7250815629959106, + "num_tokens": 366147207.0, + "step": 14286 + }, + { + "epoch": 1.5689655172413794, + "grad_norm": 2.2663707733154297, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7532641291618347, + "num_tokens": 366169838.0, + "step": 14287 + }, + { + "epoch": 1.569075334943993, + "grad_norm": 2.210930585861206, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7281545400619507, + "num_tokens": 366194235.0, + "step": 14288 + }, + { + "epoch": 1.5691851526466065, + "grad_norm": 2.103740930557251, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7111810445785522, + "num_tokens": 366222049.0, + "step": 14289 + }, + { + "epoch": 1.5692949703492203, + "grad_norm": 2.1969547271728516, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7325353622436523, + "num_tokens": 366246600.0, + "step": 14290 + }, + { + "epoch": 1.569404788051834, + "grad_norm": 1.8778539896011353, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7139865159988403, + "num_tokens": 366281184.0, + "step": 14291 + }, + { + "epoch": 1.5695146057544476, + "grad_norm": 2.141902208328247, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7311450242996216, + "num_tokens": 366307921.0, + "step": 14292 + }, + { + "epoch": 1.5696244234570613, + "grad_norm": 2.403984546661377, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7363952994346619, + "num_tokens": 366330969.0, + "step": 14293 + }, + { + "epoch": 1.5697342411596749, + "grad_norm": 2.0444860458374023, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7183154821395874, + "num_tokens": 366360724.0, + "step": 14294 + }, + { + "epoch": 1.5698440588622886, + "grad_norm": 2.284510850906372, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7242594957351685, + "num_tokens": 366385576.0, + "step": 14295 + }, + { + "epoch": 1.5699538765649024, + "grad_norm": 2.3339810371398926, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7438384294509888, + "num_tokens": 366408398.0, + "step": 14296 + }, + { + "epoch": 1.570063694267516, + "grad_norm": 2.061131238937378, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.6994255185127258, + "num_tokens": 366439578.0, + "step": 14297 + }, + { + "epoch": 1.5701735119701294, + "grad_norm": 2.4314424991607666, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.732245922088623, + "num_tokens": 366462512.0, + "step": 14298 + }, + { + "epoch": 1.5702833296727432, + "grad_norm": 2.1318323612213135, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.707841157913208, + "num_tokens": 366491464.0, + "step": 14299 + }, + { + "epoch": 1.570393147375357, + "grad_norm": 2.221029281616211, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7051499485969543, + "num_tokens": 366520121.0, + "step": 14300 + }, + { + "epoch": 1.5705029650779707, + "grad_norm": 2.179401159286499, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6964708566665649, + "num_tokens": 366547828.0, + "step": 14301 + }, + { + "epoch": 1.5706127827805842, + "grad_norm": 2.4350087642669678, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7104647755622864, + "num_tokens": 366571139.0, + "step": 14302 + }, + { + "epoch": 1.5707226004831978, + "grad_norm": 2.3722736835479736, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7425792217254639, + "num_tokens": 366593288.0, + "step": 14303 + }, + { + "epoch": 1.5708324181858115, + "grad_norm": 2.0171902179718018, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7107318639755249, + "num_tokens": 366623312.0, + "step": 14304 + }, + { + "epoch": 1.5709422358884253, + "grad_norm": 2.2564077377319336, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7416198253631592, + "num_tokens": 366646215.0, + "step": 14305 + }, + { + "epoch": 1.5710520535910388, + "grad_norm": 2.1152920722961426, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7067270874977112, + "num_tokens": 366675837.0, + "step": 14306 + }, + { + "epoch": 1.5711618712936526, + "grad_norm": 2.216414213180542, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7009033560752869, + "num_tokens": 366701687.0, + "step": 14307 + }, + { + "epoch": 1.5712716889962661, + "grad_norm": 2.140756845474243, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7437535524368286, + "num_tokens": 366726129.0, + "step": 14308 + }, + { + "epoch": 1.5713815066988799, + "grad_norm": 2.2946462631225586, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7414759397506714, + "num_tokens": 366748938.0, + "step": 14309 + }, + { + "epoch": 1.5714913244014936, + "grad_norm": 2.1649279594421387, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.728190541267395, + "num_tokens": 366777066.0, + "step": 14310 + }, + { + "epoch": 1.5716011421041072, + "grad_norm": 1.9874780178070068, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.709293007850647, + "num_tokens": 366808119.0, + "step": 14311 + }, + { + "epoch": 1.5717109598067207, + "grad_norm": 2.1438746452331543, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7160780429840088, + "num_tokens": 366835034.0, + "step": 14312 + }, + { + "epoch": 1.5718207775093345, + "grad_norm": 2.36379337310791, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7198023200035095, + "num_tokens": 366857448.0, + "step": 14313 + }, + { + "epoch": 1.5719305952119482, + "grad_norm": 2.4738593101501465, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7439447641372681, + "num_tokens": 366878911.0, + "step": 14314 + }, + { + "epoch": 1.572040412914562, + "grad_norm": 2.304969310760498, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7447999715805054, + "num_tokens": 366902056.0, + "step": 14315 + }, + { + "epoch": 1.5721502306171755, + "grad_norm": 2.0839264392852783, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7176626920700073, + "num_tokens": 366932758.0, + "step": 14316 + }, + { + "epoch": 1.572260048319789, + "grad_norm": 2.300091505050659, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6943868398666382, + "num_tokens": 366958619.0, + "step": 14317 + }, + { + "epoch": 1.5723698660224028, + "grad_norm": 2.204620599746704, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7025152444839478, + "num_tokens": 366985003.0, + "step": 14318 + }, + { + "epoch": 1.5724796837250166, + "grad_norm": 2.1775920391082764, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7066409587860107, + "num_tokens": 367013340.0, + "step": 14319 + }, + { + "epoch": 1.57258950142763, + "grad_norm": 2.2457683086395264, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7206791639328003, + "num_tokens": 367039455.0, + "step": 14320 + }, + { + "epoch": 1.5726993191302436, + "grad_norm": 2.059455633163452, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7013958692550659, + "num_tokens": 367070459.0, + "step": 14321 + }, + { + "epoch": 1.5728091368328574, + "grad_norm": 2.4880902767181396, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7222400903701782, + "num_tokens": 367092484.0, + "step": 14322 + }, + { + "epoch": 1.5729189545354711, + "grad_norm": 2.0806849002838135, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7149694561958313, + "num_tokens": 367121226.0, + "step": 14323 + }, + { + "epoch": 1.573028772238085, + "grad_norm": 2.1987099647521973, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7152602672576904, + "num_tokens": 367147961.0, + "step": 14324 + }, + { + "epoch": 1.5731385899406984, + "grad_norm": 2.4985251426696777, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7171923518180847, + "num_tokens": 367169773.0, + "step": 14325 + }, + { + "epoch": 1.573248407643312, + "grad_norm": 2.1686644554138184, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7216179370880127, + "num_tokens": 367195357.0, + "step": 14326 + }, + { + "epoch": 1.5733582253459257, + "grad_norm": 1.947522521018982, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7027151584625244, + "num_tokens": 367227704.0, + "step": 14327 + }, + { + "epoch": 1.5734680430485395, + "grad_norm": 2.3989553451538086, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7387700080871582, + "num_tokens": 367249979.0, + "step": 14328 + }, + { + "epoch": 1.5735778607511532, + "grad_norm": 2.2975125312805176, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6975343227386475, + "num_tokens": 367274387.0, + "step": 14329 + }, + { + "epoch": 1.5736876784537668, + "grad_norm": 2.1108100414276123, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6988022923469543, + "num_tokens": 367304211.0, + "step": 14330 + }, + { + "epoch": 1.5737974961563803, + "grad_norm": 2.4950203895568848, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7422917485237122, + "num_tokens": 367323595.0, + "step": 14331 + }, + { + "epoch": 1.573907313858994, + "grad_norm": 2.1776700019836426, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7194759249687195, + "num_tokens": 367348623.0, + "step": 14332 + }, + { + "epoch": 1.5740171315616078, + "grad_norm": 2.2947731018066406, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7092715501785278, + "num_tokens": 367373954.0, + "step": 14333 + }, + { + "epoch": 1.5741269492642214, + "grad_norm": 2.246669292449951, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6865257024765015, + "num_tokens": 367401386.0, + "step": 14334 + }, + { + "epoch": 1.574236766966835, + "grad_norm": 2.1601145267486572, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7004120349884033, + "num_tokens": 367427585.0, + "step": 14335 + }, + { + "epoch": 1.5743465846694487, + "grad_norm": 2.2159860134124756, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7200808525085449, + "num_tokens": 367453266.0, + "step": 14336 + }, + { + "epoch": 1.5744564023720624, + "grad_norm": 2.2411258220672607, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6961084008216858, + "num_tokens": 367479459.0, + "step": 14337 + }, + { + "epoch": 1.5745662200746762, + "grad_norm": 2.5134477615356445, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7261676788330078, + "num_tokens": 367500316.0, + "step": 14338 + }, + { + "epoch": 1.5746760377772897, + "grad_norm": 2.233212471008301, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.722785234451294, + "num_tokens": 367524900.0, + "step": 14339 + }, + { + "epoch": 1.5747858554799032, + "grad_norm": 2.1335983276367188, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7060604095458984, + "num_tokens": 367552193.0, + "step": 14340 + }, + { + "epoch": 1.574895673182517, + "grad_norm": 2.1248862743377686, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7325450778007507, + "num_tokens": 367579540.0, + "step": 14341 + }, + { + "epoch": 1.5750054908851308, + "grad_norm": 2.478330373764038, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7373389005661011, + "num_tokens": 367602349.0, + "step": 14342 + }, + { + "epoch": 1.5751153085877443, + "grad_norm": 2.5289125442504883, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.720477283000946, + "num_tokens": 367623207.0, + "step": 14343 + }, + { + "epoch": 1.575225126290358, + "grad_norm": 1.9967832565307617, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6944615840911865, + "num_tokens": 367657327.0, + "step": 14344 + }, + { + "epoch": 1.5753349439929716, + "grad_norm": 2.1297607421875, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7316053509712219, + "num_tokens": 367684215.0, + "step": 14345 + }, + { + "epoch": 1.5754447616955853, + "grad_norm": 2.2670583724975586, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.713519275188446, + "num_tokens": 367708748.0, + "step": 14346 + }, + { + "epoch": 1.575554579398199, + "grad_norm": 2.162736415863037, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.699211061000824, + "num_tokens": 367736554.0, + "step": 14347 + }, + { + "epoch": 1.5756643971008126, + "grad_norm": 2.4174394607543945, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7294264435768127, + "num_tokens": 367758735.0, + "step": 14348 + }, + { + "epoch": 1.5757742148034262, + "grad_norm": 2.0823216438293457, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6999443769454956, + "num_tokens": 367793246.0, + "step": 14349 + }, + { + "epoch": 1.57588403250604, + "grad_norm": 2.427340030670166, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7269468903541565, + "num_tokens": 367815309.0, + "step": 14350 + }, + { + "epoch": 1.5759938502086537, + "grad_norm": 2.2181777954101562, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.734548807144165, + "num_tokens": 367839735.0, + "step": 14351 + }, + { + "epoch": 1.5761036679112674, + "grad_norm": 2.2317748069763184, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7022841572761536, + "num_tokens": 367867259.0, + "step": 14352 + }, + { + "epoch": 1.576213485613881, + "grad_norm": 2.4278757572174072, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7218475341796875, + "num_tokens": 367889792.0, + "step": 14353 + }, + { + "epoch": 1.5763233033164945, + "grad_norm": 2.2744393348693848, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7175634503364563, + "num_tokens": 367914257.0, + "step": 14354 + }, + { + "epoch": 1.5764331210191083, + "grad_norm": 2.317993640899658, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7286927103996277, + "num_tokens": 367938747.0, + "step": 14355 + }, + { + "epoch": 1.576542938721722, + "grad_norm": 2.1929869651794434, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6912814378738403, + "num_tokens": 367968393.0, + "step": 14356 + }, + { + "epoch": 1.5766527564243356, + "grad_norm": 2.291048288345337, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7446966171264648, + "num_tokens": 367991279.0, + "step": 14357 + }, + { + "epoch": 1.5767625741269493, + "grad_norm": 2.074631452560425, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7478302717208862, + "num_tokens": 368020097.0, + "step": 14358 + }, + { + "epoch": 1.5768723918295628, + "grad_norm": 1.9726964235305786, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7247710227966309, + "num_tokens": 368051189.0, + "step": 14359 + }, + { + "epoch": 1.5769822095321766, + "grad_norm": 2.277219533920288, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7019888162612915, + "num_tokens": 368076705.0, + "step": 14360 + }, + { + "epoch": 1.5770920272347904, + "grad_norm": 2.213491916656494, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7108274698257446, + "num_tokens": 368103592.0, + "step": 14361 + }, + { + "epoch": 1.577201844937404, + "grad_norm": 2.298607349395752, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7206140756607056, + "num_tokens": 368128930.0, + "step": 14362 + }, + { + "epoch": 1.5773116626400174, + "grad_norm": 2.0103068351745605, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7018455862998962, + "num_tokens": 368160832.0, + "step": 14363 + }, + { + "epoch": 1.5774214803426312, + "grad_norm": 2.324582815170288, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7373768091201782, + "num_tokens": 368186379.0, + "step": 14364 + }, + { + "epoch": 1.577531298045245, + "grad_norm": 2.08931303024292, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6994490027427673, + "num_tokens": 368215017.0, + "step": 14365 + }, + { + "epoch": 1.5776411157478587, + "grad_norm": 1.9402285814285278, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7008047103881836, + "num_tokens": 368248016.0, + "step": 14366 + }, + { + "epoch": 1.5777509334504722, + "grad_norm": 2.1986472606658936, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7116155028343201, + "num_tokens": 368275045.0, + "step": 14367 + }, + { + "epoch": 1.5778607511530858, + "grad_norm": 2.4100091457366943, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7319484949111938, + "num_tokens": 368296047.0, + "step": 14368 + }, + { + "epoch": 1.5779705688556995, + "grad_norm": 1.993789553642273, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7155803442001343, + "num_tokens": 368327143.0, + "step": 14369 + }, + { + "epoch": 1.5780803865583133, + "grad_norm": 2.301109790802002, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7084079384803772, + "num_tokens": 368351687.0, + "step": 14370 + }, + { + "epoch": 1.5781902042609268, + "grad_norm": 2.183002233505249, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.710347056388855, + "num_tokens": 368378896.0, + "step": 14371 + }, + { + "epoch": 1.5783000219635406, + "grad_norm": 2.2511303424835205, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7346951961517334, + "num_tokens": 368403134.0, + "step": 14372 + }, + { + "epoch": 1.5784098396661541, + "grad_norm": 2.3428568840026855, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7186319828033447, + "num_tokens": 368426006.0, + "step": 14373 + }, + { + "epoch": 1.5785196573687679, + "grad_norm": 2.1968183517456055, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7079341411590576, + "num_tokens": 368452274.0, + "step": 14374 + }, + { + "epoch": 1.5786294750713816, + "grad_norm": 2.5680150985717773, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7469288110733032, + "num_tokens": 368471811.0, + "step": 14375 + }, + { + "epoch": 1.5787392927739952, + "grad_norm": 2.2087182998657227, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7447104454040527, + "num_tokens": 368498026.0, + "step": 14376 + }, + { + "epoch": 1.5788491104766087, + "grad_norm": 2.3405306339263916, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7056292295455933, + "num_tokens": 368520644.0, + "step": 14377 + }, + { + "epoch": 1.5789589281792225, + "grad_norm": 2.1834452152252197, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7036784887313843, + "num_tokens": 368546893.0, + "step": 14378 + }, + { + "epoch": 1.5790687458818362, + "grad_norm": 2.129255533218384, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7120015025138855, + "num_tokens": 368575043.0, + "step": 14379 + }, + { + "epoch": 1.57917856358445, + "grad_norm": 2.274259090423584, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7490991950035095, + "num_tokens": 368597246.0, + "step": 14380 + }, + { + "epoch": 1.5792883812870635, + "grad_norm": 2.144681215286255, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7143553495407104, + "num_tokens": 368624434.0, + "step": 14381 + }, + { + "epoch": 1.579398198989677, + "grad_norm": 2.3705594539642334, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7257539629936218, + "num_tokens": 368646102.0, + "step": 14382 + }, + { + "epoch": 1.5795080166922908, + "grad_norm": 2.282576560974121, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7115373611450195, + "num_tokens": 368671161.0, + "step": 14383 + }, + { + "epoch": 1.5796178343949046, + "grad_norm": 2.0749106407165527, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7244315147399902, + "num_tokens": 368698284.0, + "step": 14384 + }, + { + "epoch": 1.579727652097518, + "grad_norm": 2.216179609298706, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7196649312973022, + "num_tokens": 368723963.0, + "step": 14385 + }, + { + "epoch": 1.5798374698001316, + "grad_norm": 2.4272642135620117, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6785271763801575, + "num_tokens": 368749440.0, + "step": 14386 + }, + { + "epoch": 1.5799472875027454, + "grad_norm": 2.2621347904205322, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7129658460617065, + "num_tokens": 368774540.0, + "step": 14387 + }, + { + "epoch": 1.5800571052053591, + "grad_norm": 2.522709369659424, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7304004430770874, + "num_tokens": 368794975.0, + "step": 14388 + }, + { + "epoch": 1.580166922907973, + "grad_norm": 2.143239974975586, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7290037274360657, + "num_tokens": 368819557.0, + "step": 14389 + }, + { + "epoch": 1.5802767406105864, + "grad_norm": 2.1588025093078613, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7008152604103088, + "num_tokens": 368847359.0, + "step": 14390 + }, + { + "epoch": 1.5803865583132, + "grad_norm": 2.087493419647217, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7066274285316467, + "num_tokens": 368876928.0, + "step": 14391 + }, + { + "epoch": 1.5804963760158137, + "grad_norm": 2.05350923538208, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6811157464981079, + "num_tokens": 368909086.0, + "step": 14392 + }, + { + "epoch": 1.5806061937184275, + "grad_norm": 2.1310999393463135, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7398893237113953, + "num_tokens": 368935750.0, + "step": 14393 + }, + { + "epoch": 1.5807160114210412, + "grad_norm": 2.6757843494415283, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7212828397750854, + "num_tokens": 368956195.0, + "step": 14394 + }, + { + "epoch": 1.5808258291236548, + "grad_norm": 2.786844253540039, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7386029362678528, + "num_tokens": 368974852.0, + "step": 14395 + }, + { + "epoch": 1.5809356468262683, + "grad_norm": 2.32326602935791, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.712657630443573, + "num_tokens": 368998870.0, + "step": 14396 + }, + { + "epoch": 1.581045464528882, + "grad_norm": 2.3814375400543213, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7137123346328735, + "num_tokens": 369023862.0, + "step": 14397 + }, + { + "epoch": 1.5811552822314958, + "grad_norm": 2.2972447872161865, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7216216325759888, + "num_tokens": 369049343.0, + "step": 14398 + }, + { + "epoch": 1.5812650999341094, + "grad_norm": 2.1649727821350098, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7121241092681885, + "num_tokens": 369074458.0, + "step": 14399 + }, + { + "epoch": 1.5813749176367229, + "grad_norm": 2.406193256378174, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7251695990562439, + "num_tokens": 369095270.0, + "step": 14400 + }, + { + "epoch": 1.5814847353393366, + "grad_norm": 2.2010560035705566, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7331013679504395, + "num_tokens": 369119887.0, + "step": 14401 + }, + { + "epoch": 1.5815945530419504, + "grad_norm": 2.3203654289245605, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7253315448760986, + "num_tokens": 369145281.0, + "step": 14402 + }, + { + "epoch": 1.5817043707445642, + "grad_norm": 2.5781044960021973, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7203221321105957, + "num_tokens": 369166443.0, + "step": 14403 + }, + { + "epoch": 1.5818141884471777, + "grad_norm": 2.062061071395874, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7238624095916748, + "num_tokens": 369195365.0, + "step": 14404 + }, + { + "epoch": 1.5819240061497912, + "grad_norm": 2.084594964981079, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.697791576385498, + "num_tokens": 369224800.0, + "step": 14405 + }, + { + "epoch": 1.582033823852405, + "grad_norm": 2.194398880004883, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7264871597290039, + "num_tokens": 369249786.0, + "step": 14406 + }, + { + "epoch": 1.5821436415550187, + "grad_norm": 2.0486221313476562, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7239155769348145, + "num_tokens": 369278559.0, + "step": 14407 + }, + { + "epoch": 1.5822534592576323, + "grad_norm": 2.091430902481079, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7486909031867981, + "num_tokens": 369306274.0, + "step": 14408 + }, + { + "epoch": 1.582363276960246, + "grad_norm": 2.116264581680298, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7391132712364197, + "num_tokens": 369335406.0, + "step": 14409 + }, + { + "epoch": 1.5824730946628596, + "grad_norm": 2.0795652866363525, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7026962041854858, + "num_tokens": 369366802.0, + "step": 14410 + }, + { + "epoch": 1.5825829123654733, + "grad_norm": 2.2651214599609375, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7177258729934692, + "num_tokens": 369391865.0, + "step": 14411 + }, + { + "epoch": 1.582692730068087, + "grad_norm": 2.1649363040924072, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7518840432167053, + "num_tokens": 369417957.0, + "step": 14412 + }, + { + "epoch": 1.5828025477707006, + "grad_norm": 2.3203773498535156, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7223875522613525, + "num_tokens": 369441119.0, + "step": 14413 + }, + { + "epoch": 1.5829123654733142, + "grad_norm": 2.0830202102661133, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7207530736923218, + "num_tokens": 369469698.0, + "step": 14414 + }, + { + "epoch": 1.583022183175928, + "grad_norm": 2.173004388809204, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7116873264312744, + "num_tokens": 369497642.0, + "step": 14415 + }, + { + "epoch": 1.5831320008785417, + "grad_norm": 2.5252058506011963, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7215702533721924, + "num_tokens": 369519327.0, + "step": 14416 + }, + { + "epoch": 1.5832418185811554, + "grad_norm": 2.3104300498962402, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7182290554046631, + "num_tokens": 369544087.0, + "step": 14417 + }, + { + "epoch": 1.583351636283769, + "grad_norm": 2.144819736480713, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7133198380470276, + "num_tokens": 369571169.0, + "step": 14418 + }, + { + "epoch": 1.5834614539863825, + "grad_norm": 2.5822160243988037, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7124742269515991, + "num_tokens": 369592309.0, + "step": 14419 + }, + { + "epoch": 1.5835712716889963, + "grad_norm": 2.2008304595947266, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7026961445808411, + "num_tokens": 369621679.0, + "step": 14420 + }, + { + "epoch": 1.58368108939161, + "grad_norm": 2.4721434116363525, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7383424639701843, + "num_tokens": 369642830.0, + "step": 14421 + }, + { + "epoch": 1.5837909070942235, + "grad_norm": 2.3492584228515625, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.738834023475647, + "num_tokens": 369665267.0, + "step": 14422 + }, + { + "epoch": 1.5839007247968373, + "grad_norm": 2.5751028060913086, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7194863557815552, + "num_tokens": 369686741.0, + "step": 14423 + }, + { + "epoch": 1.5840105424994508, + "grad_norm": 2.1968061923980713, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7213811874389648, + "num_tokens": 369711151.0, + "step": 14424 + }, + { + "epoch": 1.5841203602020646, + "grad_norm": 2.3462557792663574, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7207397818565369, + "num_tokens": 369733311.0, + "step": 14425 + }, + { + "epoch": 1.5842301779046783, + "grad_norm": 2.2292113304138184, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7200160026550293, + "num_tokens": 369758633.0, + "step": 14426 + }, + { + "epoch": 1.5843399956072919, + "grad_norm": 2.023036479949951, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7117952108383179, + "num_tokens": 369790248.0, + "step": 14427 + }, + { + "epoch": 1.5844498133099054, + "grad_norm": 2.5365049839019775, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7183823585510254, + "num_tokens": 369810847.0, + "step": 14428 + }, + { + "epoch": 1.5845596310125192, + "grad_norm": 1.874568223953247, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7044666409492493, + "num_tokens": 369844602.0, + "step": 14429 + }, + { + "epoch": 1.584669448715133, + "grad_norm": 2.2869033813476562, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7242863774299622, + "num_tokens": 369868151.0, + "step": 14430 + }, + { + "epoch": 1.5847792664177467, + "grad_norm": 2.4901278018951416, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6919741034507751, + "num_tokens": 369891211.0, + "step": 14431 + }, + { + "epoch": 1.5848890841203602, + "grad_norm": 2.2376668453216553, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6953351497650146, + "num_tokens": 369920554.0, + "step": 14432 + }, + { + "epoch": 1.5849989018229738, + "grad_norm": 2.1314735412597656, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7108539342880249, + "num_tokens": 369948385.0, + "step": 14433 + }, + { + "epoch": 1.5851087195255875, + "grad_norm": 2.6332688331604004, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7496553659439087, + "num_tokens": 369967737.0, + "step": 14434 + }, + { + "epoch": 1.5852185372282013, + "grad_norm": 2.0696682929992676, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7281211614608765, + "num_tokens": 369994530.0, + "step": 14435 + }, + { + "epoch": 1.5853283549308148, + "grad_norm": 2.367932081222534, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.723433256149292, + "num_tokens": 370017246.0, + "step": 14436 + }, + { + "epoch": 1.5854381726334283, + "grad_norm": 2.5683987140655518, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7329837679862976, + "num_tokens": 370040283.0, + "step": 14437 + }, + { + "epoch": 1.585547990336042, + "grad_norm": 2.2021381855010986, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.725698709487915, + "num_tokens": 370068493.0, + "step": 14438 + }, + { + "epoch": 1.5856578080386559, + "grad_norm": 2.1187350749969482, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7383862733840942, + "num_tokens": 370095323.0, + "step": 14439 + }, + { + "epoch": 1.5857676257412696, + "grad_norm": 2.2844836711883545, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7120347023010254, + "num_tokens": 370122262.0, + "step": 14440 + }, + { + "epoch": 1.5858774434438832, + "grad_norm": 2.2642822265625, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7209705114364624, + "num_tokens": 370148379.0, + "step": 14441 + }, + { + "epoch": 1.5859872611464967, + "grad_norm": 2.138303756713867, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.715624213218689, + "num_tokens": 370175877.0, + "step": 14442 + }, + { + "epoch": 1.5860970788491104, + "grad_norm": 2.1951091289520264, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7189016342163086, + "num_tokens": 370202044.0, + "step": 14443 + }, + { + "epoch": 1.5862068965517242, + "grad_norm": 2.3581957817077637, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7095308303833008, + "num_tokens": 370224667.0, + "step": 14444 + }, + { + "epoch": 1.586316714254338, + "grad_norm": 2.1119236946105957, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.723486065864563, + "num_tokens": 370251916.0, + "step": 14445 + }, + { + "epoch": 1.5864265319569515, + "grad_norm": 2.046955108642578, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7224850654602051, + "num_tokens": 370281829.0, + "step": 14446 + }, + { + "epoch": 1.586536349659565, + "grad_norm": 2.363863468170166, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7089850902557373, + "num_tokens": 370305997.0, + "step": 14447 + }, + { + "epoch": 1.5866461673621788, + "grad_norm": 2.5066325664520264, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7154540419578552, + "num_tokens": 370327645.0, + "step": 14448 + }, + { + "epoch": 1.5867559850647925, + "grad_norm": 2.2715578079223633, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7230521440505981, + "num_tokens": 370351695.0, + "step": 14449 + }, + { + "epoch": 1.586865802767406, + "grad_norm": 2.1180036067962646, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7232130765914917, + "num_tokens": 370378784.0, + "step": 14450 + }, + { + "epoch": 1.5869756204700196, + "grad_norm": 2.3584530353546143, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7418335676193237, + "num_tokens": 370401619.0, + "step": 14451 + }, + { + "epoch": 1.5870854381726334, + "grad_norm": 2.505211591720581, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7276405096054077, + "num_tokens": 370422117.0, + "step": 14452 + }, + { + "epoch": 1.5871952558752471, + "grad_norm": 2.342094659805298, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7144294381141663, + "num_tokens": 370446515.0, + "step": 14453 + }, + { + "epoch": 1.5873050735778609, + "grad_norm": 2.28324031829834, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7141187787055969, + "num_tokens": 370474101.0, + "step": 14454 + }, + { + "epoch": 1.5874148912804744, + "grad_norm": 2.2689316272735596, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.73503577709198, + "num_tokens": 370497468.0, + "step": 14455 + }, + { + "epoch": 1.587524708983088, + "grad_norm": 2.224904775619507, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7012442946434021, + "num_tokens": 370523117.0, + "step": 14456 + }, + { + "epoch": 1.5876345266857017, + "grad_norm": 1.989586591720581, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7268809080123901, + "num_tokens": 370553566.0, + "step": 14457 + }, + { + "epoch": 1.5877443443883155, + "grad_norm": 2.2456586360931396, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7022967338562012, + "num_tokens": 370579595.0, + "step": 14458 + }, + { + "epoch": 1.5878541620909292, + "grad_norm": 2.11966872215271, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7014368772506714, + "num_tokens": 370608495.0, + "step": 14459 + }, + { + "epoch": 1.5879639797935428, + "grad_norm": 2.2237982749938965, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7138627767562866, + "num_tokens": 370635567.0, + "step": 14460 + }, + { + "epoch": 1.5880737974961563, + "grad_norm": 2.171248197555542, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7101455926895142, + "num_tokens": 370661573.0, + "step": 14461 + }, + { + "epoch": 1.58818361519877, + "grad_norm": 2.2343363761901855, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.694213330745697, + "num_tokens": 370688328.0, + "step": 14462 + }, + { + "epoch": 1.5882934329013838, + "grad_norm": 2.1493024826049805, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7201639413833618, + "num_tokens": 370715996.0, + "step": 14463 + }, + { + "epoch": 1.5884032506039973, + "grad_norm": 2.603050947189331, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7356222867965698, + "num_tokens": 370736427.0, + "step": 14464 + }, + { + "epoch": 1.5885130683066109, + "grad_norm": 2.2254526615142822, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7500042915344238, + "num_tokens": 370761215.0, + "step": 14465 + }, + { + "epoch": 1.5886228860092246, + "grad_norm": 2.232367515563965, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7105392813682556, + "num_tokens": 370787435.0, + "step": 14466 + }, + { + "epoch": 1.5887327037118384, + "grad_norm": 2.1080849170684814, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7126071453094482, + "num_tokens": 370815194.0, + "step": 14467 + }, + { + "epoch": 1.5888425214144521, + "grad_norm": 2.199726104736328, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7352698445320129, + "num_tokens": 370840228.0, + "step": 14468 + }, + { + "epoch": 1.5889523391170657, + "grad_norm": 2.3002493381500244, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7204797267913818, + "num_tokens": 370862762.0, + "step": 14469 + }, + { + "epoch": 1.5890621568196792, + "grad_norm": 2.1545093059539795, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.724160373210907, + "num_tokens": 370889930.0, + "step": 14470 + }, + { + "epoch": 1.589171974522293, + "grad_norm": 1.973419427871704, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7175847291946411, + "num_tokens": 370922242.0, + "step": 14471 + }, + { + "epoch": 1.5892817922249067, + "grad_norm": 2.510350227355957, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7349117994308472, + "num_tokens": 370945489.0, + "step": 14472 + }, + { + "epoch": 1.5893916099275203, + "grad_norm": 2.508279800415039, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7304896116256714, + "num_tokens": 370964323.0, + "step": 14473 + }, + { + "epoch": 1.589501427630134, + "grad_norm": 1.9986382722854614, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7157487273216248, + "num_tokens": 370994171.0, + "step": 14474 + }, + { + "epoch": 1.5896112453327476, + "grad_norm": 2.185044765472412, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7223540544509888, + "num_tokens": 371019411.0, + "step": 14475 + }, + { + "epoch": 1.5897210630353613, + "grad_norm": 2.191819190979004, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7104675769805908, + "num_tokens": 371047157.0, + "step": 14476 + }, + { + "epoch": 1.589830880737975, + "grad_norm": 1.9383864402770996, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7253474593162537, + "num_tokens": 371081297.0, + "step": 14477 + }, + { + "epoch": 1.5899406984405886, + "grad_norm": 2.048471450805664, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7220926284790039, + "num_tokens": 371111984.0, + "step": 14478 + }, + { + "epoch": 1.5900505161432021, + "grad_norm": 2.0455410480499268, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7251739501953125, + "num_tokens": 371139534.0, + "step": 14479 + }, + { + "epoch": 1.590160333845816, + "grad_norm": 2.357825994491577, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7065542340278625, + "num_tokens": 371165026.0, + "step": 14480 + }, + { + "epoch": 1.5902701515484297, + "grad_norm": 2.292419910430908, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7214602828025818, + "num_tokens": 371189397.0, + "step": 14481 + }, + { + "epoch": 1.5903799692510434, + "grad_norm": 2.3841991424560547, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7543736696243286, + "num_tokens": 371211137.0, + "step": 14482 + }, + { + "epoch": 1.590489786953657, + "grad_norm": 2.331157922744751, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.743630588054657, + "num_tokens": 371233013.0, + "step": 14483 + }, + { + "epoch": 1.5905996046562705, + "grad_norm": 2.075958251953125, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7383760809898376, + "num_tokens": 371261887.0, + "step": 14484 + }, + { + "epoch": 1.5907094223588842, + "grad_norm": 2.32344126701355, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7011004686355591, + "num_tokens": 371287722.0, + "step": 14485 + }, + { + "epoch": 1.590819240061498, + "grad_norm": 2.094038963317871, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7205899953842163, + "num_tokens": 371315921.0, + "step": 14486 + }, + { + "epoch": 1.5909290577641115, + "grad_norm": 2.337568759918213, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7337247729301453, + "num_tokens": 371339893.0, + "step": 14487 + }, + { + "epoch": 1.5910388754667253, + "grad_norm": 2.193970203399658, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7208176851272583, + "num_tokens": 371365490.0, + "step": 14488 + }, + { + "epoch": 1.5911486931693388, + "grad_norm": 2.530522108078003, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7114531993865967, + "num_tokens": 371387345.0, + "step": 14489 + }, + { + "epoch": 1.5912585108719526, + "grad_norm": 1.999851107597351, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7323645353317261, + "num_tokens": 371416406.0, + "step": 14490 + }, + { + "epoch": 1.5913683285745663, + "grad_norm": 1.9742389917373657, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.685764729976654, + "num_tokens": 371450704.0, + "step": 14491 + }, + { + "epoch": 1.5914781462771799, + "grad_norm": 1.9456816911697388, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.6996736526489258, + "num_tokens": 371483923.0, + "step": 14492 + }, + { + "epoch": 1.5915879639797934, + "grad_norm": 2.0164403915405273, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7030106782913208, + "num_tokens": 371515735.0, + "step": 14493 + }, + { + "epoch": 1.5916977816824072, + "grad_norm": 2.5023417472839355, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7192879915237427, + "num_tokens": 371536828.0, + "step": 14494 + }, + { + "epoch": 1.591807599385021, + "grad_norm": 2.3151705265045166, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7233295440673828, + "num_tokens": 371561807.0, + "step": 14495 + }, + { + "epoch": 1.5919174170876347, + "grad_norm": 2.1160614490509033, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7518444061279297, + "num_tokens": 371589963.0, + "step": 14496 + }, + { + "epoch": 1.5920272347902482, + "grad_norm": 1.8323041200637817, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7402347326278687, + "num_tokens": 371622448.0, + "step": 14497 + }, + { + "epoch": 1.5921370524928617, + "grad_norm": 1.8923594951629639, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6973949670791626, + "num_tokens": 371658014.0, + "step": 14498 + }, + { + "epoch": 1.5922468701954755, + "grad_norm": 2.0755879878997803, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.705592155456543, + "num_tokens": 371689178.0, + "step": 14499 + }, + { + "epoch": 1.5923566878980893, + "grad_norm": 2.2643721103668213, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7433615922927856, + "num_tokens": 371714426.0, + "step": 14500 + }, + { + "epoch": 1.5924665056007028, + "grad_norm": 2.737112045288086, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7560065984725952, + "num_tokens": 371732027.0, + "step": 14501 + }, + { + "epoch": 1.5925763233033163, + "grad_norm": 2.0712108612060547, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7186946868896484, + "num_tokens": 371761898.0, + "step": 14502 + }, + { + "epoch": 1.59268614100593, + "grad_norm": 2.3831725120544434, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.706288754940033, + "num_tokens": 371788371.0, + "step": 14503 + }, + { + "epoch": 1.5927959587085438, + "grad_norm": 2.4167673587799072, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6947387456893921, + "num_tokens": 371812135.0, + "step": 14504 + }, + { + "epoch": 1.5929057764111576, + "grad_norm": 2.5266101360321045, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7179946899414062, + "num_tokens": 371832016.0, + "step": 14505 + }, + { + "epoch": 1.5930155941137711, + "grad_norm": 2.3872947692871094, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7218770384788513, + "num_tokens": 371855819.0, + "step": 14506 + }, + { + "epoch": 1.5931254118163847, + "grad_norm": 2.1241796016693115, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7254626750946045, + "num_tokens": 371883523.0, + "step": 14507 + }, + { + "epoch": 1.5932352295189984, + "grad_norm": 2.028111457824707, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7141690254211426, + "num_tokens": 371912850.0, + "step": 14508 + }, + { + "epoch": 1.5933450472216122, + "grad_norm": 2.4563686847686768, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7211465835571289, + "num_tokens": 371933776.0, + "step": 14509 + }, + { + "epoch": 1.593454864924226, + "grad_norm": 2.2584824562072754, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7170855402946472, + "num_tokens": 371958414.0, + "step": 14510 + }, + { + "epoch": 1.5935646826268395, + "grad_norm": 2.1664059162139893, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7115014791488647, + "num_tokens": 371984899.0, + "step": 14511 + }, + { + "epoch": 1.593674500329453, + "grad_norm": 2.2145895957946777, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7267193794250488, + "num_tokens": 372010943.0, + "step": 14512 + }, + { + "epoch": 1.5937843180320668, + "grad_norm": 2.4108283519744873, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7095421552658081, + "num_tokens": 372033357.0, + "step": 14513 + }, + { + "epoch": 1.5938941357346805, + "grad_norm": 2.34089994430542, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7067884206771851, + "num_tokens": 372058543.0, + "step": 14514 + }, + { + "epoch": 1.594003953437294, + "grad_norm": 2.1953048706054688, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7131317257881165, + "num_tokens": 372084565.0, + "step": 14515 + }, + { + "epoch": 1.5941137711399076, + "grad_norm": 2.5909721851348877, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7262452244758606, + "num_tokens": 372103950.0, + "step": 14516 + }, + { + "epoch": 1.5942235888425214, + "grad_norm": 2.334444999694824, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.710662841796875, + "num_tokens": 372128120.0, + "step": 14517 + }, + { + "epoch": 1.5943334065451351, + "grad_norm": 2.083824396133423, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6952257752418518, + "num_tokens": 372159348.0, + "step": 14518 + }, + { + "epoch": 1.5944432242477489, + "grad_norm": 2.2449886798858643, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7375592589378357, + "num_tokens": 372183734.0, + "step": 14519 + }, + { + "epoch": 1.5945530419503624, + "grad_norm": 2.295990228652954, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7218048572540283, + "num_tokens": 372207044.0, + "step": 14520 + }, + { + "epoch": 1.594662859652976, + "grad_norm": 2.29282283782959, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7247682809829712, + "num_tokens": 372232679.0, + "step": 14521 + }, + { + "epoch": 1.5947726773555897, + "grad_norm": 2.352982521057129, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7114905118942261, + "num_tokens": 372256338.0, + "step": 14522 + }, + { + "epoch": 1.5948824950582035, + "grad_norm": 2.114027976989746, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6890355348587036, + "num_tokens": 372285856.0, + "step": 14523 + }, + { + "epoch": 1.5949923127608172, + "grad_norm": 2.0952115058898926, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7335335612297058, + "num_tokens": 372315292.0, + "step": 14524 + }, + { + "epoch": 1.5951021304634307, + "grad_norm": 2.314878225326538, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.711937427520752, + "num_tokens": 372339241.0, + "step": 14525 + }, + { + "epoch": 1.5952119481660443, + "grad_norm": 2.2630512714385986, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7016241550445557, + "num_tokens": 372364262.0, + "step": 14526 + }, + { + "epoch": 1.595321765868658, + "grad_norm": 2.1233036518096924, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7085719704627991, + "num_tokens": 372392001.0, + "step": 14527 + }, + { + "epoch": 1.5954315835712718, + "grad_norm": 2.224517822265625, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7252169847488403, + "num_tokens": 372418882.0, + "step": 14528 + }, + { + "epoch": 1.5955414012738853, + "grad_norm": 2.6350364685058594, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7205952405929565, + "num_tokens": 372440492.0, + "step": 14529 + }, + { + "epoch": 1.5956512189764989, + "grad_norm": 2.100921392440796, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7213424444198608, + "num_tokens": 372469168.0, + "step": 14530 + }, + { + "epoch": 1.5957610366791126, + "grad_norm": 2.4561593532562256, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7270451188087463, + "num_tokens": 372492122.0, + "step": 14531 + }, + { + "epoch": 1.5958708543817264, + "grad_norm": 2.3936803340911865, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7612758874893188, + "num_tokens": 372512907.0, + "step": 14532 + }, + { + "epoch": 1.5959806720843401, + "grad_norm": 2.0623083114624023, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7065736651420593, + "num_tokens": 372543926.0, + "step": 14533 + }, + { + "epoch": 1.5960904897869537, + "grad_norm": 2.3787968158721924, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7329159379005432, + "num_tokens": 372566024.0, + "step": 14534 + }, + { + "epoch": 1.5962003074895672, + "grad_norm": 2.4035961627960205, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7010444402694702, + "num_tokens": 372590130.0, + "step": 14535 + }, + { + "epoch": 1.596310125192181, + "grad_norm": 2.2444472312927246, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7172582149505615, + "num_tokens": 372614839.0, + "step": 14536 + }, + { + "epoch": 1.5964199428947947, + "grad_norm": 2.283276319503784, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7010034322738647, + "num_tokens": 372641449.0, + "step": 14537 + }, + { + "epoch": 1.5965297605974083, + "grad_norm": 2.2136733531951904, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7381580471992493, + "num_tokens": 372667187.0, + "step": 14538 + }, + { + "epoch": 1.596639578300022, + "grad_norm": 2.134972095489502, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7245844602584839, + "num_tokens": 372694410.0, + "step": 14539 + }, + { + "epoch": 1.5967493960026355, + "grad_norm": 2.1853902339935303, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7131551504135132, + "num_tokens": 372721881.0, + "step": 14540 + }, + { + "epoch": 1.5968592137052493, + "grad_norm": 2.1499693393707275, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7110906839370728, + "num_tokens": 372748927.0, + "step": 14541 + }, + { + "epoch": 1.596969031407863, + "grad_norm": 2.528965473175049, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7234046459197998, + "num_tokens": 372769678.0, + "step": 14542 + }, + { + "epoch": 1.5970788491104766, + "grad_norm": 2.4193506240844727, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7276531457901001, + "num_tokens": 372790521.0, + "step": 14543 + }, + { + "epoch": 1.5971886668130901, + "grad_norm": 2.341311454772949, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7267489433288574, + "num_tokens": 372812775.0, + "step": 14544 + }, + { + "epoch": 1.5972984845157039, + "grad_norm": 2.510697841644287, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7232934236526489, + "num_tokens": 372833367.0, + "step": 14545 + }, + { + "epoch": 1.5974083022183176, + "grad_norm": 2.1628341674804688, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7016202211380005, + "num_tokens": 372861059.0, + "step": 14546 + }, + { + "epoch": 1.5975181199209314, + "grad_norm": 2.3064002990722656, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7139459848403931, + "num_tokens": 372885170.0, + "step": 14547 + }, + { + "epoch": 1.597627937623545, + "grad_norm": 2.0901689529418945, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7393792867660522, + "num_tokens": 372913648.0, + "step": 14548 + }, + { + "epoch": 1.5977377553261585, + "grad_norm": 2.1953835487365723, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7333788275718689, + "num_tokens": 372939540.0, + "step": 14549 + }, + { + "epoch": 1.5978475730287722, + "grad_norm": 2.0757930278778076, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7000979781150818, + "num_tokens": 372969053.0, + "step": 14550 + }, + { + "epoch": 1.597957390731386, + "grad_norm": 2.3265268802642822, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7394726276397705, + "num_tokens": 372991584.0, + "step": 14551 + }, + { + "epoch": 1.5980672084339995, + "grad_norm": 2.2832446098327637, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7195457220077515, + "num_tokens": 373016528.0, + "step": 14552 + }, + { + "epoch": 1.5981770261366133, + "grad_norm": 2.2516016960144043, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7191612720489502, + "num_tokens": 373041360.0, + "step": 14553 + }, + { + "epoch": 1.5982868438392268, + "grad_norm": 2.213859796524048, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7306392192840576, + "num_tokens": 373068075.0, + "step": 14554 + }, + { + "epoch": 1.5983966615418406, + "grad_norm": 2.793041229248047, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7459838390350342, + "num_tokens": 373086274.0, + "step": 14555 + }, + { + "epoch": 1.5985064792444543, + "grad_norm": 2.687467336654663, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7192949056625366, + "num_tokens": 373105973.0, + "step": 14556 + }, + { + "epoch": 1.5986162969470679, + "grad_norm": 2.1936697959899902, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7113121747970581, + "num_tokens": 373135613.0, + "step": 14557 + }, + { + "epoch": 1.5987261146496814, + "grad_norm": 2.7296578884124756, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.722201943397522, + "num_tokens": 373152901.0, + "step": 14558 + }, + { + "epoch": 1.5988359323522952, + "grad_norm": 2.213123083114624, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7258238196372986, + "num_tokens": 373179190.0, + "step": 14559 + }, + { + "epoch": 1.598945750054909, + "grad_norm": 2.278651475906372, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7252461910247803, + "num_tokens": 373203937.0, + "step": 14560 + }, + { + "epoch": 1.5990555677575227, + "grad_norm": 2.2326011657714844, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7167587280273438, + "num_tokens": 373230307.0, + "step": 14561 + }, + { + "epoch": 1.5991653854601362, + "grad_norm": 2.351166009902954, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7167131900787354, + "num_tokens": 373256093.0, + "step": 14562 + }, + { + "epoch": 1.5992752031627497, + "grad_norm": 2.48710298538208, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7243891954421997, + "num_tokens": 373277704.0, + "step": 14563 + }, + { + "epoch": 1.5993850208653635, + "grad_norm": 2.1969923973083496, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7124261856079102, + "num_tokens": 373305210.0, + "step": 14564 + }, + { + "epoch": 1.5994948385679773, + "grad_norm": 2.275343656539917, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7218408584594727, + "num_tokens": 373331894.0, + "step": 14565 + }, + { + "epoch": 1.5996046562705908, + "grad_norm": 2.150078058242798, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7303599119186401, + "num_tokens": 373357036.0, + "step": 14566 + }, + { + "epoch": 1.5997144739732043, + "grad_norm": 2.223240613937378, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7111229300498962, + "num_tokens": 373384047.0, + "step": 14567 + }, + { + "epoch": 1.599824291675818, + "grad_norm": 2.216704845428467, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7074616551399231, + "num_tokens": 373410349.0, + "step": 14568 + }, + { + "epoch": 1.5999341093784318, + "grad_norm": 2.1609084606170654, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7330239415168762, + "num_tokens": 373435936.0, + "step": 14569 + }, + { + "epoch": 1.6000439270810456, + "grad_norm": 2.2279889583587646, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7110950946807861, + "num_tokens": 373462199.0, + "step": 14570 + }, + { + "epoch": 1.6001537447836591, + "grad_norm": 2.2211735248565674, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7193207740783691, + "num_tokens": 373487178.0, + "step": 14571 + }, + { + "epoch": 1.6002635624862727, + "grad_norm": 1.995916724205017, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7232487201690674, + "num_tokens": 373517279.0, + "step": 14572 + }, + { + "epoch": 1.6003733801888864, + "grad_norm": 2.3694586753845215, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7006583213806152, + "num_tokens": 373542616.0, + "step": 14573 + }, + { + "epoch": 1.6004831978915002, + "grad_norm": 2.2009479999542236, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7394799590110779, + "num_tokens": 373569471.0, + "step": 14574 + }, + { + "epoch": 1.600593015594114, + "grad_norm": 2.5975890159606934, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7301452159881592, + "num_tokens": 373589818.0, + "step": 14575 + }, + { + "epoch": 1.6007028332967275, + "grad_norm": 2.1633129119873047, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7080125212669373, + "num_tokens": 373616171.0, + "step": 14576 + }, + { + "epoch": 1.600812650999341, + "grad_norm": 2.3364181518554688, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7040103077888489, + "num_tokens": 373639415.0, + "step": 14577 + }, + { + "epoch": 1.6009224687019548, + "grad_norm": 2.7120912075042725, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7482911348342896, + "num_tokens": 373656069.0, + "step": 14578 + }, + { + "epoch": 1.6010322864045685, + "grad_norm": 2.636589288711548, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7357949614524841, + "num_tokens": 373676151.0, + "step": 14579 + }, + { + "epoch": 1.601142104107182, + "grad_norm": 2.075110912322998, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7225363254547119, + "num_tokens": 373705615.0, + "step": 14580 + }, + { + "epoch": 1.6012519218097956, + "grad_norm": 1.868652105331421, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7332497835159302, + "num_tokens": 373740844.0, + "step": 14581 + }, + { + "epoch": 1.6013617395124093, + "grad_norm": 2.217716932296753, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7105937600135803, + "num_tokens": 373769657.0, + "step": 14582 + }, + { + "epoch": 1.601471557215023, + "grad_norm": 2.4943997859954834, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7209922671318054, + "num_tokens": 373791540.0, + "step": 14583 + }, + { + "epoch": 1.6015813749176369, + "grad_norm": 2.176523208618164, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6918479800224304, + "num_tokens": 373821468.0, + "step": 14584 + }, + { + "epoch": 1.6016911926202504, + "grad_norm": 2.27620530128479, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7297427654266357, + "num_tokens": 373845244.0, + "step": 14585 + }, + { + "epoch": 1.601801010322864, + "grad_norm": 2.3092637062072754, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7152324914932251, + "num_tokens": 373871305.0, + "step": 14586 + }, + { + "epoch": 1.6019108280254777, + "grad_norm": 2.240827798843384, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6957176923751831, + "num_tokens": 373897347.0, + "step": 14587 + }, + { + "epoch": 1.6020206457280914, + "grad_norm": 2.470369815826416, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7138245105743408, + "num_tokens": 373919654.0, + "step": 14588 + }, + { + "epoch": 1.602130463430705, + "grad_norm": 2.2318952083587646, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6883544325828552, + "num_tokens": 373946668.0, + "step": 14589 + }, + { + "epoch": 1.6022402811333187, + "grad_norm": 2.0212020874023438, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7161494493484497, + "num_tokens": 373975855.0, + "step": 14590 + }, + { + "epoch": 1.6023500988359323, + "grad_norm": 2.037592649459839, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6978579759597778, + "num_tokens": 374006939.0, + "step": 14591 + }, + { + "epoch": 1.602459916538546, + "grad_norm": 3.18253231048584, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7288622856140137, + "num_tokens": 374025716.0, + "step": 14592 + }, + { + "epoch": 1.6025697342411598, + "grad_norm": 2.207510471343994, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7101051807403564, + "num_tokens": 374054507.0, + "step": 14593 + }, + { + "epoch": 1.6026795519437733, + "grad_norm": 2.116366147994995, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7225675582885742, + "num_tokens": 374083110.0, + "step": 14594 + }, + { + "epoch": 1.6027893696463869, + "grad_norm": 2.2177963256835938, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6936005353927612, + "num_tokens": 374112152.0, + "step": 14595 + }, + { + "epoch": 1.6028991873490006, + "grad_norm": 2.079632043838501, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7306565642356873, + "num_tokens": 374139356.0, + "step": 14596 + }, + { + "epoch": 1.6030090050516144, + "grad_norm": 2.0493202209472656, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7261674404144287, + "num_tokens": 374168002.0, + "step": 14597 + }, + { + "epoch": 1.6031188227542281, + "grad_norm": 2.1210925579071045, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7123868465423584, + "num_tokens": 374193344.0, + "step": 14598 + }, + { + "epoch": 1.6032286404568417, + "grad_norm": 2.2653536796569824, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7092178463935852, + "num_tokens": 374219388.0, + "step": 14599 + }, + { + "epoch": 1.6033384581594552, + "grad_norm": 2.106112003326416, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7033315896987915, + "num_tokens": 374250404.0, + "step": 14600 + }, + { + "epoch": 1.603448275862069, + "grad_norm": 2.079425573348999, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.724488377571106, + "num_tokens": 374277584.0, + "step": 14601 + }, + { + "epoch": 1.6035580935646827, + "grad_norm": 2.379033327102661, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7562721967697144, + "num_tokens": 374301117.0, + "step": 14602 + }, + { + "epoch": 1.6036679112672962, + "grad_norm": 2.14788818359375, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.726102352142334, + "num_tokens": 374327668.0, + "step": 14603 + }, + { + "epoch": 1.60377772896991, + "grad_norm": 2.275294303894043, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7310100793838501, + "num_tokens": 374351091.0, + "step": 14604 + }, + { + "epoch": 1.6038875466725235, + "grad_norm": 2.3771491050720215, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7442752122879028, + "num_tokens": 374375117.0, + "step": 14605 + }, + { + "epoch": 1.6039973643751373, + "grad_norm": 2.1011385917663574, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7246624231338501, + "num_tokens": 374402346.0, + "step": 14606 + }, + { + "epoch": 1.604107182077751, + "grad_norm": 2.212620496749878, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7175211310386658, + "num_tokens": 374429389.0, + "step": 14607 + }, + { + "epoch": 1.6042169997803646, + "grad_norm": 2.3252460956573486, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7354932427406311, + "num_tokens": 374452944.0, + "step": 14608 + }, + { + "epoch": 1.6043268174829781, + "grad_norm": 2.5204684734344482, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.756955623626709, + "num_tokens": 374472227.0, + "step": 14609 + }, + { + "epoch": 1.6044366351855919, + "grad_norm": 1.975368857383728, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7274423241615295, + "num_tokens": 374504782.0, + "step": 14610 + }, + { + "epoch": 1.6045464528882056, + "grad_norm": 2.1705029010772705, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7009181976318359, + "num_tokens": 374534162.0, + "step": 14611 + }, + { + "epoch": 1.6046562705908194, + "grad_norm": 2.1090896129608154, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7014182209968567, + "num_tokens": 374565439.0, + "step": 14612 + }, + { + "epoch": 1.604766088293433, + "grad_norm": 2.269225835800171, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7324060201644897, + "num_tokens": 374589584.0, + "step": 14613 + }, + { + "epoch": 1.6048759059960465, + "grad_norm": 2.2033402919769287, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7263854742050171, + "num_tokens": 374614427.0, + "step": 14614 + }, + { + "epoch": 1.6049857236986602, + "grad_norm": 2.0525660514831543, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7153534293174744, + "num_tokens": 374643840.0, + "step": 14615 + }, + { + "epoch": 1.605095541401274, + "grad_norm": 2.2135839462280273, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7128535509109497, + "num_tokens": 374669766.0, + "step": 14616 + }, + { + "epoch": 1.6052053591038875, + "grad_norm": 2.2095892429351807, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.70896315574646, + "num_tokens": 374695690.0, + "step": 14617 + }, + { + "epoch": 1.605315176806501, + "grad_norm": 1.9859533309936523, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7182914018630981, + "num_tokens": 374726452.0, + "step": 14618 + }, + { + "epoch": 1.6054249945091148, + "grad_norm": 2.3598732948303223, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7283830642700195, + "num_tokens": 374748749.0, + "step": 14619 + }, + { + "epoch": 1.6055348122117286, + "grad_norm": 2.0006208419799805, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7017181515693665, + "num_tokens": 374778551.0, + "step": 14620 + }, + { + "epoch": 1.6056446299143423, + "grad_norm": 2.196251392364502, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7150821089744568, + "num_tokens": 374805832.0, + "step": 14621 + }, + { + "epoch": 1.6057544476169558, + "grad_norm": 2.1998324394226074, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7150970697402954, + "num_tokens": 374833419.0, + "step": 14622 + }, + { + "epoch": 1.6058642653195694, + "grad_norm": 2.1215457916259766, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7418710589408875, + "num_tokens": 374861053.0, + "step": 14623 + }, + { + "epoch": 1.6059740830221831, + "grad_norm": 2.1989471912384033, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6983513832092285, + "num_tokens": 374887341.0, + "step": 14624 + }, + { + "epoch": 1.606083900724797, + "grad_norm": 2.686769962310791, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7307575345039368, + "num_tokens": 374906275.0, + "step": 14625 + }, + { + "epoch": 1.6061937184274107, + "grad_norm": 2.623610019683838, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7332520484924316, + "num_tokens": 374925661.0, + "step": 14626 + }, + { + "epoch": 1.6063035361300242, + "grad_norm": 2.5474371910095215, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7471078038215637, + "num_tokens": 374945471.0, + "step": 14627 + }, + { + "epoch": 1.6064133538326377, + "grad_norm": 2.1480331420898438, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7132582664489746, + "num_tokens": 374973061.0, + "step": 14628 + }, + { + "epoch": 1.6065231715352515, + "grad_norm": 2.3777196407318115, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7211549282073975, + "num_tokens": 374996789.0, + "step": 14629 + }, + { + "epoch": 1.6066329892378652, + "grad_norm": 2.281336545944214, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.710040271282196, + "num_tokens": 375024049.0, + "step": 14630 + }, + { + "epoch": 1.6067428069404788, + "grad_norm": 2.1732118129730225, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6980652809143066, + "num_tokens": 375054298.0, + "step": 14631 + }, + { + "epoch": 1.6068526246430923, + "grad_norm": 2.2736430168151855, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7206659317016602, + "num_tokens": 375079292.0, + "step": 14632 + }, + { + "epoch": 1.606962442345706, + "grad_norm": 2.430229902267456, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7108840346336365, + "num_tokens": 375102062.0, + "step": 14633 + }, + { + "epoch": 1.6070722600483198, + "grad_norm": 2.2529470920562744, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7157021760940552, + "num_tokens": 375128005.0, + "step": 14634 + }, + { + "epoch": 1.6071820777509336, + "grad_norm": 2.002953052520752, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7151436805725098, + "num_tokens": 375160328.0, + "step": 14635 + }, + { + "epoch": 1.6072918954535471, + "grad_norm": 2.2822558879852295, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7095170021057129, + "num_tokens": 375186476.0, + "step": 14636 + }, + { + "epoch": 1.6074017131561606, + "grad_norm": 2.193650007247925, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7206847071647644, + "num_tokens": 375212678.0, + "step": 14637 + }, + { + "epoch": 1.6075115308587744, + "grad_norm": 2.1256182193756104, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7261034250259399, + "num_tokens": 375241482.0, + "step": 14638 + }, + { + "epoch": 1.6076213485613882, + "grad_norm": 2.0641777515411377, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7092340588569641, + "num_tokens": 375271489.0, + "step": 14639 + }, + { + "epoch": 1.607731166264002, + "grad_norm": 2.3675272464752197, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7126638889312744, + "num_tokens": 375295727.0, + "step": 14640 + }, + { + "epoch": 1.6078409839666155, + "grad_norm": 2.4233882427215576, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7343711256980896, + "num_tokens": 375317319.0, + "step": 14641 + }, + { + "epoch": 1.607950801669229, + "grad_norm": 1.9257761240005493, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7301527857780457, + "num_tokens": 375349095.0, + "step": 14642 + }, + { + "epoch": 1.6080606193718427, + "grad_norm": 1.9874728918075562, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7177161574363708, + "num_tokens": 375380984.0, + "step": 14643 + }, + { + "epoch": 1.6081704370744565, + "grad_norm": 2.228771924972534, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.720560610294342, + "num_tokens": 375407262.0, + "step": 14644 + }, + { + "epoch": 1.60828025477707, + "grad_norm": 2.3623695373535156, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6986052989959717, + "num_tokens": 375433624.0, + "step": 14645 + }, + { + "epoch": 1.6083900724796836, + "grad_norm": 2.2813549041748047, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7308149933815002, + "num_tokens": 375457296.0, + "step": 14646 + }, + { + "epoch": 1.6084998901822973, + "grad_norm": 2.2965452671051025, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7253822684288025, + "num_tokens": 375481772.0, + "step": 14647 + }, + { + "epoch": 1.608609707884911, + "grad_norm": 2.256352186203003, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7326809167861938, + "num_tokens": 375505864.0, + "step": 14648 + }, + { + "epoch": 1.6087195255875248, + "grad_norm": 2.2467281818389893, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7208161950111389, + "num_tokens": 375531929.0, + "step": 14649 + }, + { + "epoch": 1.6088293432901384, + "grad_norm": 2.1881515979766846, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7157107591629028, + "num_tokens": 375558242.0, + "step": 14650 + }, + { + "epoch": 1.608939160992752, + "grad_norm": 2.334817886352539, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7190914154052734, + "num_tokens": 375580772.0, + "step": 14651 + }, + { + "epoch": 1.6090489786953657, + "grad_norm": 2.391355276107788, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7235589027404785, + "num_tokens": 375604662.0, + "step": 14652 + }, + { + "epoch": 1.6091587963979794, + "grad_norm": 2.6323885917663574, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7122260332107544, + "num_tokens": 375625696.0, + "step": 14653 + }, + { + "epoch": 1.609268614100593, + "grad_norm": 2.328371047973633, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7151801586151123, + "num_tokens": 375649465.0, + "step": 14654 + }, + { + "epoch": 1.6093784318032067, + "grad_norm": 2.101181983947754, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6947337985038757, + "num_tokens": 375679564.0, + "step": 14655 + }, + { + "epoch": 1.6094882495058203, + "grad_norm": 2.229473352432251, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7127591371536255, + "num_tokens": 375704946.0, + "step": 14656 + }, + { + "epoch": 1.609598067208434, + "grad_norm": 2.2226648330688477, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7073278427124023, + "num_tokens": 375732568.0, + "step": 14657 + }, + { + "epoch": 1.6097078849110478, + "grad_norm": 2.5279085636138916, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7396632432937622, + "num_tokens": 375752797.0, + "step": 14658 + }, + { + "epoch": 1.6098177026136613, + "grad_norm": 2.3239171504974365, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7234480977058411, + "num_tokens": 375775863.0, + "step": 14659 + }, + { + "epoch": 1.6099275203162748, + "grad_norm": 2.275254249572754, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7175257205963135, + "num_tokens": 375800432.0, + "step": 14660 + }, + { + "epoch": 1.6100373380188886, + "grad_norm": 1.9805704355239868, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7316492199897766, + "num_tokens": 375832108.0, + "step": 14661 + }, + { + "epoch": 1.6101471557215024, + "grad_norm": 2.2843542098999023, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.741233229637146, + "num_tokens": 375858017.0, + "step": 14662 + }, + { + "epoch": 1.6102569734241161, + "grad_norm": 2.1049795150756836, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7063432335853577, + "num_tokens": 375888849.0, + "step": 14663 + }, + { + "epoch": 1.6103667911267296, + "grad_norm": 2.1136837005615234, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.704114556312561, + "num_tokens": 375918114.0, + "step": 14664 + }, + { + "epoch": 1.6104766088293432, + "grad_norm": 2.688833475112915, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7217633724212646, + "num_tokens": 375937855.0, + "step": 14665 + }, + { + "epoch": 1.610586426531957, + "grad_norm": 2.7438790798187256, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7156938910484314, + "num_tokens": 375956163.0, + "step": 14666 + }, + { + "epoch": 1.6106962442345707, + "grad_norm": 2.15147066116333, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.741156280040741, + "num_tokens": 375984426.0, + "step": 14667 + }, + { + "epoch": 1.6108060619371842, + "grad_norm": 2.2408509254455566, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7220234870910645, + "num_tokens": 376010796.0, + "step": 14668 + }, + { + "epoch": 1.610915879639798, + "grad_norm": 2.1005678176879883, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7013731002807617, + "num_tokens": 376039983.0, + "step": 14669 + }, + { + "epoch": 1.6110256973424115, + "grad_norm": 2.13193941116333, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7137277126312256, + "num_tokens": 376067302.0, + "step": 14670 + }, + { + "epoch": 1.6111355150450253, + "grad_norm": 2.269230604171753, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7087730765342712, + "num_tokens": 376093660.0, + "step": 14671 + }, + { + "epoch": 1.611245332747639, + "grad_norm": 1.9540541172027588, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.6970316767692566, + "num_tokens": 376125064.0, + "step": 14672 + }, + { + "epoch": 1.6113551504502526, + "grad_norm": 2.254038095474243, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.6994504928588867, + "num_tokens": 376152200.0, + "step": 14673 + }, + { + "epoch": 1.611464968152866, + "grad_norm": 2.162606954574585, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7115727663040161, + "num_tokens": 376180436.0, + "step": 14674 + }, + { + "epoch": 1.6115747858554799, + "grad_norm": 2.2397096157073975, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7304633259773254, + "num_tokens": 376206719.0, + "step": 14675 + }, + { + "epoch": 1.6116846035580936, + "grad_norm": 2.507275104522705, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7162343263626099, + "num_tokens": 376228075.0, + "step": 14676 + }, + { + "epoch": 1.6117944212607074, + "grad_norm": 2.012606620788574, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7118827104568481, + "num_tokens": 376258383.0, + "step": 14677 + }, + { + "epoch": 1.611904238963321, + "grad_norm": 2.440382719039917, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7078132629394531, + "num_tokens": 376281528.0, + "step": 14678 + }, + { + "epoch": 1.6120140566659344, + "grad_norm": 2.203584671020508, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7348352670669556, + "num_tokens": 376307024.0, + "step": 14679 + }, + { + "epoch": 1.6121238743685482, + "grad_norm": 2.6607768535614014, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7404308915138245, + "num_tokens": 376325205.0, + "step": 14680 + }, + { + "epoch": 1.612233692071162, + "grad_norm": 2.4854257106781006, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7010425329208374, + "num_tokens": 376346266.0, + "step": 14681 + }, + { + "epoch": 1.6123435097737755, + "grad_norm": 2.536226272583008, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7346044778823853, + "num_tokens": 376366120.0, + "step": 14682 + }, + { + "epoch": 1.612453327476389, + "grad_norm": 2.2096846103668213, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7317831516265869, + "num_tokens": 376395786.0, + "step": 14683 + }, + { + "epoch": 1.6125631451790028, + "grad_norm": 2.2648401260375977, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7147888541221619, + "num_tokens": 376422040.0, + "step": 14684 + }, + { + "epoch": 1.6126729628816165, + "grad_norm": 2.2588424682617188, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.729804515838623, + "num_tokens": 376447974.0, + "step": 14685 + }, + { + "epoch": 1.6127827805842303, + "grad_norm": 2.3334388732910156, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7294815182685852, + "num_tokens": 376472574.0, + "step": 14686 + }, + { + "epoch": 1.6128925982868438, + "grad_norm": 2.1118898391723633, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7177708148956299, + "num_tokens": 376502381.0, + "step": 14687 + }, + { + "epoch": 1.6130024159894574, + "grad_norm": 2.344214916229248, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.719100832939148, + "num_tokens": 376526422.0, + "step": 14688 + }, + { + "epoch": 1.6131122336920711, + "grad_norm": 2.4364545345306396, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7243027687072754, + "num_tokens": 376548149.0, + "step": 14689 + }, + { + "epoch": 1.6132220513946849, + "grad_norm": 2.5229289531707764, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7099891304969788, + "num_tokens": 376568810.0, + "step": 14690 + }, + { + "epoch": 1.6133318690972986, + "grad_norm": 2.4503819942474365, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7282765507698059, + "num_tokens": 376590793.0, + "step": 14691 + }, + { + "epoch": 1.6134416867999122, + "grad_norm": 2.296846389770508, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7136200666427612, + "num_tokens": 376616297.0, + "step": 14692 + }, + { + "epoch": 1.6135515045025257, + "grad_norm": 2.3090052604675293, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7151656150817871, + "num_tokens": 376640092.0, + "step": 14693 + }, + { + "epoch": 1.6136613222051395, + "grad_norm": 2.3217313289642334, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7177532315254211, + "num_tokens": 376665454.0, + "step": 14694 + }, + { + "epoch": 1.6137711399077532, + "grad_norm": 2.584681749343872, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7216947674751282, + "num_tokens": 376685659.0, + "step": 14695 + }, + { + "epoch": 1.6138809576103668, + "grad_norm": 2.5071749687194824, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7231038212776184, + "num_tokens": 376707533.0, + "step": 14696 + }, + { + "epoch": 1.6139907753129803, + "grad_norm": 2.1714723110198975, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7201773524284363, + "num_tokens": 376734564.0, + "step": 14697 + }, + { + "epoch": 1.614100593015594, + "grad_norm": 2.239231824874878, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7080974578857422, + "num_tokens": 376759661.0, + "step": 14698 + }, + { + "epoch": 1.6142104107182078, + "grad_norm": 2.742375373840332, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7258439064025879, + "num_tokens": 376776975.0, + "step": 14699 + }, + { + "epoch": 1.6143202284208216, + "grad_norm": 2.2478229999542236, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7197895050048828, + "num_tokens": 376804100.0, + "step": 14700 + }, + { + "epoch": 1.614430046123435, + "grad_norm": 2.290160655975342, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7087795734405518, + "num_tokens": 376829838.0, + "step": 14701 + }, + { + "epoch": 1.6145398638260486, + "grad_norm": 2.303260326385498, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.717766284942627, + "num_tokens": 376855559.0, + "step": 14702 + }, + { + "epoch": 1.6146496815286624, + "grad_norm": 2.326845645904541, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7195773124694824, + "num_tokens": 376879873.0, + "step": 14703 + }, + { + "epoch": 1.6147594992312762, + "grad_norm": 2.4181182384490967, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7215948104858398, + "num_tokens": 376900474.0, + "step": 14704 + }, + { + "epoch": 1.61486931693389, + "grad_norm": 2.3112080097198486, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7094284296035767, + "num_tokens": 376927236.0, + "step": 14705 + }, + { + "epoch": 1.6149791346365034, + "grad_norm": 2.157209873199463, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7260793447494507, + "num_tokens": 376954357.0, + "step": 14706 + }, + { + "epoch": 1.615088952339117, + "grad_norm": 2.370345115661621, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7192298173904419, + "num_tokens": 376976904.0, + "step": 14707 + }, + { + "epoch": 1.6151987700417307, + "grad_norm": 2.350419759750366, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7086864709854126, + "num_tokens": 377001009.0, + "step": 14708 + }, + { + "epoch": 1.6153085877443445, + "grad_norm": 2.1003201007843018, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7356805801391602, + "num_tokens": 377029701.0, + "step": 14709 + }, + { + "epoch": 1.615418405446958, + "grad_norm": 1.9235479831695557, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7013866901397705, + "num_tokens": 377063264.0, + "step": 14710 + }, + { + "epoch": 1.6155282231495716, + "grad_norm": 2.0189504623413086, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6937092542648315, + "num_tokens": 377094968.0, + "step": 14711 + }, + { + "epoch": 1.6156380408521853, + "grad_norm": 2.2185003757476807, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7149032354354858, + "num_tokens": 377119726.0, + "step": 14712 + }, + { + "epoch": 1.615747858554799, + "grad_norm": 2.179177761077881, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7096992135047913, + "num_tokens": 377144888.0, + "step": 14713 + }, + { + "epoch": 1.6158576762574128, + "grad_norm": 2.3706729412078857, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7176827192306519, + "num_tokens": 377172556.0, + "step": 14714 + }, + { + "epoch": 1.6159674939600264, + "grad_norm": 2.8835437297821045, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.744887113571167, + "num_tokens": 377189859.0, + "step": 14715 + }, + { + "epoch": 1.61607731166264, + "grad_norm": 2.307785749435425, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7116433382034302, + "num_tokens": 377214856.0, + "step": 14716 + }, + { + "epoch": 1.6161871293652537, + "grad_norm": 2.337148904800415, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7283568978309631, + "num_tokens": 377237301.0, + "step": 14717 + }, + { + "epoch": 1.6162969470678674, + "grad_norm": 2.2072091102600098, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7365527749061584, + "num_tokens": 377261253.0, + "step": 14718 + }, + { + "epoch": 1.616406764770481, + "grad_norm": 2.163247585296631, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7146357893943787, + "num_tokens": 377289222.0, + "step": 14719 + }, + { + "epoch": 1.6165165824730947, + "grad_norm": 2.5547285079956055, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.719628095626831, + "num_tokens": 377311623.0, + "step": 14720 + }, + { + "epoch": 1.6166264001757082, + "grad_norm": 2.385805130004883, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7094560861587524, + "num_tokens": 377336392.0, + "step": 14721 + }, + { + "epoch": 1.616736217878322, + "grad_norm": 2.388422727584839, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.731452465057373, + "num_tokens": 377358036.0, + "step": 14722 + }, + { + "epoch": 1.6168460355809358, + "grad_norm": 2.102675199508667, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7293207049369812, + "num_tokens": 377386497.0, + "step": 14723 + }, + { + "epoch": 1.6169558532835493, + "grad_norm": 2.0724332332611084, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7093228101730347, + "num_tokens": 377417407.0, + "step": 14724 + }, + { + "epoch": 1.6170656709861628, + "grad_norm": 2.2085752487182617, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6872442364692688, + "num_tokens": 377443431.0, + "step": 14725 + }, + { + "epoch": 1.6171754886887766, + "grad_norm": 1.9049845933914185, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7172681093215942, + "num_tokens": 377475955.0, + "step": 14726 + }, + { + "epoch": 1.6172853063913903, + "grad_norm": 2.2999210357666016, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.718009352684021, + "num_tokens": 377500796.0, + "step": 14727 + }, + { + "epoch": 1.617395124094004, + "grad_norm": 2.352900743484497, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7072070837020874, + "num_tokens": 377525110.0, + "step": 14728 + }, + { + "epoch": 1.6175049417966176, + "grad_norm": 1.9572149515151978, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7150431871414185, + "num_tokens": 377557292.0, + "step": 14729 + }, + { + "epoch": 1.6176147594992312, + "grad_norm": 2.2406036853790283, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.704805850982666, + "num_tokens": 377586164.0, + "step": 14730 + }, + { + "epoch": 1.617724577201845, + "grad_norm": 2.102090358734131, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6993541717529297, + "num_tokens": 377614868.0, + "step": 14731 + }, + { + "epoch": 1.6178343949044587, + "grad_norm": 2.319251537322998, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7330530881881714, + "num_tokens": 377638580.0, + "step": 14732 + }, + { + "epoch": 1.6179442126070722, + "grad_norm": 1.842613935470581, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6972891688346863, + "num_tokens": 377676854.0, + "step": 14733 + }, + { + "epoch": 1.618054030309686, + "grad_norm": 2.513843059539795, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7110571265220642, + "num_tokens": 377700947.0, + "step": 14734 + }, + { + "epoch": 1.6181638480122995, + "grad_norm": 2.3080661296844482, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7243194580078125, + "num_tokens": 377724552.0, + "step": 14735 + }, + { + "epoch": 1.6182736657149133, + "grad_norm": 2.6156628131866455, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7440811991691589, + "num_tokens": 377744458.0, + "step": 14736 + }, + { + "epoch": 1.618383483417527, + "grad_norm": 2.5124175548553467, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.7161551117897034, + "num_tokens": 377770140.0, + "step": 14737 + }, + { + "epoch": 1.6184933011201406, + "grad_norm": 2.3385303020477295, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7194249629974365, + "num_tokens": 377796825.0, + "step": 14738 + }, + { + "epoch": 1.618603118822754, + "grad_norm": 2.2119641304016113, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7010112404823303, + "num_tokens": 377823165.0, + "step": 14739 + }, + { + "epoch": 1.6187129365253679, + "grad_norm": 2.4691720008850098, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7554855346679688, + "num_tokens": 377843909.0, + "step": 14740 + }, + { + "epoch": 1.6188227542279816, + "grad_norm": 2.1767079830169678, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7084696292877197, + "num_tokens": 377870873.0, + "step": 14741 + }, + { + "epoch": 1.6189325719305954, + "grad_norm": 2.0668704509735107, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7096455693244934, + "num_tokens": 377902523.0, + "step": 14742 + }, + { + "epoch": 1.619042389633209, + "grad_norm": 1.9486628770828247, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7238942384719849, + "num_tokens": 377934782.0, + "step": 14743 + }, + { + "epoch": 1.6191522073358224, + "grad_norm": 2.5448615550994873, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7460688352584839, + "num_tokens": 377955052.0, + "step": 14744 + }, + { + "epoch": 1.6192620250384362, + "grad_norm": 2.19842267036438, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7309091091156006, + "num_tokens": 377980666.0, + "step": 14745 + }, + { + "epoch": 1.61937184274105, + "grad_norm": 2.218416690826416, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7162414789199829, + "num_tokens": 378006083.0, + "step": 14746 + }, + { + "epoch": 1.6194816604436635, + "grad_norm": 2.4536798000335693, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7229822874069214, + "num_tokens": 378028773.0, + "step": 14747 + }, + { + "epoch": 1.619591478146277, + "grad_norm": 2.126055955886841, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6966632008552551, + "num_tokens": 378056909.0, + "step": 14748 + }, + { + "epoch": 1.6197012958488908, + "grad_norm": 2.0686726570129395, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7351404428482056, + "num_tokens": 378086046.0, + "step": 14749 + }, + { + "epoch": 1.6198111135515045, + "grad_norm": 2.32167911529541, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7346364259719849, + "num_tokens": 378109957.0, + "step": 14750 + }, + { + "epoch": 1.6199209312541183, + "grad_norm": 2.2379491329193115, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7507272958755493, + "num_tokens": 378131984.0, + "step": 14751 + }, + { + "epoch": 1.6200307489567318, + "grad_norm": 2.0894935131073, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7241690754890442, + "num_tokens": 378159807.0, + "step": 14752 + }, + { + "epoch": 1.6201405666593454, + "grad_norm": 2.3069536685943604, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7325316071510315, + "num_tokens": 378182910.0, + "step": 14753 + }, + { + "epoch": 1.6202503843619591, + "grad_norm": 2.3278441429138184, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7381759881973267, + "num_tokens": 378204089.0, + "step": 14754 + }, + { + "epoch": 1.6203602020645729, + "grad_norm": 2.155256509780884, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7247961163520813, + "num_tokens": 378232683.0, + "step": 14755 + }, + { + "epoch": 1.6204700197671866, + "grad_norm": 2.2826080322265625, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7181304693222046, + "num_tokens": 378257380.0, + "step": 14756 + }, + { + "epoch": 1.6205798374698002, + "grad_norm": 2.109562635421753, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7142317295074463, + "num_tokens": 378286102.0, + "step": 14757 + }, + { + "epoch": 1.6206896551724137, + "grad_norm": 2.13828706741333, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.712498664855957, + "num_tokens": 378315705.0, + "step": 14758 + }, + { + "epoch": 1.6207994728750275, + "grad_norm": 2.5181984901428223, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7228315472602844, + "num_tokens": 378336386.0, + "step": 14759 + }, + { + "epoch": 1.6209092905776412, + "grad_norm": 2.1282031536102295, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7368274927139282, + "num_tokens": 378365431.0, + "step": 14760 + }, + { + "epoch": 1.6210191082802548, + "grad_norm": 2.354813575744629, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7157611846923828, + "num_tokens": 378390137.0, + "step": 14761 + }, + { + "epoch": 1.6211289259828683, + "grad_norm": 2.1947033405303955, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7219136357307434, + "num_tokens": 378417528.0, + "step": 14762 + }, + { + "epoch": 1.621238743685482, + "grad_norm": 2.0376193523406982, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.706749439239502, + "num_tokens": 378450164.0, + "step": 14763 + }, + { + "epoch": 1.6213485613880958, + "grad_norm": 2.2209715843200684, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7355459332466125, + "num_tokens": 378475412.0, + "step": 14764 + }, + { + "epoch": 1.6214583790907096, + "grad_norm": 2.1217074394226074, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7221888899803162, + "num_tokens": 378502708.0, + "step": 14765 + }, + { + "epoch": 1.621568196793323, + "grad_norm": 2.4613072872161865, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7115516662597656, + "num_tokens": 378525269.0, + "step": 14766 + }, + { + "epoch": 1.6216780144959366, + "grad_norm": 2.4225521087646484, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.715205729007721, + "num_tokens": 378548575.0, + "step": 14767 + }, + { + "epoch": 1.6217878321985504, + "grad_norm": 2.4349312782287598, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7395687103271484, + "num_tokens": 378571785.0, + "step": 14768 + }, + { + "epoch": 1.6218976499011641, + "grad_norm": 2.2396798133850098, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7147597074508667, + "num_tokens": 378598047.0, + "step": 14769 + }, + { + "epoch": 1.6220074676037777, + "grad_norm": 2.4150702953338623, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7237159013748169, + "num_tokens": 378622232.0, + "step": 14770 + }, + { + "epoch": 1.6221172853063914, + "grad_norm": 2.0293800830841064, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6971486806869507, + "num_tokens": 378653364.0, + "step": 14771 + }, + { + "epoch": 1.622227103009005, + "grad_norm": 2.217825412750244, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6916738152503967, + "num_tokens": 378682347.0, + "step": 14772 + }, + { + "epoch": 1.6223369207116187, + "grad_norm": 2.6289689540863037, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7105913162231445, + "num_tokens": 378703010.0, + "step": 14773 + }, + { + "epoch": 1.6224467384142325, + "grad_norm": 2.156428337097168, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7096231579780579, + "num_tokens": 378729529.0, + "step": 14774 + }, + { + "epoch": 1.622556556116846, + "grad_norm": 2.1000781059265137, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7392454147338867, + "num_tokens": 378756201.0, + "step": 14775 + }, + { + "epoch": 1.6226663738194596, + "grad_norm": 2.3582942485809326, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7342456579208374, + "num_tokens": 378779156.0, + "step": 14776 + }, + { + "epoch": 1.6227761915220733, + "grad_norm": 2.3614416122436523, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7252365946769714, + "num_tokens": 378803203.0, + "step": 14777 + }, + { + "epoch": 1.622886009224687, + "grad_norm": 2.2844974994659424, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7000511884689331, + "num_tokens": 378829790.0, + "step": 14778 + }, + { + "epoch": 1.6229958269273008, + "grad_norm": 2.103410482406616, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.717047393321991, + "num_tokens": 378860244.0, + "step": 14779 + }, + { + "epoch": 1.6231056446299144, + "grad_norm": 2.273319721221924, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7143521308898926, + "num_tokens": 378887048.0, + "step": 14780 + }, + { + "epoch": 1.623215462332528, + "grad_norm": 2.422426700592041, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7117125988006592, + "num_tokens": 378910665.0, + "step": 14781 + }, + { + "epoch": 1.6233252800351416, + "grad_norm": 2.087754726409912, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7244415283203125, + "num_tokens": 378939375.0, + "step": 14782 + }, + { + "epoch": 1.6234350977377554, + "grad_norm": 2.2424867153167725, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7149908542633057, + "num_tokens": 378963620.0, + "step": 14783 + }, + { + "epoch": 1.623544915440369, + "grad_norm": 2.094695806503296, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7283496260643005, + "num_tokens": 378992679.0, + "step": 14784 + }, + { + "epoch": 1.6236547331429827, + "grad_norm": 2.3512771129608154, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.715447187423706, + "num_tokens": 379016569.0, + "step": 14785 + }, + { + "epoch": 1.6237645508455962, + "grad_norm": 1.9226746559143066, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7157915234565735, + "num_tokens": 379048438.0, + "step": 14786 + }, + { + "epoch": 1.62387436854821, + "grad_norm": 2.0691990852355957, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7268046140670776, + "num_tokens": 379076334.0, + "step": 14787 + }, + { + "epoch": 1.6239841862508237, + "grad_norm": 2.3024325370788574, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7330660820007324, + "num_tokens": 379100221.0, + "step": 14788 + }, + { + "epoch": 1.6240940039534373, + "grad_norm": 2.198835849761963, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7119851112365723, + "num_tokens": 379127696.0, + "step": 14789 + }, + { + "epoch": 1.6242038216560508, + "grad_norm": 2.2677433490753174, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7091991305351257, + "num_tokens": 379155131.0, + "step": 14790 + }, + { + "epoch": 1.6243136393586646, + "grad_norm": 1.9319661855697632, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.6998783349990845, + "num_tokens": 379189616.0, + "step": 14791 + }, + { + "epoch": 1.6244234570612783, + "grad_norm": 2.066953420639038, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7005470991134644, + "num_tokens": 379220761.0, + "step": 14792 + }, + { + "epoch": 1.624533274763892, + "grad_norm": 1.9831146001815796, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7157696485519409, + "num_tokens": 379251390.0, + "step": 14793 + }, + { + "epoch": 1.6246430924665056, + "grad_norm": 2.255323648452759, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7347778677940369, + "num_tokens": 379276437.0, + "step": 14794 + }, + { + "epoch": 1.6247529101691192, + "grad_norm": 2.3261220455169678, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7225192785263062, + "num_tokens": 379300317.0, + "step": 14795 + }, + { + "epoch": 1.624862727871733, + "grad_norm": 2.1035072803497314, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.697742223739624, + "num_tokens": 379330574.0, + "step": 14796 + }, + { + "epoch": 1.6249725455743467, + "grad_norm": 2.60905385017395, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7545946836471558, + "num_tokens": 379348816.0, + "step": 14797 + }, + { + "epoch": 1.6250823632769602, + "grad_norm": 2.2684614658355713, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7059985995292664, + "num_tokens": 379374586.0, + "step": 14798 + }, + { + "epoch": 1.6251921809795737, + "grad_norm": 2.3264734745025635, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.735033392906189, + "num_tokens": 379398245.0, + "step": 14799 + }, + { + "epoch": 1.6253019986821875, + "grad_norm": 2.2364540100097656, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7231518030166626, + "num_tokens": 379424987.0, + "step": 14800 + }, + { + "epoch": 1.6254118163848013, + "grad_norm": 2.035717487335205, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7107813358306885, + "num_tokens": 379456756.0, + "step": 14801 + }, + { + "epoch": 1.625521634087415, + "grad_norm": 2.1804215908050537, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7278170585632324, + "num_tokens": 379481977.0, + "step": 14802 + }, + { + "epoch": 1.6256314517900285, + "grad_norm": 2.3166401386260986, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7114250659942627, + "num_tokens": 379506511.0, + "step": 14803 + }, + { + "epoch": 1.625741269492642, + "grad_norm": 2.3454129695892334, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7306157350540161, + "num_tokens": 379529110.0, + "step": 14804 + }, + { + "epoch": 1.6258510871952558, + "grad_norm": 2.1036200523376465, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7428890466690063, + "num_tokens": 379556163.0, + "step": 14805 + }, + { + "epoch": 1.6259609048978696, + "grad_norm": 1.9746397733688354, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7060196399688721, + "num_tokens": 379587572.0, + "step": 14806 + }, + { + "epoch": 1.6260707226004834, + "grad_norm": 2.0309009552001953, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7207372188568115, + "num_tokens": 379617987.0, + "step": 14807 + }, + { + "epoch": 1.6261805403030969, + "grad_norm": 2.3133060932159424, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7198566198348999, + "num_tokens": 379644659.0, + "step": 14808 + }, + { + "epoch": 1.6262903580057104, + "grad_norm": 2.209606409072876, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7177951335906982, + "num_tokens": 379670943.0, + "step": 14809 + }, + { + "epoch": 1.6264001757083242, + "grad_norm": 2.2103869915008545, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7444232106208801, + "num_tokens": 379694888.0, + "step": 14810 + }, + { + "epoch": 1.626509993410938, + "grad_norm": 2.5783004760742188, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7126578688621521, + "num_tokens": 379715473.0, + "step": 14811 + }, + { + "epoch": 1.6266198111135515, + "grad_norm": 2.4897561073303223, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7158209085464478, + "num_tokens": 379737368.0, + "step": 14812 + }, + { + "epoch": 1.626729628816165, + "grad_norm": 2.368347644805908, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7380077838897705, + "num_tokens": 379761543.0, + "step": 14813 + }, + { + "epoch": 1.6268394465187788, + "grad_norm": 2.1284682750701904, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7193634510040283, + "num_tokens": 379791853.0, + "step": 14814 + }, + { + "epoch": 1.6269492642213925, + "grad_norm": 2.302330493927002, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7197084426879883, + "num_tokens": 379815633.0, + "step": 14815 + }, + { + "epoch": 1.6270590819240063, + "grad_norm": 2.1384494304656982, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7246006727218628, + "num_tokens": 379842324.0, + "step": 14816 + }, + { + "epoch": 1.6271688996266198, + "grad_norm": 2.0211806297302246, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7187747359275818, + "num_tokens": 379873552.0, + "step": 14817 + }, + { + "epoch": 1.6272787173292333, + "grad_norm": 2.512622833251953, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7508724927902222, + "num_tokens": 379892509.0, + "step": 14818 + }, + { + "epoch": 1.627388535031847, + "grad_norm": 2.3576290607452393, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7122398614883423, + "num_tokens": 379918061.0, + "step": 14819 + }, + { + "epoch": 1.6274983527344609, + "grad_norm": 1.998484492301941, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7170053720474243, + "num_tokens": 379948087.0, + "step": 14820 + }, + { + "epoch": 1.6276081704370746, + "grad_norm": 2.083347797393799, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7219551801681519, + "num_tokens": 379976345.0, + "step": 14821 + }, + { + "epoch": 1.6277179881396882, + "grad_norm": 2.1481051445007324, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7389001250267029, + "num_tokens": 380001735.0, + "step": 14822 + }, + { + "epoch": 1.6278278058423017, + "grad_norm": 2.4117579460144043, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7365247011184692, + "num_tokens": 380024182.0, + "step": 14823 + }, + { + "epoch": 1.6279376235449154, + "grad_norm": 2.374260902404785, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7315804958343506, + "num_tokens": 380048128.0, + "step": 14824 + }, + { + "epoch": 1.6280474412475292, + "grad_norm": 2.0517497062683105, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7179754972457886, + "num_tokens": 380078004.0, + "step": 14825 + }, + { + "epoch": 1.6281572589501427, + "grad_norm": 2.3043899536132812, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7040264010429382, + "num_tokens": 380103521.0, + "step": 14826 + }, + { + "epoch": 1.6282670766527563, + "grad_norm": 2.098600387573242, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7161344289779663, + "num_tokens": 380133259.0, + "step": 14827 + }, + { + "epoch": 1.62837689435537, + "grad_norm": 2.242482900619507, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.717146098613739, + "num_tokens": 380160062.0, + "step": 14828 + }, + { + "epoch": 1.6284867120579838, + "grad_norm": 2.2896246910095215, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7012604475021362, + "num_tokens": 380186356.0, + "step": 14829 + }, + { + "epoch": 1.6285965297605975, + "grad_norm": 2.123310089111328, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7038559317588806, + "num_tokens": 380215295.0, + "step": 14830 + }, + { + "epoch": 1.628706347463211, + "grad_norm": 2.162243127822876, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7222861051559448, + "num_tokens": 380242828.0, + "step": 14831 + }, + { + "epoch": 1.6288161651658246, + "grad_norm": 2.1889662742614746, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7060525417327881, + "num_tokens": 380269013.0, + "step": 14832 + }, + { + "epoch": 1.6289259828684384, + "grad_norm": 2.422177791595459, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7271592617034912, + "num_tokens": 380292970.0, + "step": 14833 + }, + { + "epoch": 1.6290358005710521, + "grad_norm": 2.356604814529419, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7151966691017151, + "num_tokens": 380315437.0, + "step": 14834 + }, + { + "epoch": 1.6291456182736657, + "grad_norm": 2.4155588150024414, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7375131249427795, + "num_tokens": 380338612.0, + "step": 14835 + }, + { + "epoch": 1.6292554359762794, + "grad_norm": 2.1339006423950195, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.701593816280365, + "num_tokens": 380366653.0, + "step": 14836 + }, + { + "epoch": 1.629365253678893, + "grad_norm": 1.891860842704773, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6888537406921387, + "num_tokens": 380402968.0, + "step": 14837 + }, + { + "epoch": 1.6294750713815067, + "grad_norm": 2.7641103267669678, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.734062671661377, + "num_tokens": 380420311.0, + "step": 14838 + }, + { + "epoch": 1.6295848890841205, + "grad_norm": 2.4129161834716797, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6975336670875549, + "num_tokens": 380444913.0, + "step": 14839 + }, + { + "epoch": 1.629694706786734, + "grad_norm": 1.9935870170593262, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6978096961975098, + "num_tokens": 380476500.0, + "step": 14840 + }, + { + "epoch": 1.6298045244893475, + "grad_norm": 2.4292805194854736, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7379986047744751, + "num_tokens": 380497964.0, + "step": 14841 + }, + { + "epoch": 1.6299143421919613, + "grad_norm": 2.3753857612609863, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7243829965591431, + "num_tokens": 380521502.0, + "step": 14842 + }, + { + "epoch": 1.630024159894575, + "grad_norm": 2.221879005432129, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7011844515800476, + "num_tokens": 380547603.0, + "step": 14843 + }, + { + "epoch": 1.6301339775971888, + "grad_norm": 2.2760181427001953, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7232047319412231, + "num_tokens": 380572654.0, + "step": 14844 + }, + { + "epoch": 1.6302437952998023, + "grad_norm": 2.180819272994995, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7167226672172546, + "num_tokens": 380600475.0, + "step": 14845 + }, + { + "epoch": 1.6303536130024159, + "grad_norm": 2.2507994174957275, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7160906195640564, + "num_tokens": 380629687.0, + "step": 14846 + }, + { + "epoch": 1.6304634307050296, + "grad_norm": 2.1933019161224365, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.73481684923172, + "num_tokens": 380657118.0, + "step": 14847 + }, + { + "epoch": 1.6305732484076434, + "grad_norm": 2.2251579761505127, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7106784582138062, + "num_tokens": 380683166.0, + "step": 14848 + }, + { + "epoch": 1.630683066110257, + "grad_norm": 2.2897212505340576, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7219887375831604, + "num_tokens": 380707108.0, + "step": 14849 + }, + { + "epoch": 1.6307928838128707, + "grad_norm": 2.3868954181671143, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7282528877258301, + "num_tokens": 380732609.0, + "step": 14850 + }, + { + "epoch": 1.6309027015154842, + "grad_norm": 2.310135841369629, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.734188437461853, + "num_tokens": 380756241.0, + "step": 14851 + }, + { + "epoch": 1.631012519218098, + "grad_norm": 2.224013566970825, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7184577584266663, + "num_tokens": 380783347.0, + "step": 14852 + }, + { + "epoch": 1.6311223369207117, + "grad_norm": 2.3136496543884277, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7526301145553589, + "num_tokens": 380805137.0, + "step": 14853 + }, + { + "epoch": 1.6312321546233253, + "grad_norm": 2.6076533794403076, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7248129844665527, + "num_tokens": 380824668.0, + "step": 14854 + }, + { + "epoch": 1.6313419723259388, + "grad_norm": 2.273622989654541, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7312510013580322, + "num_tokens": 380854656.0, + "step": 14855 + }, + { + "epoch": 1.6314517900285526, + "grad_norm": 2.093946695327759, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.711246132850647, + "num_tokens": 380886220.0, + "step": 14856 + }, + { + "epoch": 1.6315616077311663, + "grad_norm": 2.0736382007598877, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7644938230514526, + "num_tokens": 380912626.0, + "step": 14857 + }, + { + "epoch": 1.63167142543378, + "grad_norm": 2.500990390777588, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7029275894165039, + "num_tokens": 380934944.0, + "step": 14858 + }, + { + "epoch": 1.6317812431363936, + "grad_norm": 2.3224217891693115, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7261073589324951, + "num_tokens": 380959215.0, + "step": 14859 + }, + { + "epoch": 1.6318910608390071, + "grad_norm": 2.2232818603515625, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.709197998046875, + "num_tokens": 380987814.0, + "step": 14860 + }, + { + "epoch": 1.632000878541621, + "grad_norm": 2.256969928741455, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.719733715057373, + "num_tokens": 381011736.0, + "step": 14861 + }, + { + "epoch": 1.6321106962442347, + "grad_norm": 2.2516937255859375, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7451515793800354, + "num_tokens": 381035686.0, + "step": 14862 + }, + { + "epoch": 1.6322205139468482, + "grad_norm": 2.1209933757781982, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7270032167434692, + "num_tokens": 381064570.0, + "step": 14863 + }, + { + "epoch": 1.6323303316494617, + "grad_norm": 2.1386966705322266, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7130626440048218, + "num_tokens": 381091812.0, + "step": 14864 + }, + { + "epoch": 1.6324401493520755, + "grad_norm": 2.3956758975982666, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.732850193977356, + "num_tokens": 381113374.0, + "step": 14865 + }, + { + "epoch": 1.6325499670546892, + "grad_norm": 2.2535765171051025, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.6988601088523865, + "num_tokens": 381139188.0, + "step": 14866 + }, + { + "epoch": 1.632659784757303, + "grad_norm": 2.2763404846191406, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7252051830291748, + "num_tokens": 381164512.0, + "step": 14867 + }, + { + "epoch": 1.6327696024599165, + "grad_norm": 2.465278148651123, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7397798299789429, + "num_tokens": 381186228.0, + "step": 14868 + }, + { + "epoch": 1.63287942016253, + "grad_norm": 2.07136607170105, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6978441476821899, + "num_tokens": 381216623.0, + "step": 14869 + }, + { + "epoch": 1.6329892378651438, + "grad_norm": 2.12489652633667, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7267814874649048, + "num_tokens": 381242743.0, + "step": 14870 + }, + { + "epoch": 1.6330990555677576, + "grad_norm": 2.4349920749664307, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6918667554855347, + "num_tokens": 381268516.0, + "step": 14871 + }, + { + "epoch": 1.6332088732703713, + "grad_norm": 2.4277942180633545, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7238104343414307, + "num_tokens": 381291418.0, + "step": 14872 + }, + { + "epoch": 1.6333186909729849, + "grad_norm": 2.4044575691223145, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7071378231048584, + "num_tokens": 381317839.0, + "step": 14873 + }, + { + "epoch": 1.6334285086755984, + "grad_norm": 2.3291215896606445, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7333947420120239, + "num_tokens": 381339629.0, + "step": 14874 + }, + { + "epoch": 1.6335383263782122, + "grad_norm": 2.318267822265625, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7318242788314819, + "num_tokens": 381364636.0, + "step": 14875 + }, + { + "epoch": 1.633648144080826, + "grad_norm": 2.1718404293060303, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7554740905761719, + "num_tokens": 381389206.0, + "step": 14876 + }, + { + "epoch": 1.6337579617834395, + "grad_norm": 1.9739012718200684, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7104315757751465, + "num_tokens": 381422729.0, + "step": 14877 + }, + { + "epoch": 1.633867779486053, + "grad_norm": 2.525125503540039, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7309198379516602, + "num_tokens": 381443311.0, + "step": 14878 + }, + { + "epoch": 1.6339775971886668, + "grad_norm": 2.3964760303497314, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7208210825920105, + "num_tokens": 381468036.0, + "step": 14879 + }, + { + "epoch": 1.6340874148912805, + "grad_norm": 2.245481252670288, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7070188522338867, + "num_tokens": 381494558.0, + "step": 14880 + }, + { + "epoch": 1.6341972325938943, + "grad_norm": 2.6228461265563965, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7052906155586243, + "num_tokens": 381516078.0, + "step": 14881 + }, + { + "epoch": 1.6343070502965078, + "grad_norm": 2.5201220512390137, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7314580678939819, + "num_tokens": 381537079.0, + "step": 14882 + }, + { + "epoch": 1.6344168679991213, + "grad_norm": 2.3713340759277344, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7250381112098694, + "num_tokens": 381562960.0, + "step": 14883 + }, + { + "epoch": 1.634526685701735, + "grad_norm": 2.059375286102295, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7296273708343506, + "num_tokens": 381592329.0, + "step": 14884 + }, + { + "epoch": 1.6346365034043489, + "grad_norm": 2.3270580768585205, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7289987802505493, + "num_tokens": 381616345.0, + "step": 14885 + }, + { + "epoch": 1.6347463211069626, + "grad_norm": 2.031484365463257, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7322641015052795, + "num_tokens": 381647185.0, + "step": 14886 + }, + { + "epoch": 1.6348561388095761, + "grad_norm": 2.359574556350708, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7288801670074463, + "num_tokens": 381669642.0, + "step": 14887 + }, + { + "epoch": 1.6349659565121897, + "grad_norm": 2.42679762840271, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.726133406162262, + "num_tokens": 381693934.0, + "step": 14888 + }, + { + "epoch": 1.6350757742148034, + "grad_norm": 2.700284004211426, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7337496280670166, + "num_tokens": 381715023.0, + "step": 14889 + }, + { + "epoch": 1.6351855919174172, + "grad_norm": 2.8873250484466553, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7487098574638367, + "num_tokens": 381731552.0, + "step": 14890 + }, + { + "epoch": 1.6352954096200307, + "grad_norm": 2.4254536628723145, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7010424137115479, + "num_tokens": 381756504.0, + "step": 14891 + }, + { + "epoch": 1.6354052273226443, + "grad_norm": 2.4102439880371094, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7246218323707581, + "num_tokens": 381778326.0, + "step": 14892 + }, + { + "epoch": 1.635515045025258, + "grad_norm": 2.11684513092041, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7132443189620972, + "num_tokens": 381809416.0, + "step": 14893 + }, + { + "epoch": 1.6356248627278718, + "grad_norm": 2.185875654220581, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6885911226272583, + "num_tokens": 381836868.0, + "step": 14894 + }, + { + "epoch": 1.6357346804304855, + "grad_norm": 2.469820737838745, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7234194278717041, + "num_tokens": 381857647.0, + "step": 14895 + }, + { + "epoch": 1.635844498133099, + "grad_norm": 2.336280345916748, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7351371645927429, + "num_tokens": 381880940.0, + "step": 14896 + }, + { + "epoch": 1.6359543158357126, + "grad_norm": 2.0820884704589844, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7116906642913818, + "num_tokens": 381909067.0, + "step": 14897 + }, + { + "epoch": 1.6360641335383264, + "grad_norm": 2.155411958694458, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7325825095176697, + "num_tokens": 381936725.0, + "step": 14898 + }, + { + "epoch": 1.6361739512409401, + "grad_norm": 2.2168476581573486, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6996521949768066, + "num_tokens": 381962831.0, + "step": 14899 + }, + { + "epoch": 1.6362837689435537, + "grad_norm": 2.464782238006592, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.721433162689209, + "num_tokens": 381984701.0, + "step": 14900 + }, + { + "epoch": 1.6363935866461674, + "grad_norm": 2.8332903385162354, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7520668506622314, + "num_tokens": 382001671.0, + "step": 14901 + }, + { + "epoch": 1.636503404348781, + "grad_norm": 2.356870412826538, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7433210611343384, + "num_tokens": 382025593.0, + "step": 14902 + }, + { + "epoch": 1.6366132220513947, + "grad_norm": 2.0359983444213867, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7222480773925781, + "num_tokens": 382054250.0, + "step": 14903 + }, + { + "epoch": 1.6367230397540085, + "grad_norm": 2.248810291290283, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7186834812164307, + "num_tokens": 382080215.0, + "step": 14904 + }, + { + "epoch": 1.636832857456622, + "grad_norm": 2.266712188720703, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7219606041908264, + "num_tokens": 382109201.0, + "step": 14905 + }, + { + "epoch": 1.6369426751592355, + "grad_norm": 2.0769803524017334, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7185407876968384, + "num_tokens": 382136876.0, + "step": 14906 + }, + { + "epoch": 1.6370524928618493, + "grad_norm": 2.736936330795288, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7296444773674011, + "num_tokens": 382157300.0, + "step": 14907 + }, + { + "epoch": 1.637162310564463, + "grad_norm": 1.9864566326141357, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7287817001342773, + "num_tokens": 382186122.0, + "step": 14908 + }, + { + "epoch": 1.6372721282670768, + "grad_norm": 2.0931153297424316, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7001276612281799, + "num_tokens": 382216028.0, + "step": 14909 + }, + { + "epoch": 1.6373819459696903, + "grad_norm": 2.201237440109253, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7262672185897827, + "num_tokens": 382240449.0, + "step": 14910 + }, + { + "epoch": 1.6374917636723039, + "grad_norm": 2.218418836593628, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.6987854242324829, + "num_tokens": 382266155.0, + "step": 14911 + }, + { + "epoch": 1.6376015813749176, + "grad_norm": 2.353135347366333, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.737662136554718, + "num_tokens": 382289794.0, + "step": 14912 + }, + { + "epoch": 1.6377113990775314, + "grad_norm": 2.335857629776001, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6981523036956787, + "num_tokens": 382313591.0, + "step": 14913 + }, + { + "epoch": 1.637821216780145, + "grad_norm": 2.0868473052978516, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7347559332847595, + "num_tokens": 382342543.0, + "step": 14914 + }, + { + "epoch": 1.6379310344827587, + "grad_norm": 2.239583969116211, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7014776468276978, + "num_tokens": 382369014.0, + "step": 14915 + }, + { + "epoch": 1.6380408521853722, + "grad_norm": 2.5718257427215576, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7134074568748474, + "num_tokens": 382389053.0, + "step": 14916 + }, + { + "epoch": 1.638150669887986, + "grad_norm": 2.2622547149658203, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7164269685745239, + "num_tokens": 382414690.0, + "step": 14917 + }, + { + "epoch": 1.6382604875905997, + "grad_norm": 2.509018659591675, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7248220443725586, + "num_tokens": 382436120.0, + "step": 14918 + }, + { + "epoch": 1.6383703052932133, + "grad_norm": 2.473498821258545, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7423050999641418, + "num_tokens": 382456874.0, + "step": 14919 + }, + { + "epoch": 1.6384801229958268, + "grad_norm": 2.2627334594726562, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7188151478767395, + "num_tokens": 382482742.0, + "step": 14920 + }, + { + "epoch": 1.6385899406984406, + "grad_norm": 2.233140468597412, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7057052850723267, + "num_tokens": 382511035.0, + "step": 14921 + }, + { + "epoch": 1.6386997584010543, + "grad_norm": 2.192986249923706, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7165801525115967, + "num_tokens": 382537454.0, + "step": 14922 + }, + { + "epoch": 1.638809576103668, + "grad_norm": 2.0545477867126465, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7401993274688721, + "num_tokens": 382565037.0, + "step": 14923 + }, + { + "epoch": 1.6389193938062816, + "grad_norm": 1.9523965120315552, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.688579797744751, + "num_tokens": 382598045.0, + "step": 14924 + }, + { + "epoch": 1.6390292115088951, + "grad_norm": 2.193074941635132, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7279255390167236, + "num_tokens": 382625442.0, + "step": 14925 + }, + { + "epoch": 1.639139029211509, + "grad_norm": 2.340789556503296, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7188611030578613, + "num_tokens": 382649428.0, + "step": 14926 + }, + { + "epoch": 1.6392488469141226, + "grad_norm": 2.2452003955841064, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7455725073814392, + "num_tokens": 382672321.0, + "step": 14927 + }, + { + "epoch": 1.6393586646167362, + "grad_norm": 2.130849599838257, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7030133008956909, + "num_tokens": 382702972.0, + "step": 14928 + }, + { + "epoch": 1.6394684823193497, + "grad_norm": 2.085649251937866, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7318595051765442, + "num_tokens": 382731149.0, + "step": 14929 + }, + { + "epoch": 1.6395783000219635, + "grad_norm": 2.074734926223755, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7212987542152405, + "num_tokens": 382760766.0, + "step": 14930 + }, + { + "epoch": 1.6396881177245772, + "grad_norm": 1.893615484237671, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7158737182617188, + "num_tokens": 382793953.0, + "step": 14931 + }, + { + "epoch": 1.639797935427191, + "grad_norm": 2.2466721534729004, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7373566627502441, + "num_tokens": 382817197.0, + "step": 14932 + }, + { + "epoch": 1.6399077531298045, + "grad_norm": 2.0569047927856445, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.706301212310791, + "num_tokens": 382847746.0, + "step": 14933 + }, + { + "epoch": 1.640017570832418, + "grad_norm": 2.3509929180145264, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7085453271865845, + "num_tokens": 382872370.0, + "step": 14934 + }, + { + "epoch": 1.6401273885350318, + "grad_norm": 2.2726659774780273, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7428573369979858, + "num_tokens": 382896744.0, + "step": 14935 + }, + { + "epoch": 1.6402372062376456, + "grad_norm": 2.3860270977020264, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7222476601600647, + "num_tokens": 382919969.0, + "step": 14936 + }, + { + "epoch": 1.6403470239402593, + "grad_norm": 2.2801003456115723, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7383023500442505, + "num_tokens": 382944317.0, + "step": 14937 + }, + { + "epoch": 1.6404568416428729, + "grad_norm": 2.2029271125793457, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7154929637908936, + "num_tokens": 382970281.0, + "step": 14938 + }, + { + "epoch": 1.6405666593454864, + "grad_norm": 2.1092188358306885, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.713371992111206, + "num_tokens": 383000892.0, + "step": 14939 + }, + { + "epoch": 1.6406764770481002, + "grad_norm": 2.174170970916748, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7143571972846985, + "num_tokens": 383027387.0, + "step": 14940 + }, + { + "epoch": 1.640786294750714, + "grad_norm": 2.3282313346862793, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.715950608253479, + "num_tokens": 383051972.0, + "step": 14941 + }, + { + "epoch": 1.6408961124533274, + "grad_norm": 2.348041534423828, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7214108109474182, + "num_tokens": 383075249.0, + "step": 14942 + }, + { + "epoch": 1.641005930155941, + "grad_norm": 2.196962594985962, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7282936573028564, + "num_tokens": 383101587.0, + "step": 14943 + }, + { + "epoch": 1.6411157478585547, + "grad_norm": 2.272643804550171, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7104291915893555, + "num_tokens": 383126605.0, + "step": 14944 + }, + { + "epoch": 1.6412255655611685, + "grad_norm": 2.3080832958221436, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7143330574035645, + "num_tokens": 383152460.0, + "step": 14945 + }, + { + "epoch": 1.6413353832637823, + "grad_norm": 2.556635618209839, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7229855060577393, + "num_tokens": 383173916.0, + "step": 14946 + }, + { + "epoch": 1.6414452009663958, + "grad_norm": 2.2401793003082275, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7257311344146729, + "num_tokens": 383198645.0, + "step": 14947 + }, + { + "epoch": 1.6415550186690093, + "grad_norm": 2.022538900375366, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7222302556037903, + "num_tokens": 383228493.0, + "step": 14948 + }, + { + "epoch": 1.641664836371623, + "grad_norm": 2.263213872909546, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7159661054611206, + "num_tokens": 383255194.0, + "step": 14949 + }, + { + "epoch": 1.6417746540742368, + "grad_norm": 2.3400630950927734, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7098220586776733, + "num_tokens": 383279242.0, + "step": 14950 + }, + { + "epoch": 1.6418844717768504, + "grad_norm": 2.418097734451294, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7461251020431519, + "num_tokens": 383300314.0, + "step": 14951 + }, + { + "epoch": 1.6419942894794641, + "grad_norm": 1.982475996017456, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7038524150848389, + "num_tokens": 383333126.0, + "step": 14952 + }, + { + "epoch": 1.6421041071820777, + "grad_norm": 2.2705142498016357, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7136655449867249, + "num_tokens": 383357658.0, + "step": 14953 + }, + { + "epoch": 1.6422139248846914, + "grad_norm": 2.679009199142456, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7273337244987488, + "num_tokens": 383377352.0, + "step": 14954 + }, + { + "epoch": 1.6423237425873052, + "grad_norm": 2.1705172061920166, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6937218904495239, + "num_tokens": 383406073.0, + "step": 14955 + }, + { + "epoch": 1.6424335602899187, + "grad_norm": 2.2140426635742188, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7227174043655396, + "num_tokens": 383432674.0, + "step": 14956 + }, + { + "epoch": 1.6425433779925322, + "grad_norm": 2.496866464614868, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7322385311126709, + "num_tokens": 383453791.0, + "step": 14957 + }, + { + "epoch": 1.642653195695146, + "grad_norm": 2.5416882038116455, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7376401424407959, + "num_tokens": 383473350.0, + "step": 14958 + }, + { + "epoch": 1.6427630133977598, + "grad_norm": 2.2340028285980225, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7318881750106812, + "num_tokens": 383500062.0, + "step": 14959 + }, + { + "epoch": 1.6428728311003735, + "grad_norm": 2.2476940155029297, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7117093801498413, + "num_tokens": 383528644.0, + "step": 14960 + }, + { + "epoch": 1.642982648802987, + "grad_norm": 2.541233539581299, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7267931699752808, + "num_tokens": 383548685.0, + "step": 14961 + }, + { + "epoch": 1.6430924665056006, + "grad_norm": 2.2239270210266113, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7178593873977661, + "num_tokens": 383573071.0, + "step": 14962 + }, + { + "epoch": 1.6432022842082143, + "grad_norm": 2.251727342605591, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.716092586517334, + "num_tokens": 383599068.0, + "step": 14963 + }, + { + "epoch": 1.643312101910828, + "grad_norm": 2.065520763397217, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6860976219177246, + "num_tokens": 383630616.0, + "step": 14964 + }, + { + "epoch": 1.6434219196134416, + "grad_norm": 2.3108062744140625, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7016966342926025, + "num_tokens": 383656231.0, + "step": 14965 + }, + { + "epoch": 1.6435317373160554, + "grad_norm": 2.17991304397583, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7098200917243958, + "num_tokens": 383682682.0, + "step": 14966 + }, + { + "epoch": 1.643641555018669, + "grad_norm": 2.23921537399292, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7417082786560059, + "num_tokens": 383706484.0, + "step": 14967 + }, + { + "epoch": 1.6437513727212827, + "grad_norm": 2.534139394760132, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7352109551429749, + "num_tokens": 383726726.0, + "step": 14968 + }, + { + "epoch": 1.6438611904238964, + "grad_norm": 2.0535471439361572, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7202840447425842, + "num_tokens": 383759342.0, + "step": 14969 + }, + { + "epoch": 1.64397100812651, + "grad_norm": 2.3067288398742676, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.727812647819519, + "num_tokens": 383782312.0, + "step": 14970 + }, + { + "epoch": 1.6440808258291235, + "grad_norm": 2.1903302669525146, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7046369314193726, + "num_tokens": 383811494.0, + "step": 14971 + }, + { + "epoch": 1.6441906435317373, + "grad_norm": 2.250173807144165, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7269369959831238, + "num_tokens": 383837259.0, + "step": 14972 + }, + { + "epoch": 1.644300461234351, + "grad_norm": 2.4731247425079346, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7424789667129517, + "num_tokens": 383858275.0, + "step": 14973 + }, + { + "epoch": 1.6444102789369648, + "grad_norm": 2.1783127784729004, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7204700708389282, + "num_tokens": 383885020.0, + "step": 14974 + }, + { + "epoch": 1.6445200966395783, + "grad_norm": 2.266911745071411, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7415890097618103, + "num_tokens": 383910757.0, + "step": 14975 + }, + { + "epoch": 1.6446299143421919, + "grad_norm": 2.633462905883789, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7217903137207031, + "num_tokens": 383930526.0, + "step": 14976 + }, + { + "epoch": 1.6447397320448056, + "grad_norm": 2.2671186923980713, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7050947546958923, + "num_tokens": 383956023.0, + "step": 14977 + }, + { + "epoch": 1.6448495497474194, + "grad_norm": 2.1931941509246826, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7131433486938477, + "num_tokens": 383982221.0, + "step": 14978 + }, + { + "epoch": 1.644959367450033, + "grad_norm": 2.0118026733398438, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7080298066139221, + "num_tokens": 384014680.0, + "step": 14979 + }, + { + "epoch": 1.6450691851526464, + "grad_norm": 2.23504900932312, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7321054935455322, + "num_tokens": 384040867.0, + "step": 14980 + }, + { + "epoch": 1.6451790028552602, + "grad_norm": 2.435696601867676, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7170774340629578, + "num_tokens": 384064582.0, + "step": 14981 + }, + { + "epoch": 1.645288820557874, + "grad_norm": 2.262561559677124, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7220745086669922, + "num_tokens": 384091118.0, + "step": 14982 + }, + { + "epoch": 1.6453986382604877, + "grad_norm": 2.268991708755493, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7214034199714661, + "num_tokens": 384116737.0, + "step": 14983 + }, + { + "epoch": 1.6455084559631012, + "grad_norm": 2.1731951236724854, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6935241222381592, + "num_tokens": 384146609.0, + "step": 14984 + }, + { + "epoch": 1.6456182736657148, + "grad_norm": 2.1443920135498047, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7164560556411743, + "num_tokens": 384176172.0, + "step": 14985 + }, + { + "epoch": 1.6457280913683285, + "grad_norm": 1.941300868988037, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7020769715309143, + "num_tokens": 384208838.0, + "step": 14986 + }, + { + "epoch": 1.6458379090709423, + "grad_norm": 2.2222018241882324, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6891183853149414, + "num_tokens": 384236022.0, + "step": 14987 + }, + { + "epoch": 1.645947726773556, + "grad_norm": 2.2330431938171387, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7218832969665527, + "num_tokens": 384261479.0, + "step": 14988 + }, + { + "epoch": 1.6460575444761696, + "grad_norm": 2.5086758136749268, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7253282070159912, + "num_tokens": 384283865.0, + "step": 14989 + }, + { + "epoch": 1.6461673621787831, + "grad_norm": 2.43264102935791, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6995603442192078, + "num_tokens": 384308157.0, + "step": 14990 + }, + { + "epoch": 1.6462771798813969, + "grad_norm": 2.2150957584381104, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7494651079177856, + "num_tokens": 384332290.0, + "step": 14991 + }, + { + "epoch": 1.6463869975840106, + "grad_norm": 2.1963603496551514, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7129654884338379, + "num_tokens": 384361341.0, + "step": 14992 + }, + { + "epoch": 1.6464968152866242, + "grad_norm": 2.4258556365966797, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7428805828094482, + "num_tokens": 384382462.0, + "step": 14993 + }, + { + "epoch": 1.6466066329892377, + "grad_norm": 2.3447110652923584, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7207565307617188, + "num_tokens": 384407674.0, + "step": 14994 + }, + { + "epoch": 1.6467164506918515, + "grad_norm": 2.019141674041748, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7020372152328491, + "num_tokens": 384437657.0, + "step": 14995 + }, + { + "epoch": 1.6468262683944652, + "grad_norm": 2.434096336364746, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7174161672592163, + "num_tokens": 384459439.0, + "step": 14996 + }, + { + "epoch": 1.646936086097079, + "grad_norm": 2.219093084335327, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7041052579879761, + "num_tokens": 384487094.0, + "step": 14997 + }, + { + "epoch": 1.6470459037996925, + "grad_norm": 2.1705570220947266, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6981590986251831, + "num_tokens": 384514120.0, + "step": 14998 + }, + { + "epoch": 1.647155721502306, + "grad_norm": 2.2959437370300293, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7287263870239258, + "num_tokens": 384539985.0, + "step": 14999 + }, + { + "epoch": 1.6472655392049198, + "grad_norm": 2.483405828475952, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7044672966003418, + "num_tokens": 384563187.0, + "step": 15000 + }, + { + "epoch": 1.6473753569075336, + "grad_norm": 2.153034210205078, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7624129056930542, + "num_tokens": 384586216.0, + "step": 15001 + }, + { + "epoch": 1.6474851746101473, + "grad_norm": 2.1304309368133545, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6914018392562866, + "num_tokens": 384613432.0, + "step": 15002 + }, + { + "epoch": 1.6475949923127609, + "grad_norm": 2.2752089500427246, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7101830244064331, + "num_tokens": 384638434.0, + "step": 15003 + }, + { + "epoch": 1.6477048100153744, + "grad_norm": 2.5217647552490234, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7129045128822327, + "num_tokens": 384661151.0, + "step": 15004 + }, + { + "epoch": 1.6478146277179881, + "grad_norm": 2.225513458251953, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.716703474521637, + "num_tokens": 384687693.0, + "step": 15005 + }, + { + "epoch": 1.647924445420602, + "grad_norm": 2.3084871768951416, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7305498719215393, + "num_tokens": 384711781.0, + "step": 15006 + }, + { + "epoch": 1.6480342631232154, + "grad_norm": 2.2729671001434326, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7293225526809692, + "num_tokens": 384735890.0, + "step": 15007 + }, + { + "epoch": 1.648144080825829, + "grad_norm": 2.4072835445404053, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7209721207618713, + "num_tokens": 384759599.0, + "step": 15008 + }, + { + "epoch": 1.6482538985284427, + "grad_norm": 2.128877639770508, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7217861413955688, + "num_tokens": 384788563.0, + "step": 15009 + }, + { + "epoch": 1.6483637162310565, + "grad_norm": 2.115262269973755, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6932144165039062, + "num_tokens": 384818941.0, + "step": 15010 + }, + { + "epoch": 1.6484735339336702, + "grad_norm": 2.4674158096313477, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7079004049301147, + "num_tokens": 384841578.0, + "step": 15011 + }, + { + "epoch": 1.6485833516362838, + "grad_norm": 2.017277956008911, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6920437812805176, + "num_tokens": 384874641.0, + "step": 15012 + }, + { + "epoch": 1.6486931693388973, + "grad_norm": 2.0307540893554688, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.723595380783081, + "num_tokens": 384905184.0, + "step": 15013 + }, + { + "epoch": 1.648802987041511, + "grad_norm": 2.2451765537261963, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7103028297424316, + "num_tokens": 384931642.0, + "step": 15014 + }, + { + "epoch": 1.6489128047441248, + "grad_norm": 2.2405717372894287, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.735154390335083, + "num_tokens": 384957791.0, + "step": 15015 + }, + { + "epoch": 1.6490226224467384, + "grad_norm": 2.334088087081909, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7290992736816406, + "num_tokens": 384982521.0, + "step": 15016 + }, + { + "epoch": 1.6491324401493521, + "grad_norm": 2.2204957008361816, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7170150876045227, + "num_tokens": 385008817.0, + "step": 15017 + }, + { + "epoch": 1.6492422578519657, + "grad_norm": 2.2184765338897705, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.754496693611145, + "num_tokens": 385033325.0, + "step": 15018 + }, + { + "epoch": 1.6493520755545794, + "grad_norm": 1.9983152151107788, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7145720720291138, + "num_tokens": 385063148.0, + "step": 15019 + }, + { + "epoch": 1.6494618932571932, + "grad_norm": 2.151632785797119, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7062076330184937, + "num_tokens": 385092298.0, + "step": 15020 + }, + { + "epoch": 1.6495717109598067, + "grad_norm": 2.1139986515045166, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7194089293479919, + "num_tokens": 385122749.0, + "step": 15021 + }, + { + "epoch": 1.6496815286624202, + "grad_norm": 1.8247817754745483, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7011709213256836, + "num_tokens": 385159899.0, + "step": 15022 + }, + { + "epoch": 1.649791346365034, + "grad_norm": 2.5206282138824463, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7179821729660034, + "num_tokens": 385183288.0, + "step": 15023 + }, + { + "epoch": 1.6499011640676478, + "grad_norm": 2.1742522716522217, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7343951463699341, + "num_tokens": 385209437.0, + "step": 15024 + }, + { + "epoch": 1.6500109817702615, + "grad_norm": 2.2181544303894043, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7096695303916931, + "num_tokens": 385236058.0, + "step": 15025 + }, + { + "epoch": 1.650120799472875, + "grad_norm": 2.1335561275482178, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7197502255439758, + "num_tokens": 385265993.0, + "step": 15026 + }, + { + "epoch": 1.6502306171754886, + "grad_norm": 2.223538875579834, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7023544907569885, + "num_tokens": 385292184.0, + "step": 15027 + }, + { + "epoch": 1.6503404348781023, + "grad_norm": 2.0288190841674805, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7072328925132751, + "num_tokens": 385324423.0, + "step": 15028 + }, + { + "epoch": 1.650450252580716, + "grad_norm": 2.3314826488494873, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.701547384262085, + "num_tokens": 385350154.0, + "step": 15029 + }, + { + "epoch": 1.6505600702833296, + "grad_norm": 2.278022527694702, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7519078850746155, + "num_tokens": 385374725.0, + "step": 15030 + }, + { + "epoch": 1.6506698879859434, + "grad_norm": 2.199538469314575, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7141456604003906, + "num_tokens": 385400779.0, + "step": 15031 + }, + { + "epoch": 1.650779705688557, + "grad_norm": 2.2061374187469482, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7175503969192505, + "num_tokens": 385427993.0, + "step": 15032 + }, + { + "epoch": 1.6508895233911707, + "grad_norm": 2.142148733139038, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7209545969963074, + "num_tokens": 385453492.0, + "step": 15033 + }, + { + "epoch": 1.6509993410937844, + "grad_norm": 2.349769353866577, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7275257110595703, + "num_tokens": 385481167.0, + "step": 15034 + }, + { + "epoch": 1.651109158796398, + "grad_norm": 2.2478268146514893, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.732123851776123, + "num_tokens": 385506177.0, + "step": 15035 + }, + { + "epoch": 1.6512189764990115, + "grad_norm": 2.2813615798950195, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7233537435531616, + "num_tokens": 385531495.0, + "step": 15036 + }, + { + "epoch": 1.6513287942016253, + "grad_norm": 2.2900636196136475, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7377641797065735, + "num_tokens": 385554266.0, + "step": 15037 + }, + { + "epoch": 1.651438611904239, + "grad_norm": 2.34525728225708, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7213070392608643, + "num_tokens": 385579005.0, + "step": 15038 + }, + { + "epoch": 1.6515484296068528, + "grad_norm": 2.2494800090789795, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6985752582550049, + "num_tokens": 385605420.0, + "step": 15039 + }, + { + "epoch": 1.6516582473094663, + "grad_norm": 2.252776622772217, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7216991186141968, + "num_tokens": 385631117.0, + "step": 15040 + }, + { + "epoch": 1.6517680650120798, + "grad_norm": 2.0816590785980225, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7170913219451904, + "num_tokens": 385660741.0, + "step": 15041 + }, + { + "epoch": 1.6518778827146936, + "grad_norm": 2.0724480152130127, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7165333032608032, + "num_tokens": 385689699.0, + "step": 15042 + }, + { + "epoch": 1.6519877004173074, + "grad_norm": 1.920518398284912, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6935037970542908, + "num_tokens": 385721347.0, + "step": 15043 + }, + { + "epoch": 1.652097518119921, + "grad_norm": 2.0729377269744873, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7124451398849487, + "num_tokens": 385751670.0, + "step": 15044 + }, + { + "epoch": 1.6522073358225344, + "grad_norm": 2.432699203491211, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7111306190490723, + "num_tokens": 385773741.0, + "step": 15045 + }, + { + "epoch": 1.6523171535251482, + "grad_norm": 2.4977269172668457, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7175438404083252, + "num_tokens": 385795870.0, + "step": 15046 + }, + { + "epoch": 1.652426971227762, + "grad_norm": 2.351428985595703, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7175701260566711, + "num_tokens": 385819125.0, + "step": 15047 + }, + { + "epoch": 1.6525367889303757, + "grad_norm": 2.1250288486480713, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7223861217498779, + "num_tokens": 385847431.0, + "step": 15048 + }, + { + "epoch": 1.6526466066329892, + "grad_norm": 2.105867862701416, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7034050226211548, + "num_tokens": 385876246.0, + "step": 15049 + }, + { + "epoch": 1.6527564243356028, + "grad_norm": 2.2500131130218506, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7265755534172058, + "num_tokens": 385901738.0, + "step": 15050 + }, + { + "epoch": 1.6528662420382165, + "grad_norm": 2.460254192352295, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7163957357406616, + "num_tokens": 385923434.0, + "step": 15051 + }, + { + "epoch": 1.6529760597408303, + "grad_norm": 2.3799877166748047, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.730457067489624, + "num_tokens": 385944610.0, + "step": 15052 + }, + { + "epoch": 1.653085877443444, + "grad_norm": 2.2619364261627197, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7214575409889221, + "num_tokens": 385970313.0, + "step": 15053 + }, + { + "epoch": 1.6531956951460576, + "grad_norm": 2.398555040359497, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7089844346046448, + "num_tokens": 385992740.0, + "step": 15054 + }, + { + "epoch": 1.653305512848671, + "grad_norm": 2.6560492515563965, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7248737812042236, + "num_tokens": 386014010.0, + "step": 15055 + }, + { + "epoch": 1.6534153305512849, + "grad_norm": 2.3106024265289307, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.6997447609901428, + "num_tokens": 386038628.0, + "step": 15056 + }, + { + "epoch": 1.6535251482538986, + "grad_norm": 2.105633020401001, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7532438039779663, + "num_tokens": 386064142.0, + "step": 15057 + }, + { + "epoch": 1.6536349659565122, + "grad_norm": 2.3675055503845215, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7254431247711182, + "num_tokens": 386087662.0, + "step": 15058 + }, + { + "epoch": 1.6537447836591257, + "grad_norm": 2.1330249309539795, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7185890078544617, + "num_tokens": 386113281.0, + "step": 15059 + }, + { + "epoch": 1.6538546013617395, + "grad_norm": 2.426093101501465, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.730320930480957, + "num_tokens": 386134392.0, + "step": 15060 + }, + { + "epoch": 1.6539644190643532, + "grad_norm": 2.255970001220703, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7510490417480469, + "num_tokens": 386156351.0, + "step": 15061 + }, + { + "epoch": 1.654074236766967, + "grad_norm": 2.40549898147583, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7081111669540405, + "num_tokens": 386180139.0, + "step": 15062 + }, + { + "epoch": 1.6541840544695805, + "grad_norm": 2.0717177391052246, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7016324996948242, + "num_tokens": 386209566.0, + "step": 15063 + }, + { + "epoch": 1.654293872172194, + "grad_norm": 2.123070478439331, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7166460752487183, + "num_tokens": 386237967.0, + "step": 15064 + }, + { + "epoch": 1.6544036898748078, + "grad_norm": 2.3199403285980225, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7279003858566284, + "num_tokens": 386263775.0, + "step": 15065 + }, + { + "epoch": 1.6545135075774215, + "grad_norm": 2.3130571842193604, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7258578538894653, + "num_tokens": 386288744.0, + "step": 15066 + }, + { + "epoch": 1.6546233252800353, + "grad_norm": 2.33894681930542, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7269810438156128, + "num_tokens": 386314305.0, + "step": 15067 + }, + { + "epoch": 1.6547331429826488, + "grad_norm": 2.4708950519561768, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7372258305549622, + "num_tokens": 386337687.0, + "step": 15068 + }, + { + "epoch": 1.6548429606852624, + "grad_norm": 2.283151865005493, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7167162895202637, + "num_tokens": 386361561.0, + "step": 15069 + }, + { + "epoch": 1.6549527783878761, + "grad_norm": 2.2026572227478027, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7395171523094177, + "num_tokens": 386388255.0, + "step": 15070 + }, + { + "epoch": 1.65506259609049, + "grad_norm": 1.9248749017715454, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7229697704315186, + "num_tokens": 386420415.0, + "step": 15071 + }, + { + "epoch": 1.6551724137931034, + "grad_norm": 2.182234764099121, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7211399674415588, + "num_tokens": 386447018.0, + "step": 15072 + }, + { + "epoch": 1.655282231495717, + "grad_norm": 2.536656618118286, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7209577560424805, + "num_tokens": 386467186.0, + "step": 15073 + }, + { + "epoch": 1.6553920491983307, + "grad_norm": 2.7552876472473145, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7240031957626343, + "num_tokens": 386485038.0, + "step": 15074 + }, + { + "epoch": 1.6555018669009445, + "grad_norm": 1.9952033758163452, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7030711770057678, + "num_tokens": 386515301.0, + "step": 15075 + }, + { + "epoch": 1.6556116846035582, + "grad_norm": 2.0225141048431396, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7242335081100464, + "num_tokens": 386545936.0, + "step": 15076 + }, + { + "epoch": 1.6557215023061718, + "grad_norm": 2.4159536361694336, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7278105616569519, + "num_tokens": 386569857.0, + "step": 15077 + }, + { + "epoch": 1.6558313200087853, + "grad_norm": 2.4667587280273438, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7180531024932861, + "num_tokens": 386592091.0, + "step": 15078 + }, + { + "epoch": 1.655941137711399, + "grad_norm": 2.7439465522766113, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7196058630943298, + "num_tokens": 386614586.0, + "step": 15079 + }, + { + "epoch": 1.6560509554140128, + "grad_norm": 2.2051961421966553, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7272989749908447, + "num_tokens": 386643887.0, + "step": 15080 + }, + { + "epoch": 1.6561607731166264, + "grad_norm": 2.3809804916381836, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7061121463775635, + "num_tokens": 386666980.0, + "step": 15081 + }, + { + "epoch": 1.65627059081924, + "grad_norm": 2.0689828395843506, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.703705906867981, + "num_tokens": 386697395.0, + "step": 15082 + }, + { + "epoch": 1.6563804085218536, + "grad_norm": 2.1223387718200684, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7171828746795654, + "num_tokens": 386724165.0, + "step": 15083 + }, + { + "epoch": 1.6564902262244674, + "grad_norm": 2.4947268962860107, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7360049486160278, + "num_tokens": 386744839.0, + "step": 15084 + }, + { + "epoch": 1.6566000439270812, + "grad_norm": 2.2522470951080322, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7320435643196106, + "num_tokens": 386771753.0, + "step": 15085 + }, + { + "epoch": 1.6567098616296947, + "grad_norm": 1.9975990056991577, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7208805084228516, + "num_tokens": 386800721.0, + "step": 15086 + }, + { + "epoch": 1.6568196793323082, + "grad_norm": 1.99830961227417, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7233173847198486, + "num_tokens": 386831272.0, + "step": 15087 + }, + { + "epoch": 1.656929497034922, + "grad_norm": 2.4306304454803467, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7438290119171143, + "num_tokens": 386852499.0, + "step": 15088 + }, + { + "epoch": 1.6570393147375357, + "grad_norm": 2.293504476547241, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7468304634094238, + "num_tokens": 386877502.0, + "step": 15089 + }, + { + "epoch": 1.6571491324401495, + "grad_norm": 2.1954236030578613, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7219028472900391, + "num_tokens": 386904176.0, + "step": 15090 + }, + { + "epoch": 1.657258950142763, + "grad_norm": 1.986755132675171, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7309958934783936, + "num_tokens": 386933614.0, + "step": 15091 + }, + { + "epoch": 1.6573687678453766, + "grad_norm": 2.183546543121338, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7050459384918213, + "num_tokens": 386961373.0, + "step": 15092 + }, + { + "epoch": 1.6574785855479903, + "grad_norm": 2.4344472885131836, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7401928901672363, + "num_tokens": 386982765.0, + "step": 15093 + }, + { + "epoch": 1.657588403250604, + "grad_norm": 2.2448618412017822, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7371789813041687, + "num_tokens": 387007691.0, + "step": 15094 + }, + { + "epoch": 1.6576982209532176, + "grad_norm": 2.5676748752593994, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7291480302810669, + "num_tokens": 387029439.0, + "step": 15095 + }, + { + "epoch": 1.6578080386558314, + "grad_norm": 2.2103233337402344, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7121708393096924, + "num_tokens": 387058372.0, + "step": 15096 + }, + { + "epoch": 1.657917856358445, + "grad_norm": 2.2783374786376953, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7285099029541016, + "num_tokens": 387083446.0, + "step": 15097 + }, + { + "epoch": 1.6580276740610587, + "grad_norm": 2.059196949005127, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7141346335411072, + "num_tokens": 387113187.0, + "step": 15098 + }, + { + "epoch": 1.6581374917636724, + "grad_norm": 2.113748788833618, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7268126010894775, + "num_tokens": 387141140.0, + "step": 15099 + }, + { + "epoch": 1.658247309466286, + "grad_norm": 2.1418709754943848, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7388761043548584, + "num_tokens": 387166223.0, + "step": 15100 + }, + { + "epoch": 1.6583571271688995, + "grad_norm": 2.7238993644714355, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7406774759292603, + "num_tokens": 387184770.0, + "step": 15101 + }, + { + "epoch": 1.6584669448715132, + "grad_norm": 2.209625720977783, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7303920984268188, + "num_tokens": 387208867.0, + "step": 15102 + }, + { + "epoch": 1.658576762574127, + "grad_norm": 2.180418014526367, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7188397645950317, + "num_tokens": 387234479.0, + "step": 15103 + }, + { + "epoch": 1.6586865802767408, + "grad_norm": 2.5253448486328125, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7520079612731934, + "num_tokens": 387253660.0, + "step": 15104 + }, + { + "epoch": 1.6587963979793543, + "grad_norm": 2.299471139907837, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7175761461257935, + "num_tokens": 387279839.0, + "step": 15105 + }, + { + "epoch": 1.6589062156819678, + "grad_norm": 2.2776377201080322, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7018718123435974, + "num_tokens": 387304515.0, + "step": 15106 + }, + { + "epoch": 1.6590160333845816, + "grad_norm": 2.0823380947113037, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.6974683403968811, + "num_tokens": 387334226.0, + "step": 15107 + }, + { + "epoch": 1.6591258510871953, + "grad_norm": 2.173548460006714, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7251219153404236, + "num_tokens": 387360031.0, + "step": 15108 + }, + { + "epoch": 1.6592356687898089, + "grad_norm": 2.2294669151306152, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7587002515792847, + "num_tokens": 387382412.0, + "step": 15109 + }, + { + "epoch": 1.6593454864924224, + "grad_norm": 2.356513261795044, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7208560705184937, + "num_tokens": 387404212.0, + "step": 15110 + }, + { + "epoch": 1.6594553041950362, + "grad_norm": 2.1151411533355713, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7023385763168335, + "num_tokens": 387431050.0, + "step": 15111 + }, + { + "epoch": 1.65956512189765, + "grad_norm": 1.9785377979278564, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6971755027770996, + "num_tokens": 387462384.0, + "step": 15112 + }, + { + "epoch": 1.6596749396002637, + "grad_norm": 2.3156707286834717, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7397294640541077, + "num_tokens": 387483164.0, + "step": 15113 + }, + { + "epoch": 1.6597847573028772, + "grad_norm": 2.2900314331054688, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7225605845451355, + "num_tokens": 387507721.0, + "step": 15114 + }, + { + "epoch": 1.6598945750054908, + "grad_norm": 2.2021865844726562, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7005254626274109, + "num_tokens": 387533945.0, + "step": 15115 + }, + { + "epoch": 1.6600043927081045, + "grad_norm": 2.2137563228607178, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7150065302848816, + "num_tokens": 387561369.0, + "step": 15116 + }, + { + "epoch": 1.6601142104107183, + "grad_norm": 2.2411625385284424, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.722825825214386, + "num_tokens": 387588222.0, + "step": 15117 + }, + { + "epoch": 1.660224028113332, + "grad_norm": 1.9809579849243164, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7114330530166626, + "num_tokens": 387618771.0, + "step": 15118 + }, + { + "epoch": 1.6603338458159456, + "grad_norm": 2.29152512550354, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7158961892127991, + "num_tokens": 387645535.0, + "step": 15119 + }, + { + "epoch": 1.660443663518559, + "grad_norm": 2.5743138790130615, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7325003147125244, + "num_tokens": 387666251.0, + "step": 15120 + }, + { + "epoch": 1.6605534812211729, + "grad_norm": 2.0880684852600098, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7373175621032715, + "num_tokens": 387691701.0, + "step": 15121 + }, + { + "epoch": 1.6606632989237866, + "grad_norm": 2.1205825805664062, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7509986162185669, + "num_tokens": 387716032.0, + "step": 15122 + }, + { + "epoch": 1.6607731166264001, + "grad_norm": 2.0123257637023926, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7152222990989685, + "num_tokens": 387746839.0, + "step": 15123 + }, + { + "epoch": 1.6608829343290137, + "grad_norm": 2.544919967651367, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7226530313491821, + "num_tokens": 387769232.0, + "step": 15124 + }, + { + "epoch": 1.6609927520316274, + "grad_norm": 2.304946184158325, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.721981406211853, + "num_tokens": 387794653.0, + "step": 15125 + }, + { + "epoch": 1.6611025697342412, + "grad_norm": 2.084179401397705, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7137128114700317, + "num_tokens": 387823354.0, + "step": 15126 + }, + { + "epoch": 1.661212387436855, + "grad_norm": 2.3793370723724365, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7142187356948853, + "num_tokens": 387845757.0, + "step": 15127 + }, + { + "epoch": 1.6613222051394685, + "grad_norm": 2.0354137420654297, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7124916315078735, + "num_tokens": 387874941.0, + "step": 15128 + }, + { + "epoch": 1.661432022842082, + "grad_norm": 2.319997549057007, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7303425669670105, + "num_tokens": 387899875.0, + "step": 15129 + }, + { + "epoch": 1.6615418405446958, + "grad_norm": 2.1128883361816406, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7333940267562866, + "num_tokens": 387928673.0, + "step": 15130 + }, + { + "epoch": 1.6616516582473095, + "grad_norm": 2.1036739349365234, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.703311562538147, + "num_tokens": 387959165.0, + "step": 15131 + }, + { + "epoch": 1.661761475949923, + "grad_norm": 2.0728888511657715, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7148357629776001, + "num_tokens": 387986274.0, + "step": 15132 + }, + { + "epoch": 1.6618712936525368, + "grad_norm": 2.1803247928619385, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7263954877853394, + "num_tokens": 388012768.0, + "step": 15133 + }, + { + "epoch": 1.6619811113551504, + "grad_norm": 2.0592546463012695, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7257119417190552, + "num_tokens": 388041521.0, + "step": 15134 + }, + { + "epoch": 1.6620909290577641, + "grad_norm": 2.258298635482788, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7013353109359741, + "num_tokens": 388067494.0, + "step": 15135 + }, + { + "epoch": 1.6622007467603779, + "grad_norm": 2.1978137493133545, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7084371447563171, + "num_tokens": 388094457.0, + "step": 15136 + }, + { + "epoch": 1.6623105644629914, + "grad_norm": 2.3725974559783936, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7248894572257996, + "num_tokens": 388116799.0, + "step": 15137 + }, + { + "epoch": 1.662420382165605, + "grad_norm": 2.050100564956665, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7247140407562256, + "num_tokens": 388148482.0, + "step": 15138 + }, + { + "epoch": 1.6625301998682187, + "grad_norm": 2.180293321609497, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7187840938568115, + "num_tokens": 388174016.0, + "step": 15139 + }, + { + "epoch": 1.6626400175708325, + "grad_norm": 1.7613729238510132, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6916896104812622, + "num_tokens": 388211655.0, + "step": 15140 + }, + { + "epoch": 1.6627498352734462, + "grad_norm": 2.734846353530884, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7451385259628296, + "num_tokens": 388229574.0, + "step": 15141 + }, + { + "epoch": 1.6628596529760598, + "grad_norm": 2.103623390197754, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7343518733978271, + "num_tokens": 388255346.0, + "step": 15142 + }, + { + "epoch": 1.6629694706786733, + "grad_norm": 2.517986297607422, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7135952711105347, + "num_tokens": 388277707.0, + "step": 15143 + }, + { + "epoch": 1.663079288381287, + "grad_norm": 2.2553696632385254, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7362757921218872, + "num_tokens": 388301190.0, + "step": 15144 + }, + { + "epoch": 1.6631891060839008, + "grad_norm": 2.4329514503479004, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.739647388458252, + "num_tokens": 388325167.0, + "step": 15145 + }, + { + "epoch": 1.6632989237865143, + "grad_norm": 2.387091636657715, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7442598342895508, + "num_tokens": 388346620.0, + "step": 15146 + }, + { + "epoch": 1.663408741489128, + "grad_norm": 2.162451982498169, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7057051658630371, + "num_tokens": 388373629.0, + "step": 15147 + }, + { + "epoch": 1.6635185591917416, + "grad_norm": 2.210726261138916, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7341970205307007, + "num_tokens": 388397290.0, + "step": 15148 + }, + { + "epoch": 1.6636283768943554, + "grad_norm": 2.1895062923431396, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.6961984038352966, + "num_tokens": 388425136.0, + "step": 15149 + }, + { + "epoch": 1.6637381945969691, + "grad_norm": 2.050840377807617, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.732388436794281, + "num_tokens": 388452860.0, + "step": 15150 + }, + { + "epoch": 1.6638480122995827, + "grad_norm": 2.568582057952881, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.721408486366272, + "num_tokens": 388472520.0, + "step": 15151 + }, + { + "epoch": 1.6639578300021962, + "grad_norm": 2.3616158962249756, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7006829977035522, + "num_tokens": 388497396.0, + "step": 15152 + }, + { + "epoch": 1.66406764770481, + "grad_norm": 2.6363134384155273, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7229865193367004, + "num_tokens": 388517646.0, + "step": 15153 + }, + { + "epoch": 1.6641774654074237, + "grad_norm": 2.2273569107055664, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7044733762741089, + "num_tokens": 388544758.0, + "step": 15154 + }, + { + "epoch": 1.6642872831100375, + "grad_norm": 2.047783374786377, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.716894268989563, + "num_tokens": 388574530.0, + "step": 15155 + }, + { + "epoch": 1.664397100812651, + "grad_norm": 2.1734678745269775, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7148334980010986, + "num_tokens": 388601790.0, + "step": 15156 + }, + { + "epoch": 1.6645069185152646, + "grad_norm": 2.075101137161255, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7450913190841675, + "num_tokens": 388630809.0, + "step": 15157 + }, + { + "epoch": 1.6646167362178783, + "grad_norm": 2.3470211029052734, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7089320421218872, + "num_tokens": 388655045.0, + "step": 15158 + }, + { + "epoch": 1.664726553920492, + "grad_norm": 2.368821620941162, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.708430826663971, + "num_tokens": 388681193.0, + "step": 15159 + }, + { + "epoch": 1.6648363716231056, + "grad_norm": 2.0795509815216064, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7374820709228516, + "num_tokens": 388709181.0, + "step": 15160 + }, + { + "epoch": 1.6649461893257194, + "grad_norm": 2.27205753326416, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7363059520721436, + "num_tokens": 388733636.0, + "step": 15161 + }, + { + "epoch": 1.665056007028333, + "grad_norm": 1.9759092330932617, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7070774435997009, + "num_tokens": 388768207.0, + "step": 15162 + }, + { + "epoch": 1.6651658247309467, + "grad_norm": 1.9418866634368896, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.71208256483078, + "num_tokens": 388801834.0, + "step": 15163 + }, + { + "epoch": 1.6652756424335604, + "grad_norm": 2.2440807819366455, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6931705474853516, + "num_tokens": 388827716.0, + "step": 15164 + }, + { + "epoch": 1.665385460136174, + "grad_norm": 2.3557496070861816, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7191840410232544, + "num_tokens": 388850243.0, + "step": 15165 + }, + { + "epoch": 1.6654952778387875, + "grad_norm": 2.24650239944458, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7291268706321716, + "num_tokens": 388876415.0, + "step": 15166 + }, + { + "epoch": 1.6656050955414012, + "grad_norm": 2.379619836807251, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.726885974407196, + "num_tokens": 388897880.0, + "step": 15167 + }, + { + "epoch": 1.665714913244015, + "grad_norm": 2.097498893737793, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.707664966583252, + "num_tokens": 388927612.0, + "step": 15168 + }, + { + "epoch": 1.6658247309466288, + "grad_norm": 2.332261562347412, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7417513728141785, + "num_tokens": 388950551.0, + "step": 15169 + }, + { + "epoch": 1.6659345486492423, + "grad_norm": 2.059232711791992, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7081681489944458, + "num_tokens": 388978590.0, + "step": 15170 + }, + { + "epoch": 1.6660443663518558, + "grad_norm": 2.1680386066436768, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.732239842414856, + "num_tokens": 389006917.0, + "step": 15171 + }, + { + "epoch": 1.6661541840544696, + "grad_norm": 2.1049113273620605, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6963419318199158, + "num_tokens": 389037390.0, + "step": 15172 + }, + { + "epoch": 1.6662640017570833, + "grad_norm": 2.1387569904327393, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7240451574325562, + "num_tokens": 389063867.0, + "step": 15173 + }, + { + "epoch": 1.6663738194596969, + "grad_norm": 2.0468366146087646, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7163617610931396, + "num_tokens": 389093671.0, + "step": 15174 + }, + { + "epoch": 1.6664836371623104, + "grad_norm": 2.1447553634643555, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7273398637771606, + "num_tokens": 389119335.0, + "step": 15175 + }, + { + "epoch": 1.6665934548649242, + "grad_norm": 2.023615837097168, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7335215210914612, + "num_tokens": 389150528.0, + "step": 15176 + }, + { + "epoch": 1.666703272567538, + "grad_norm": 2.3202385902404785, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6962727904319763, + "num_tokens": 389177247.0, + "step": 15177 + }, + { + "epoch": 1.6668130902701517, + "grad_norm": 2.050750970840454, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7292609214782715, + "num_tokens": 389205055.0, + "step": 15178 + }, + { + "epoch": 1.6669229079727652, + "grad_norm": 2.261812925338745, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7229150533676147, + "num_tokens": 389230440.0, + "step": 15179 + }, + { + "epoch": 1.6670327256753787, + "grad_norm": 2.0375020503997803, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6939032077789307, + "num_tokens": 389261610.0, + "step": 15180 + }, + { + "epoch": 1.6671425433779925, + "grad_norm": 2.420212507247925, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7353532314300537, + "num_tokens": 389283902.0, + "step": 15181 + }, + { + "epoch": 1.6672523610806063, + "grad_norm": 2.2561545372009277, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.742851972579956, + "num_tokens": 389309012.0, + "step": 15182 + }, + { + "epoch": 1.66736217878322, + "grad_norm": 2.269226551055908, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7377640604972839, + "num_tokens": 389331894.0, + "step": 15183 + }, + { + "epoch": 1.6674719964858336, + "grad_norm": 2.3421738147735596, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7339314222335815, + "num_tokens": 389354808.0, + "step": 15184 + }, + { + "epoch": 1.667581814188447, + "grad_norm": 2.5561399459838867, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7233948111534119, + "num_tokens": 389375849.0, + "step": 15185 + }, + { + "epoch": 1.6676916318910608, + "grad_norm": 2.1182596683502197, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7058836817741394, + "num_tokens": 389404461.0, + "step": 15186 + }, + { + "epoch": 1.6678014495936746, + "grad_norm": 2.4632484912872314, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7181891202926636, + "num_tokens": 389426670.0, + "step": 15187 + }, + { + "epoch": 1.6679112672962881, + "grad_norm": 2.3600547313690186, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7084988355636597, + "num_tokens": 389450639.0, + "step": 15188 + }, + { + "epoch": 1.6680210849989017, + "grad_norm": 2.1581599712371826, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.707695484161377, + "num_tokens": 389478949.0, + "step": 15189 + }, + { + "epoch": 1.6681309027015154, + "grad_norm": 2.4101803302764893, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7134034633636475, + "num_tokens": 389502334.0, + "step": 15190 + }, + { + "epoch": 1.6682407204041292, + "grad_norm": 2.015475273132324, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7433418035507202, + "num_tokens": 389531178.0, + "step": 15191 + }, + { + "epoch": 1.668350538106743, + "grad_norm": 2.018578052520752, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7273131608963013, + "num_tokens": 389562306.0, + "step": 15192 + }, + { + "epoch": 1.6684603558093565, + "grad_norm": 2.166088819503784, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6996611952781677, + "num_tokens": 389589977.0, + "step": 15193 + }, + { + "epoch": 1.66857017351197, + "grad_norm": 2.481874942779541, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7414013147354126, + "num_tokens": 389610666.0, + "step": 15194 + }, + { + "epoch": 1.6686799912145838, + "grad_norm": 2.556797981262207, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7350055575370789, + "num_tokens": 389629548.0, + "step": 15195 + }, + { + "epoch": 1.6687898089171975, + "grad_norm": 2.206295967102051, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7277258634567261, + "num_tokens": 389655419.0, + "step": 15196 + }, + { + "epoch": 1.668899626619811, + "grad_norm": 2.254096031188965, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7288722395896912, + "num_tokens": 389680066.0, + "step": 15197 + }, + { + "epoch": 1.6690094443224248, + "grad_norm": 2.2863271236419678, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7133399248123169, + "num_tokens": 389705964.0, + "step": 15198 + }, + { + "epoch": 1.6691192620250384, + "grad_norm": 1.884010672569275, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7395725250244141, + "num_tokens": 389738771.0, + "step": 15199 + }, + { + "epoch": 1.669229079727652, + "grad_norm": 2.2389628887176514, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7227208018302917, + "num_tokens": 389764452.0, + "step": 15200 + }, + { + "epoch": 1.6693388974302659, + "grad_norm": 2.208026170730591, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7204601764678955, + "num_tokens": 389791442.0, + "step": 15201 + }, + { + "epoch": 1.6694487151328794, + "grad_norm": 2.158411741256714, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7146938443183899, + "num_tokens": 389818432.0, + "step": 15202 + }, + { + "epoch": 1.669558532835493, + "grad_norm": 2.1876261234283447, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7204273343086243, + "num_tokens": 389844997.0, + "step": 15203 + }, + { + "epoch": 1.6696683505381067, + "grad_norm": 2.2273221015930176, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7453401684761047, + "num_tokens": 389869085.0, + "step": 15204 + }, + { + "epoch": 1.6697781682407205, + "grad_norm": 2.4228038787841797, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.709235668182373, + "num_tokens": 389893481.0, + "step": 15205 + }, + { + "epoch": 1.6698879859433342, + "grad_norm": 2.1363964080810547, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7212262749671936, + "num_tokens": 389921773.0, + "step": 15206 + }, + { + "epoch": 1.6699978036459477, + "grad_norm": 2.065871000289917, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7083322405815125, + "num_tokens": 389950761.0, + "step": 15207 + }, + { + "epoch": 1.6701076213485613, + "grad_norm": 2.1432979106903076, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7143666744232178, + "num_tokens": 389978506.0, + "step": 15208 + }, + { + "epoch": 1.670217439051175, + "grad_norm": 2.464820623397827, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7110485434532166, + "num_tokens": 390000767.0, + "step": 15209 + }, + { + "epoch": 1.6703272567537888, + "grad_norm": 2.538590908050537, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7134148478507996, + "num_tokens": 390021954.0, + "step": 15210 + }, + { + "epoch": 1.6704370744564023, + "grad_norm": 2.4244437217712402, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7139488458633423, + "num_tokens": 390044450.0, + "step": 15211 + }, + { + "epoch": 1.670546892159016, + "grad_norm": 2.3081624507904053, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7396552562713623, + "num_tokens": 390067070.0, + "step": 15212 + }, + { + "epoch": 1.6706567098616296, + "grad_norm": 2.74299693107605, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7381241917610168, + "num_tokens": 390084582.0, + "step": 15213 + }, + { + "epoch": 1.6707665275642434, + "grad_norm": 2.351290464401245, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7303987145423889, + "num_tokens": 390105697.0, + "step": 15214 + }, + { + "epoch": 1.6708763452668571, + "grad_norm": 2.538147211074829, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7152190804481506, + "num_tokens": 390127499.0, + "step": 15215 + }, + { + "epoch": 1.6709861629694707, + "grad_norm": 2.2179105281829834, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7121127843856812, + "num_tokens": 390155297.0, + "step": 15216 + }, + { + "epoch": 1.6710959806720842, + "grad_norm": 2.1759252548217773, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.71647047996521, + "num_tokens": 390181119.0, + "step": 15217 + }, + { + "epoch": 1.671205798374698, + "grad_norm": 2.2564947605133057, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7193648815155029, + "num_tokens": 390206741.0, + "step": 15218 + }, + { + "epoch": 1.6713156160773117, + "grad_norm": 2.1964409351348877, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7362245321273804, + "num_tokens": 390231540.0, + "step": 15219 + }, + { + "epoch": 1.6714254337799255, + "grad_norm": 2.136544942855835, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7125347852706909, + "num_tokens": 390260963.0, + "step": 15220 + }, + { + "epoch": 1.671535251482539, + "grad_norm": 2.083707571029663, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7279267311096191, + "num_tokens": 390288244.0, + "step": 15221 + }, + { + "epoch": 1.6716450691851525, + "grad_norm": 2.2236711978912354, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6967461109161377, + "num_tokens": 390315103.0, + "step": 15222 + }, + { + "epoch": 1.6717548868877663, + "grad_norm": 2.28849458694458, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.7129004001617432, + "num_tokens": 390338978.0, + "step": 15223 + }, + { + "epoch": 1.67186470459038, + "grad_norm": 2.2352592945098877, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7209317684173584, + "num_tokens": 390365908.0, + "step": 15224 + }, + { + "epoch": 1.6719745222929936, + "grad_norm": 2.2080557346343994, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7155723571777344, + "num_tokens": 390391754.0, + "step": 15225 + }, + { + "epoch": 1.6720843399956071, + "grad_norm": 2.3601653575897217, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7108328342437744, + "num_tokens": 390414308.0, + "step": 15226 + }, + { + "epoch": 1.6721941576982209, + "grad_norm": 1.9580048322677612, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7251040935516357, + "num_tokens": 390445074.0, + "step": 15227 + }, + { + "epoch": 1.6723039754008346, + "grad_norm": 2.643791437149048, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7460945248603821, + "num_tokens": 390462773.0, + "step": 15228 + }, + { + "epoch": 1.6724137931034484, + "grad_norm": 2.089982032775879, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7059073448181152, + "num_tokens": 390492458.0, + "step": 15229 + }, + { + "epoch": 1.672523610806062, + "grad_norm": 2.196497917175293, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7225450277328491, + "num_tokens": 390520088.0, + "step": 15230 + }, + { + "epoch": 1.6726334285086755, + "grad_norm": 2.4480414390563965, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7312505841255188, + "num_tokens": 390541392.0, + "step": 15231 + }, + { + "epoch": 1.6727432462112892, + "grad_norm": 2.1492698192596436, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7118754386901855, + "num_tokens": 390568882.0, + "step": 15232 + }, + { + "epoch": 1.672853063913903, + "grad_norm": 2.5793957710266113, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7326227426528931, + "num_tokens": 390589838.0, + "step": 15233 + }, + { + "epoch": 1.6729628816165167, + "grad_norm": 2.7521347999572754, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7273438572883606, + "num_tokens": 390607518.0, + "step": 15234 + }, + { + "epoch": 1.6730726993191303, + "grad_norm": 2.658989667892456, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7309671640396118, + "num_tokens": 390627946.0, + "step": 15235 + }, + { + "epoch": 1.6731825170217438, + "grad_norm": 2.2875118255615234, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7370764017105103, + "num_tokens": 390651831.0, + "step": 15236 + }, + { + "epoch": 1.6732923347243576, + "grad_norm": 2.1884608268737793, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7121717929840088, + "num_tokens": 390678885.0, + "step": 15237 + }, + { + "epoch": 1.6734021524269713, + "grad_norm": 2.5384469032287598, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7363569736480713, + "num_tokens": 390698223.0, + "step": 15238 + }, + { + "epoch": 1.6735119701295849, + "grad_norm": 2.8146188259124756, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7441767454147339, + "num_tokens": 390716087.0, + "step": 15239 + }, + { + "epoch": 1.6736217878321984, + "grad_norm": 2.2773678302764893, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7266730070114136, + "num_tokens": 390739590.0, + "step": 15240 + }, + { + "epoch": 1.6737316055348122, + "grad_norm": 2.3903069496154785, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7342674732208252, + "num_tokens": 390762859.0, + "step": 15241 + }, + { + "epoch": 1.673841423237426, + "grad_norm": 2.2366840839385986, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7254358530044556, + "num_tokens": 390787040.0, + "step": 15242 + }, + { + "epoch": 1.6739512409400397, + "grad_norm": 2.235079050064087, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7197273969650269, + "num_tokens": 390813205.0, + "step": 15243 + }, + { + "epoch": 1.6740610586426532, + "grad_norm": 2.212028741836548, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7486971616744995, + "num_tokens": 390837561.0, + "step": 15244 + }, + { + "epoch": 1.6741708763452667, + "grad_norm": 2.243112564086914, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6975358724594116, + "num_tokens": 390864596.0, + "step": 15245 + }, + { + "epoch": 1.6742806940478805, + "grad_norm": 2.4907047748565674, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7274036407470703, + "num_tokens": 390885522.0, + "step": 15246 + }, + { + "epoch": 1.6743905117504942, + "grad_norm": 2.036132335662842, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7184708118438721, + "num_tokens": 390914327.0, + "step": 15247 + }, + { + "epoch": 1.674500329453108, + "grad_norm": 1.9391247034072876, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7278550267219543, + "num_tokens": 390948232.0, + "step": 15248 + }, + { + "epoch": 1.6746101471557215, + "grad_norm": 2.304116725921631, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7136020660400391, + "num_tokens": 390971562.0, + "step": 15249 + }, + { + "epoch": 1.674719964858335, + "grad_norm": 2.1649670600891113, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7274206876754761, + "num_tokens": 390999032.0, + "step": 15250 + }, + { + "epoch": 1.6748297825609488, + "grad_norm": 1.9734712839126587, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.723015546798706, + "num_tokens": 391030893.0, + "step": 15251 + }, + { + "epoch": 1.6749396002635626, + "grad_norm": 2.1799187660217285, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7114462852478027, + "num_tokens": 391056950.0, + "step": 15252 + }, + { + "epoch": 1.6750494179661761, + "grad_norm": 2.2051196098327637, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.694907009601593, + "num_tokens": 391083979.0, + "step": 15253 + }, + { + "epoch": 1.6751592356687897, + "grad_norm": 2.215806245803833, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7109394073486328, + "num_tokens": 391110538.0, + "step": 15254 + }, + { + "epoch": 1.6752690533714034, + "grad_norm": 2.8003575801849365, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7337296605110168, + "num_tokens": 391129310.0, + "step": 15255 + }, + { + "epoch": 1.6753788710740172, + "grad_norm": 2.2800345420837402, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7267547845840454, + "num_tokens": 391153269.0, + "step": 15256 + }, + { + "epoch": 1.675488688776631, + "grad_norm": 2.3433258533477783, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7108093500137329, + "num_tokens": 391179262.0, + "step": 15257 + }, + { + "epoch": 1.6755985064792445, + "grad_norm": 2.1366167068481445, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7053805589675903, + "num_tokens": 391206818.0, + "step": 15258 + }, + { + "epoch": 1.675708324181858, + "grad_norm": 2.1432743072509766, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7181517481803894, + "num_tokens": 391233956.0, + "step": 15259 + }, + { + "epoch": 1.6758181418844718, + "grad_norm": 2.0078670978546143, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7192466855049133, + "num_tokens": 391263312.0, + "step": 15260 + }, + { + "epoch": 1.6759279595870855, + "grad_norm": 2.325455904006958, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7078019380569458, + "num_tokens": 391288807.0, + "step": 15261 + }, + { + "epoch": 1.676037777289699, + "grad_norm": 2.000765562057495, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7146615386009216, + "num_tokens": 391320132.0, + "step": 15262 + }, + { + "epoch": 1.6761475949923128, + "grad_norm": 2.035033702850342, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7142056226730347, + "num_tokens": 391352491.0, + "step": 15263 + }, + { + "epoch": 1.6762574126949263, + "grad_norm": 2.3385894298553467, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7542033195495605, + "num_tokens": 391375325.0, + "step": 15264 + }, + { + "epoch": 1.67636723039754, + "grad_norm": 2.0287482738494873, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.702590823173523, + "num_tokens": 391404623.0, + "step": 15265 + }, + { + "epoch": 1.6764770481001539, + "grad_norm": 2.2480554580688477, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7365527749061584, + "num_tokens": 391429368.0, + "step": 15266 + }, + { + "epoch": 1.6765868658027674, + "grad_norm": 2.0626304149627686, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7133697867393494, + "num_tokens": 391459322.0, + "step": 15267 + }, + { + "epoch": 1.676696683505381, + "grad_norm": 2.275073766708374, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7319721579551697, + "num_tokens": 391484578.0, + "step": 15268 + }, + { + "epoch": 1.6768065012079947, + "grad_norm": 2.1598165035247803, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7114332914352417, + "num_tokens": 391512041.0, + "step": 15269 + }, + { + "epoch": 1.6769163189106084, + "grad_norm": 2.026149034500122, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7323493957519531, + "num_tokens": 391539917.0, + "step": 15270 + }, + { + "epoch": 1.6770261366132222, + "grad_norm": 2.0779643058776855, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7196081280708313, + "num_tokens": 391569192.0, + "step": 15271 + }, + { + "epoch": 1.6771359543158357, + "grad_norm": 2.0101959705352783, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7104175686836243, + "num_tokens": 391601179.0, + "step": 15272 + }, + { + "epoch": 1.6772457720184493, + "grad_norm": 2.1764895915985107, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7092970609664917, + "num_tokens": 391629476.0, + "step": 15273 + }, + { + "epoch": 1.677355589721063, + "grad_norm": 2.127953052520752, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7044789791107178, + "num_tokens": 391658109.0, + "step": 15274 + }, + { + "epoch": 1.6774654074236768, + "grad_norm": 2.02422833442688, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7078148126602173, + "num_tokens": 391688288.0, + "step": 15275 + }, + { + "epoch": 1.6775752251262903, + "grad_norm": 2.441741466522217, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7172578573226929, + "num_tokens": 391710607.0, + "step": 15276 + }, + { + "epoch": 1.677685042828904, + "grad_norm": 2.154538631439209, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7201225757598877, + "num_tokens": 391739339.0, + "step": 15277 + }, + { + "epoch": 1.6777948605315176, + "grad_norm": 2.223633289337158, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7480231523513794, + "num_tokens": 391764298.0, + "step": 15278 + }, + { + "epoch": 1.6779046782341314, + "grad_norm": 2.4420993328094482, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7186018824577332, + "num_tokens": 391785868.0, + "step": 15279 + }, + { + "epoch": 1.6780144959367451, + "grad_norm": 2.2691080570220947, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7144163846969604, + "num_tokens": 391811109.0, + "step": 15280 + }, + { + "epoch": 1.6781243136393587, + "grad_norm": 2.0537941455841064, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7034546136856079, + "num_tokens": 391841779.0, + "step": 15281 + }, + { + "epoch": 1.6782341313419722, + "grad_norm": 2.2819299697875977, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7208115458488464, + "num_tokens": 391865722.0, + "step": 15282 + }, + { + "epoch": 1.678343949044586, + "grad_norm": 2.2612054347991943, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7296327948570251, + "num_tokens": 391890663.0, + "step": 15283 + }, + { + "epoch": 1.6784537667471997, + "grad_norm": 2.35585618019104, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7152700424194336, + "num_tokens": 391915225.0, + "step": 15284 + }, + { + "epoch": 1.6785635844498135, + "grad_norm": 2.4069392681121826, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7656084299087524, + "num_tokens": 391936581.0, + "step": 15285 + }, + { + "epoch": 1.678673402152427, + "grad_norm": 2.3331375122070312, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7470825910568237, + "num_tokens": 391958823.0, + "step": 15286 + }, + { + "epoch": 1.6787832198550405, + "grad_norm": 2.160088062286377, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7081469297409058, + "num_tokens": 391988396.0, + "step": 15287 + }, + { + "epoch": 1.6788930375576543, + "grad_norm": 2.480977773666382, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7319921255111694, + "num_tokens": 392008570.0, + "step": 15288 + }, + { + "epoch": 1.679002855260268, + "grad_norm": 2.381714344024658, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7031267881393433, + "num_tokens": 392032801.0, + "step": 15289 + }, + { + "epoch": 1.6791126729628816, + "grad_norm": 2.4181907176971436, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7322429418563843, + "num_tokens": 392054934.0, + "step": 15290 + }, + { + "epoch": 1.6792224906654951, + "grad_norm": 2.3417186737060547, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.716606080532074, + "num_tokens": 392078915.0, + "step": 15291 + }, + { + "epoch": 1.6793323083681089, + "grad_norm": 2.031886100769043, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7427423596382141, + "num_tokens": 392107143.0, + "step": 15292 + }, + { + "epoch": 1.6794421260707226, + "grad_norm": 2.3578131198883057, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7117248773574829, + "num_tokens": 392133129.0, + "step": 15293 + }, + { + "epoch": 1.6795519437733364, + "grad_norm": 2.220723867416382, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7142304182052612, + "num_tokens": 392159447.0, + "step": 15294 + }, + { + "epoch": 1.67966176147595, + "grad_norm": 2.2420036792755127, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7141574025154114, + "num_tokens": 392184808.0, + "step": 15295 + }, + { + "epoch": 1.6797715791785635, + "grad_norm": 2.279453754425049, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7102611064910889, + "num_tokens": 392209490.0, + "step": 15296 + }, + { + "epoch": 1.6798813968811772, + "grad_norm": 2.3313589096069336, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7415064573287964, + "num_tokens": 392231219.0, + "step": 15297 + }, + { + "epoch": 1.679991214583791, + "grad_norm": 2.183938980102539, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.6957631707191467, + "num_tokens": 392258371.0, + "step": 15298 + }, + { + "epoch": 1.6801010322864047, + "grad_norm": 2.408454656600952, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7109981775283813, + "num_tokens": 392282149.0, + "step": 15299 + }, + { + "epoch": 1.6802108499890183, + "grad_norm": 2.0330653190612793, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7417780756950378, + "num_tokens": 392311049.0, + "step": 15300 + }, + { + "epoch": 1.6803206676916318, + "grad_norm": 2.3141562938690186, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6982893347740173, + "num_tokens": 392338968.0, + "step": 15301 + }, + { + "epoch": 1.6804304853942456, + "grad_norm": 1.9862865209579468, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6891860365867615, + "num_tokens": 392374778.0, + "step": 15302 + }, + { + "epoch": 1.6805403030968593, + "grad_norm": 2.1743040084838867, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7198421955108643, + "num_tokens": 392402673.0, + "step": 15303 + }, + { + "epoch": 1.6806501207994728, + "grad_norm": 2.2198874950408936, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7328883409500122, + "num_tokens": 392430962.0, + "step": 15304 + }, + { + "epoch": 1.6807599385020864, + "grad_norm": 2.345794439315796, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7138216495513916, + "num_tokens": 392453306.0, + "step": 15305 + }, + { + "epoch": 1.6808697562047001, + "grad_norm": 2.050428867340088, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.70013827085495, + "num_tokens": 392483097.0, + "step": 15306 + }, + { + "epoch": 1.680979573907314, + "grad_norm": 1.9391286373138428, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7107861042022705, + "num_tokens": 392515381.0, + "step": 15307 + }, + { + "epoch": 1.6810893916099277, + "grad_norm": 2.411785840988159, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7266857624053955, + "num_tokens": 392535923.0, + "step": 15308 + }, + { + "epoch": 1.6811992093125412, + "grad_norm": 2.609221935272217, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7388466000556946, + "num_tokens": 392555562.0, + "step": 15309 + }, + { + "epoch": 1.6813090270151547, + "grad_norm": 2.132337808609009, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6971427202224731, + "num_tokens": 392585984.0, + "step": 15310 + }, + { + "epoch": 1.6814188447177685, + "grad_norm": 2.171081304550171, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.701636791229248, + "num_tokens": 392612261.0, + "step": 15311 + }, + { + "epoch": 1.6815286624203822, + "grad_norm": 2.0449023246765137, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7104620337486267, + "num_tokens": 392643353.0, + "step": 15312 + }, + { + "epoch": 1.681638480122996, + "grad_norm": 2.538782835006714, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7515435814857483, + "num_tokens": 392664664.0, + "step": 15313 + }, + { + "epoch": 1.6817482978256095, + "grad_norm": 2.639754295349121, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7282156944274902, + "num_tokens": 392683351.0, + "step": 15314 + }, + { + "epoch": 1.681858115528223, + "grad_norm": 2.5758728981018066, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7343065142631531, + "num_tokens": 392702524.0, + "step": 15315 + }, + { + "epoch": 1.6819679332308368, + "grad_norm": 2.5810649394989014, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7320330739021301, + "num_tokens": 392724855.0, + "step": 15316 + }, + { + "epoch": 1.6820777509334506, + "grad_norm": 2.3674652576446533, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7357038259506226, + "num_tokens": 392748173.0, + "step": 15317 + }, + { + "epoch": 1.6821875686360641, + "grad_norm": 2.0196545124053955, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7137331366539001, + "num_tokens": 392780085.0, + "step": 15318 + }, + { + "epoch": 1.6822973863386776, + "grad_norm": 2.2220447063446045, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6816241145133972, + "num_tokens": 392808244.0, + "step": 15319 + }, + { + "epoch": 1.6824072040412914, + "grad_norm": 2.465446949005127, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7080110311508179, + "num_tokens": 392831546.0, + "step": 15320 + }, + { + "epoch": 1.6825170217439052, + "grad_norm": 2.096979856491089, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7333590984344482, + "num_tokens": 392859667.0, + "step": 15321 + }, + { + "epoch": 1.682626839446519, + "grad_norm": 2.076805591583252, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7145116329193115, + "num_tokens": 392889697.0, + "step": 15322 + }, + { + "epoch": 1.6827366571491325, + "grad_norm": 2.523339033126831, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.73859041929245, + "num_tokens": 392910208.0, + "step": 15323 + }, + { + "epoch": 1.682846474851746, + "grad_norm": 2.1637003421783447, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7101664543151855, + "num_tokens": 392936085.0, + "step": 15324 + }, + { + "epoch": 1.6829562925543597, + "grad_norm": 2.6067984104156494, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7650622129440308, + "num_tokens": 392954873.0, + "step": 15325 + }, + { + "epoch": 1.6830661102569735, + "grad_norm": 2.3824806213378906, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.731146514415741, + "num_tokens": 392977331.0, + "step": 15326 + }, + { + "epoch": 1.683175927959587, + "grad_norm": 2.1694765090942383, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.6943244934082031, + "num_tokens": 393007726.0, + "step": 15327 + }, + { + "epoch": 1.6832857456622008, + "grad_norm": 2.6058475971221924, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7222376465797424, + "num_tokens": 393027406.0, + "step": 15328 + }, + { + "epoch": 1.6833955633648143, + "grad_norm": 2.464480400085449, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7281039953231812, + "num_tokens": 393048062.0, + "step": 15329 + }, + { + "epoch": 1.683505381067428, + "grad_norm": 2.4329886436462402, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.746756374835968, + "num_tokens": 393068657.0, + "step": 15330 + }, + { + "epoch": 1.6836151987700418, + "grad_norm": 2.260885000228882, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7301888465881348, + "num_tokens": 393092273.0, + "step": 15331 + }, + { + "epoch": 1.6837250164726554, + "grad_norm": 2.0673720836639404, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.727324366569519, + "num_tokens": 393120421.0, + "step": 15332 + }, + { + "epoch": 1.683834834175269, + "grad_norm": 2.266085147857666, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7084558010101318, + "num_tokens": 393146165.0, + "step": 15333 + }, + { + "epoch": 1.6839446518778827, + "grad_norm": 2.0747721195220947, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7289491295814514, + "num_tokens": 393172486.0, + "step": 15334 + }, + { + "epoch": 1.6840544695804964, + "grad_norm": 1.7975413799285889, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7199066877365112, + "num_tokens": 393209761.0, + "step": 15335 + }, + { + "epoch": 1.6841642872831102, + "grad_norm": 2.4513323307037354, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.718854546546936, + "num_tokens": 393233861.0, + "step": 15336 + }, + { + "epoch": 1.6842741049857237, + "grad_norm": 2.0624685287475586, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7239693999290466, + "num_tokens": 393262460.0, + "step": 15337 + }, + { + "epoch": 1.6843839226883373, + "grad_norm": 2.0778517723083496, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.729733943939209, + "num_tokens": 393290121.0, + "step": 15338 + }, + { + "epoch": 1.684493740390951, + "grad_norm": 2.291524648666382, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7106890678405762, + "num_tokens": 393318058.0, + "step": 15339 + }, + { + "epoch": 1.6846035580935648, + "grad_norm": 2.205066204071045, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7594228386878967, + "num_tokens": 393342559.0, + "step": 15340 + }, + { + "epoch": 1.6847133757961783, + "grad_norm": 2.3631584644317627, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7501000165939331, + "num_tokens": 393364460.0, + "step": 15341 + }, + { + "epoch": 1.684823193498792, + "grad_norm": 2.431471586227417, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.71393883228302, + "num_tokens": 393388353.0, + "step": 15342 + }, + { + "epoch": 1.6849330112014056, + "grad_norm": 2.1968531608581543, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7316033840179443, + "num_tokens": 393413801.0, + "step": 15343 + }, + { + "epoch": 1.6850428289040194, + "grad_norm": 2.5995101928710938, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7534676790237427, + "num_tokens": 393432148.0, + "step": 15344 + }, + { + "epoch": 1.685152646606633, + "grad_norm": 2.395714282989502, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7551709413528442, + "num_tokens": 393452999.0, + "step": 15345 + }, + { + "epoch": 1.6852624643092466, + "grad_norm": 2.1628406047821045, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7081586718559265, + "num_tokens": 393481622.0, + "step": 15346 + }, + { + "epoch": 1.6853722820118602, + "grad_norm": 2.1729021072387695, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7439481616020203, + "num_tokens": 393507418.0, + "step": 15347 + }, + { + "epoch": 1.685482099714474, + "grad_norm": 2.157414674758911, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7266126871109009, + "num_tokens": 393534784.0, + "step": 15348 + }, + { + "epoch": 1.6855919174170877, + "grad_norm": 2.472050428390503, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7037293910980225, + "num_tokens": 393557884.0, + "step": 15349 + }, + { + "epoch": 1.6857017351197014, + "grad_norm": 2.2249555587768555, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7411326169967651, + "num_tokens": 393583875.0, + "step": 15350 + }, + { + "epoch": 1.685811552822315, + "grad_norm": 2.103818893432617, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7128440737724304, + "num_tokens": 393611841.0, + "step": 15351 + }, + { + "epoch": 1.6859213705249285, + "grad_norm": 2.2423171997070312, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7155070304870605, + "num_tokens": 393636852.0, + "step": 15352 + }, + { + "epoch": 1.6860311882275423, + "grad_norm": 2.102984666824341, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7074905633926392, + "num_tokens": 393665931.0, + "step": 15353 + }, + { + "epoch": 1.686141005930156, + "grad_norm": 2.0120797157287598, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7018314599990845, + "num_tokens": 393696351.0, + "step": 15354 + }, + { + "epoch": 1.6862508236327696, + "grad_norm": 2.183788776397705, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7196332812309265, + "num_tokens": 393723607.0, + "step": 15355 + }, + { + "epoch": 1.686360641335383, + "grad_norm": 2.3883016109466553, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7195456027984619, + "num_tokens": 393745387.0, + "step": 15356 + }, + { + "epoch": 1.6864704590379969, + "grad_norm": 2.872499704360962, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7375680208206177, + "num_tokens": 393762611.0, + "step": 15357 + }, + { + "epoch": 1.6865802767406106, + "grad_norm": 2.074552297592163, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6826688051223755, + "num_tokens": 393794664.0, + "step": 15358 + }, + { + "epoch": 1.6866900944432244, + "grad_norm": 2.394728183746338, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7316559553146362, + "num_tokens": 393816508.0, + "step": 15359 + }, + { + "epoch": 1.686799912145838, + "grad_norm": 2.721017360687256, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7200281620025635, + "num_tokens": 393836266.0, + "step": 15360 + }, + { + "epoch": 1.6869097298484514, + "grad_norm": 2.377816677093506, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7269635200500488, + "num_tokens": 393860018.0, + "step": 15361 + }, + { + "epoch": 1.6870195475510652, + "grad_norm": 2.169633388519287, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7133162021636963, + "num_tokens": 393887122.0, + "step": 15362 + }, + { + "epoch": 1.687129365253679, + "grad_norm": 2.27291202545166, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7047968506813049, + "num_tokens": 393912913.0, + "step": 15363 + }, + { + "epoch": 1.6872391829562927, + "grad_norm": 2.130603313446045, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7228546738624573, + "num_tokens": 393940644.0, + "step": 15364 + }, + { + "epoch": 1.6873490006589063, + "grad_norm": 2.4156384468078613, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7350156307220459, + "num_tokens": 393965079.0, + "step": 15365 + }, + { + "epoch": 1.6874588183615198, + "grad_norm": 2.062713861465454, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.7035793662071228, + "num_tokens": 393996958.0, + "step": 15366 + }, + { + "epoch": 1.6875686360641335, + "grad_norm": 2.245042324066162, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7072321176528931, + "num_tokens": 394023069.0, + "step": 15367 + }, + { + "epoch": 1.6876784537667473, + "grad_norm": 2.0494801998138428, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7267378568649292, + "num_tokens": 394051973.0, + "step": 15368 + }, + { + "epoch": 1.6877882714693608, + "grad_norm": 2.4370436668395996, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7656664848327637, + "num_tokens": 394073815.0, + "step": 15369 + }, + { + "epoch": 1.6878980891719744, + "grad_norm": 2.3966965675354004, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7182000875473022, + "num_tokens": 394097045.0, + "step": 15370 + }, + { + "epoch": 1.6880079068745881, + "grad_norm": 2.3662352561950684, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7049517631530762, + "num_tokens": 394121169.0, + "step": 15371 + }, + { + "epoch": 1.6881177245772019, + "grad_norm": 2.6212143898010254, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7359552979469299, + "num_tokens": 394140875.0, + "step": 15372 + }, + { + "epoch": 1.6882275422798156, + "grad_norm": 1.9758782386779785, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7131403088569641, + "num_tokens": 394171761.0, + "step": 15373 + }, + { + "epoch": 1.6883373599824292, + "grad_norm": 1.9957079887390137, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7195772528648376, + "num_tokens": 394203541.0, + "step": 15374 + }, + { + "epoch": 1.6884471776850427, + "grad_norm": 2.1811273097991943, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7024461030960083, + "num_tokens": 394230321.0, + "step": 15375 + }, + { + "epoch": 1.6885569953876565, + "grad_norm": 2.556853771209717, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7229275703430176, + "num_tokens": 394252083.0, + "step": 15376 + }, + { + "epoch": 1.6886668130902702, + "grad_norm": 2.245561361312866, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7151919603347778, + "num_tokens": 394278674.0, + "step": 15377 + }, + { + "epoch": 1.6887766307928838, + "grad_norm": 2.087902545928955, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7042337656021118, + "num_tokens": 394310120.0, + "step": 15378 + }, + { + "epoch": 1.6888864484954975, + "grad_norm": 2.390479564666748, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7171214818954468, + "num_tokens": 394335039.0, + "step": 15379 + }, + { + "epoch": 1.688996266198111, + "grad_norm": 2.573988676071167, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7342470288276672, + "num_tokens": 394355403.0, + "step": 15380 + }, + { + "epoch": 1.6891060839007248, + "grad_norm": 2.159844160079956, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7192583680152893, + "num_tokens": 394383195.0, + "step": 15381 + }, + { + "epoch": 1.6892159016033386, + "grad_norm": 2.1161117553710938, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7123292684555054, + "num_tokens": 394410333.0, + "step": 15382 + }, + { + "epoch": 1.689325719305952, + "grad_norm": 1.8951082229614258, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7019597887992859, + "num_tokens": 394447082.0, + "step": 15383 + }, + { + "epoch": 1.6894355370085656, + "grad_norm": 2.1363525390625, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7144920229911804, + "num_tokens": 394474676.0, + "step": 15384 + }, + { + "epoch": 1.6895453547111794, + "grad_norm": 2.311225652694702, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7059357166290283, + "num_tokens": 394501944.0, + "step": 15385 + }, + { + "epoch": 1.6896551724137931, + "grad_norm": 2.3628759384155273, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7015345692634583, + "num_tokens": 394525795.0, + "step": 15386 + }, + { + "epoch": 1.689764990116407, + "grad_norm": 2.138779640197754, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.712031364440918, + "num_tokens": 394554635.0, + "step": 15387 + }, + { + "epoch": 1.6898748078190204, + "grad_norm": 2.4663000106811523, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7093397378921509, + "num_tokens": 394577135.0, + "step": 15388 + }, + { + "epoch": 1.689984625521634, + "grad_norm": 2.1906933784484863, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.728500247001648, + "num_tokens": 394602823.0, + "step": 15389 + }, + { + "epoch": 1.6900944432242477, + "grad_norm": 2.3459880352020264, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7067638635635376, + "num_tokens": 394626168.0, + "step": 15390 + }, + { + "epoch": 1.6902042609268615, + "grad_norm": 2.2386083602905273, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7283411026000977, + "num_tokens": 394652273.0, + "step": 15391 + }, + { + "epoch": 1.690314078629475, + "grad_norm": 1.8897987604141235, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7242810726165771, + "num_tokens": 394684517.0, + "step": 15392 + }, + { + "epoch": 1.6904238963320888, + "grad_norm": 2.073943614959717, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7158594131469727, + "num_tokens": 394714735.0, + "step": 15393 + }, + { + "epoch": 1.6905337140347023, + "grad_norm": 2.41454815864563, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7310937643051147, + "num_tokens": 394737609.0, + "step": 15394 + }, + { + "epoch": 1.690643531737316, + "grad_norm": 2.4641551971435547, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7122964859008789, + "num_tokens": 394759895.0, + "step": 15395 + }, + { + "epoch": 1.6907533494399298, + "grad_norm": 2.305190324783325, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7286112904548645, + "num_tokens": 394783849.0, + "step": 15396 + }, + { + "epoch": 1.6908631671425434, + "grad_norm": 2.1844496726989746, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7036831378936768, + "num_tokens": 394809457.0, + "step": 15397 + }, + { + "epoch": 1.690972984845157, + "grad_norm": 2.0492749214172363, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6944006681442261, + "num_tokens": 394843800.0, + "step": 15398 + }, + { + "epoch": 1.6910828025477707, + "grad_norm": 2.353233814239502, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7204432487487793, + "num_tokens": 394868223.0, + "step": 15399 + }, + { + "epoch": 1.6911926202503844, + "grad_norm": 2.166212797164917, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7092114686965942, + "num_tokens": 394895488.0, + "step": 15400 + }, + { + "epoch": 1.6913024379529982, + "grad_norm": 2.250906229019165, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7116122841835022, + "num_tokens": 394921744.0, + "step": 15401 + }, + { + "epoch": 1.6914122556556117, + "grad_norm": 2.2315151691436768, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7256903648376465, + "num_tokens": 394947953.0, + "step": 15402 + }, + { + "epoch": 1.6915220733582252, + "grad_norm": 2.213264226913452, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7185859680175781, + "num_tokens": 394975201.0, + "step": 15403 + }, + { + "epoch": 1.691631891060839, + "grad_norm": 1.9358866214752197, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7026002407073975, + "num_tokens": 395009683.0, + "step": 15404 + }, + { + "epoch": 1.6917417087634528, + "grad_norm": 2.152569055557251, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7184484004974365, + "num_tokens": 395037260.0, + "step": 15405 + }, + { + "epoch": 1.6918515264660663, + "grad_norm": 2.2051610946655273, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7158035039901733, + "num_tokens": 395064500.0, + "step": 15406 + }, + { + "epoch": 1.6919613441686798, + "grad_norm": 1.895490050315857, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7005114555358887, + "num_tokens": 395098526.0, + "step": 15407 + }, + { + "epoch": 1.6920711618712936, + "grad_norm": 2.4553298950195312, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7279634475708008, + "num_tokens": 395120607.0, + "step": 15408 + }, + { + "epoch": 1.6921809795739073, + "grad_norm": 2.1863839626312256, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7101418972015381, + "num_tokens": 395151598.0, + "step": 15409 + }, + { + "epoch": 1.692290797276521, + "grad_norm": 1.9713311195373535, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7241893410682678, + "num_tokens": 395182092.0, + "step": 15410 + }, + { + "epoch": 1.6924006149791346, + "grad_norm": 1.9227231740951538, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7007886171340942, + "num_tokens": 395216099.0, + "step": 15411 + }, + { + "epoch": 1.6925104326817482, + "grad_norm": 2.295095682144165, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7284498810768127, + "num_tokens": 395239157.0, + "step": 15412 + }, + { + "epoch": 1.692620250384362, + "grad_norm": 2.231473922729492, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6983723640441895, + "num_tokens": 395266603.0, + "step": 15413 + }, + { + "epoch": 1.6927300680869757, + "grad_norm": 1.930936574935913, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7442122101783752, + "num_tokens": 395297103.0, + "step": 15414 + }, + { + "epoch": 1.6928398857895894, + "grad_norm": 2.418579339981079, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7330746054649353, + "num_tokens": 395318901.0, + "step": 15415 + }, + { + "epoch": 1.692949703492203, + "grad_norm": 2.0574886798858643, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7177253365516663, + "num_tokens": 395345504.0, + "step": 15416 + }, + { + "epoch": 1.6930595211948165, + "grad_norm": 2.3748064041137695, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7332960367202759, + "num_tokens": 395366722.0, + "step": 15417 + }, + { + "epoch": 1.6931693388974303, + "grad_norm": 2.2011425495147705, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7279262542724609, + "num_tokens": 395391613.0, + "step": 15418 + }, + { + "epoch": 1.693279156600044, + "grad_norm": 2.1870341300964355, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7372615337371826, + "num_tokens": 395416861.0, + "step": 15419 + }, + { + "epoch": 1.6933889743026576, + "grad_norm": 2.3879926204681396, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7328174710273743, + "num_tokens": 395437868.0, + "step": 15420 + }, + { + "epoch": 1.693498792005271, + "grad_norm": 2.851846933364868, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7424773573875427, + "num_tokens": 395455219.0, + "step": 15421 + }, + { + "epoch": 1.6936086097078848, + "grad_norm": 1.8765032291412354, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.6999993920326233, + "num_tokens": 395491653.0, + "step": 15422 + }, + { + "epoch": 1.6937184274104986, + "grad_norm": 2.381958246231079, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7140635251998901, + "num_tokens": 395514835.0, + "step": 15423 + }, + { + "epoch": 1.6938282451131124, + "grad_norm": 2.2863101959228516, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.698351263999939, + "num_tokens": 395539547.0, + "step": 15424 + }, + { + "epoch": 1.693938062815726, + "grad_norm": 2.1469039916992188, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7093490362167358, + "num_tokens": 395568364.0, + "step": 15425 + }, + { + "epoch": 1.6940478805183394, + "grad_norm": 2.1895077228546143, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7045704126358032, + "num_tokens": 395594647.0, + "step": 15426 + }, + { + "epoch": 1.6941576982209532, + "grad_norm": 2.4932057857513428, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7338632941246033, + "num_tokens": 395616421.0, + "step": 15427 + }, + { + "epoch": 1.694267515923567, + "grad_norm": 2.3933591842651367, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7466607689857483, + "num_tokens": 395639629.0, + "step": 15428 + }, + { + "epoch": 1.6943773336261807, + "grad_norm": 2.4787862300872803, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7302484512329102, + "num_tokens": 395662087.0, + "step": 15429 + }, + { + "epoch": 1.6944871513287942, + "grad_norm": 2.1711902618408203, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7173898816108704, + "num_tokens": 395688737.0, + "step": 15430 + }, + { + "epoch": 1.6945969690314078, + "grad_norm": 2.1898276805877686, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7453448176383972, + "num_tokens": 395714354.0, + "step": 15431 + }, + { + "epoch": 1.6947067867340215, + "grad_norm": 2.1756601333618164, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7228735685348511, + "num_tokens": 395742510.0, + "step": 15432 + }, + { + "epoch": 1.6948166044366353, + "grad_norm": 2.114358425140381, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7220486402511597, + "num_tokens": 395768976.0, + "step": 15433 + }, + { + "epoch": 1.6949264221392488, + "grad_norm": 2.0441806316375732, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7263520956039429, + "num_tokens": 395797503.0, + "step": 15434 + }, + { + "epoch": 1.6950362398418624, + "grad_norm": 1.9672062397003174, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6958606243133545, + "num_tokens": 395830528.0, + "step": 15435 + }, + { + "epoch": 1.6951460575444761, + "grad_norm": 2.311614513397217, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7085726261138916, + "num_tokens": 395854654.0, + "step": 15436 + }, + { + "epoch": 1.6952558752470899, + "grad_norm": 2.459249258041382, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7127043008804321, + "num_tokens": 395877753.0, + "step": 15437 + }, + { + "epoch": 1.6953656929497036, + "grad_norm": 2.1238865852355957, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7061305046081543, + "num_tokens": 395902739.0, + "step": 15438 + }, + { + "epoch": 1.6954755106523172, + "grad_norm": 2.0602059364318848, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7061254978179932, + "num_tokens": 395934004.0, + "step": 15439 + }, + { + "epoch": 1.6955853283549307, + "grad_norm": 2.136944055557251, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7194087505340576, + "num_tokens": 395959572.0, + "step": 15440 + }, + { + "epoch": 1.6956951460575445, + "grad_norm": 2.1886491775512695, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7143508195877075, + "num_tokens": 395986058.0, + "step": 15441 + }, + { + "epoch": 1.6958049637601582, + "grad_norm": 2.7167840003967285, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7286144495010376, + "num_tokens": 396005177.0, + "step": 15442 + }, + { + "epoch": 1.6959147814627717, + "grad_norm": 2.2416322231292725, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7247152328491211, + "num_tokens": 396029816.0, + "step": 15443 + }, + { + "epoch": 1.6960245991653855, + "grad_norm": 2.397451400756836, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.718694806098938, + "num_tokens": 396053533.0, + "step": 15444 + }, + { + "epoch": 1.696134416867999, + "grad_norm": 2.5502147674560547, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7206224799156189, + "num_tokens": 396074441.0, + "step": 15445 + }, + { + "epoch": 1.6962442345706128, + "grad_norm": 2.2523996829986572, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7593103647232056, + "num_tokens": 396098770.0, + "step": 15446 + }, + { + "epoch": 1.6963540522732266, + "grad_norm": 2.0908966064453125, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7206684350967407, + "num_tokens": 396124763.0, + "step": 15447 + }, + { + "epoch": 1.69646386997584, + "grad_norm": 2.2665693759918213, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7379084825515747, + "num_tokens": 396148146.0, + "step": 15448 + }, + { + "epoch": 1.6965736876784536, + "grad_norm": 2.264397621154785, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.715428352355957, + "num_tokens": 396172954.0, + "step": 15449 + }, + { + "epoch": 1.6966835053810674, + "grad_norm": 2.0389912128448486, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7359098196029663, + "num_tokens": 396202714.0, + "step": 15450 + }, + { + "epoch": 1.6967933230836811, + "grad_norm": 2.121908664703369, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7289004325866699, + "num_tokens": 396228397.0, + "step": 15451 + }, + { + "epoch": 1.696903140786295, + "grad_norm": 2.4719419479370117, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7540521621704102, + "num_tokens": 396248793.0, + "step": 15452 + }, + { + "epoch": 1.6970129584889084, + "grad_norm": 1.9871248006820679, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7056690454483032, + "num_tokens": 396279687.0, + "step": 15453 + }, + { + "epoch": 1.697122776191522, + "grad_norm": 2.3633134365081787, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7267002463340759, + "num_tokens": 396304143.0, + "step": 15454 + }, + { + "epoch": 1.6972325938941357, + "grad_norm": 2.2785837650299072, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7265485525131226, + "num_tokens": 396330624.0, + "step": 15455 + }, + { + "epoch": 1.6973424115967495, + "grad_norm": 2.423088312149048, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7205132246017456, + "num_tokens": 396353041.0, + "step": 15456 + }, + { + "epoch": 1.697452229299363, + "grad_norm": 2.492870807647705, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7389159798622131, + "num_tokens": 396374665.0, + "step": 15457 + }, + { + "epoch": 1.6975620470019768, + "grad_norm": 2.0832269191741943, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7081067562103271, + "num_tokens": 396403383.0, + "step": 15458 + }, + { + "epoch": 1.6976718647045903, + "grad_norm": 2.417529582977295, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7027620077133179, + "num_tokens": 396426005.0, + "step": 15459 + }, + { + "epoch": 1.697781682407204, + "grad_norm": 2.410223960876465, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7388111352920532, + "num_tokens": 396448005.0, + "step": 15460 + }, + { + "epoch": 1.6978915001098178, + "grad_norm": 2.485671281814575, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7301145195960999, + "num_tokens": 396470643.0, + "step": 15461 + }, + { + "epoch": 1.6980013178124314, + "grad_norm": 2.225346803665161, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7132009863853455, + "num_tokens": 396495942.0, + "step": 15462 + }, + { + "epoch": 1.698111135515045, + "grad_norm": 2.567883253097534, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7247463464736938, + "num_tokens": 396518842.0, + "step": 15463 + }, + { + "epoch": 1.6982209532176586, + "grad_norm": 1.8708531856536865, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.713456928730011, + "num_tokens": 396550857.0, + "step": 15464 + }, + { + "epoch": 1.6983307709202724, + "grad_norm": 2.330137252807617, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7197030782699585, + "num_tokens": 396573071.0, + "step": 15465 + }, + { + "epoch": 1.6984405886228862, + "grad_norm": 2.2568228244781494, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7129524946212769, + "num_tokens": 396598090.0, + "step": 15466 + }, + { + "epoch": 1.6985504063254997, + "grad_norm": 2.276540756225586, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.709270715713501, + "num_tokens": 396623088.0, + "step": 15467 + }, + { + "epoch": 1.6986602240281132, + "grad_norm": 2.277453899383545, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7261413931846619, + "num_tokens": 396649854.0, + "step": 15468 + }, + { + "epoch": 1.698770041730727, + "grad_norm": 1.913049578666687, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7197756767272949, + "num_tokens": 396683597.0, + "step": 15469 + }, + { + "epoch": 1.6988798594333407, + "grad_norm": 1.9599285125732422, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7054475545883179, + "num_tokens": 396713734.0, + "step": 15470 + }, + { + "epoch": 1.6989896771359543, + "grad_norm": 2.359340190887451, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7120456099510193, + "num_tokens": 396737432.0, + "step": 15471 + }, + { + "epoch": 1.6990994948385678, + "grad_norm": 2.01019549369812, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7070955038070679, + "num_tokens": 396766983.0, + "step": 15472 + }, + { + "epoch": 1.6992093125411816, + "grad_norm": 2.0539634227752686, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6950260996818542, + "num_tokens": 396795032.0, + "step": 15473 + }, + { + "epoch": 1.6993191302437953, + "grad_norm": 2.7771854400634766, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7136543989181519, + "num_tokens": 396812955.0, + "step": 15474 + }, + { + "epoch": 1.699428947946409, + "grad_norm": 2.2373993396759033, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.721534252166748, + "num_tokens": 396837842.0, + "step": 15475 + }, + { + "epoch": 1.6995387656490226, + "grad_norm": 2.1604421138763428, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7344484329223633, + "num_tokens": 396863028.0, + "step": 15476 + }, + { + "epoch": 1.6996485833516362, + "grad_norm": 2.3735744953155518, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7059527039527893, + "num_tokens": 396887620.0, + "step": 15477 + }, + { + "epoch": 1.69975840105425, + "grad_norm": 2.36283278465271, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7236590385437012, + "num_tokens": 396909013.0, + "step": 15478 + }, + { + "epoch": 1.6998682187568637, + "grad_norm": 2.4911508560180664, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7392696142196655, + "num_tokens": 396929248.0, + "step": 15479 + }, + { + "epoch": 1.6999780364594774, + "grad_norm": 2.422820568084717, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7191314697265625, + "num_tokens": 396952388.0, + "step": 15480 + }, + { + "epoch": 1.700087854162091, + "grad_norm": 2.6193549633026123, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7471178770065308, + "num_tokens": 396971353.0, + "step": 15481 + }, + { + "epoch": 1.7001976718647045, + "grad_norm": 2.1833462715148926, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7159339189529419, + "num_tokens": 396998071.0, + "step": 15482 + }, + { + "epoch": 1.7003074895673183, + "grad_norm": 2.398585796356201, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7368611097335815, + "num_tokens": 397020374.0, + "step": 15483 + }, + { + "epoch": 1.700417307269932, + "grad_norm": 2.1510326862335205, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7083504796028137, + "num_tokens": 397048006.0, + "step": 15484 + }, + { + "epoch": 1.7005271249725455, + "grad_norm": 1.9421839714050293, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7182712554931641, + "num_tokens": 397078478.0, + "step": 15485 + }, + { + "epoch": 1.700636942675159, + "grad_norm": 1.9152247905731201, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.704241156578064, + "num_tokens": 397111996.0, + "step": 15486 + }, + { + "epoch": 1.7007467603777728, + "grad_norm": 2.2341365814208984, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7231345176696777, + "num_tokens": 397137217.0, + "step": 15487 + }, + { + "epoch": 1.7008565780803866, + "grad_norm": 2.421531915664673, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.725641667842865, + "num_tokens": 397159951.0, + "step": 15488 + }, + { + "epoch": 1.7009663957830004, + "grad_norm": 2.3156583309173584, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7283285856246948, + "num_tokens": 397183261.0, + "step": 15489 + }, + { + "epoch": 1.7010762134856139, + "grad_norm": 2.4793946743011475, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7328199148178101, + "num_tokens": 397206382.0, + "step": 15490 + }, + { + "epoch": 1.7011860311882274, + "grad_norm": 2.014223098754883, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.694374680519104, + "num_tokens": 397239998.0, + "step": 15491 + }, + { + "epoch": 1.7012958488908412, + "grad_norm": 2.145780563354492, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7270426750183105, + "num_tokens": 397264272.0, + "step": 15492 + }, + { + "epoch": 1.701405666593455, + "grad_norm": 2.206526279449463, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7356019020080566, + "num_tokens": 397288791.0, + "step": 15493 + }, + { + "epoch": 1.7015154842960687, + "grad_norm": 2.193305015563965, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7331469058990479, + "num_tokens": 397316711.0, + "step": 15494 + }, + { + "epoch": 1.7016253019986822, + "grad_norm": 2.1720235347747803, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7113871574401855, + "num_tokens": 397344225.0, + "step": 15495 + }, + { + "epoch": 1.7017351197012958, + "grad_norm": 2.3169808387756348, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7130930423736572, + "num_tokens": 397368956.0, + "step": 15496 + }, + { + "epoch": 1.7018449374039095, + "grad_norm": 2.2075443267822266, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.727695643901825, + "num_tokens": 397394118.0, + "step": 15497 + }, + { + "epoch": 1.7019547551065233, + "grad_norm": 2.1257288455963135, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.7122082710266113, + "num_tokens": 397422760.0, + "step": 15498 + }, + { + "epoch": 1.7020645728091368, + "grad_norm": 2.235365867614746, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7491065263748169, + "num_tokens": 397447378.0, + "step": 15499 + }, + { + "epoch": 1.7021743905117503, + "grad_norm": 2.083846092224121, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6976076364517212, + "num_tokens": 397474712.0, + "step": 15500 + }, + { + "epoch": 1.702284208214364, + "grad_norm": 2.2179603576660156, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7171909809112549, + "num_tokens": 397499628.0, + "step": 15501 + }, + { + "epoch": 1.7023940259169779, + "grad_norm": 2.269197463989258, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7442587614059448, + "num_tokens": 397523179.0, + "step": 15502 + }, + { + "epoch": 1.7025038436195916, + "grad_norm": 2.0214040279388428, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6886000633239746, + "num_tokens": 397554553.0, + "step": 15503 + }, + { + "epoch": 1.7026136613222052, + "grad_norm": 2.3444793224334717, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7145418524742126, + "num_tokens": 397579864.0, + "step": 15504 + }, + { + "epoch": 1.7027234790248187, + "grad_norm": 2.6952600479125977, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7284843325614929, + "num_tokens": 397599264.0, + "step": 15505 + }, + { + "epoch": 1.7028332967274324, + "grad_norm": 2.6185050010681152, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7284346222877502, + "num_tokens": 397620326.0, + "step": 15506 + }, + { + "epoch": 1.7029431144300462, + "grad_norm": 2.032980442047119, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7134738564491272, + "num_tokens": 397649790.0, + "step": 15507 + }, + { + "epoch": 1.7030529321326597, + "grad_norm": 2.2804770469665527, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7152959108352661, + "num_tokens": 397676364.0, + "step": 15508 + }, + { + "epoch": 1.7031627498352735, + "grad_norm": 2.415926456451416, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7415928840637207, + "num_tokens": 397698641.0, + "step": 15509 + }, + { + "epoch": 1.703272567537887, + "grad_norm": 2.299386739730835, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7410167455673218, + "num_tokens": 397721721.0, + "step": 15510 + }, + { + "epoch": 1.7033823852405008, + "grad_norm": 1.8623539209365845, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.715398907661438, + "num_tokens": 397757177.0, + "step": 15511 + }, + { + "epoch": 1.7034922029431145, + "grad_norm": 2.513378381729126, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7319977879524231, + "num_tokens": 397780872.0, + "step": 15512 + }, + { + "epoch": 1.703602020645728, + "grad_norm": 2.659156084060669, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7366659641265869, + "num_tokens": 397798568.0, + "step": 15513 + }, + { + "epoch": 1.7037118383483416, + "grad_norm": 2.2792696952819824, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7218886613845825, + "num_tokens": 397825237.0, + "step": 15514 + }, + { + "epoch": 1.7038216560509554, + "grad_norm": 2.258608102798462, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7340643405914307, + "num_tokens": 397848940.0, + "step": 15515 + }, + { + "epoch": 1.7039314737535691, + "grad_norm": 2.2026069164276123, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.706615686416626, + "num_tokens": 397874440.0, + "step": 15516 + }, + { + "epoch": 1.7040412914561829, + "grad_norm": 2.3471319675445557, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7215871810913086, + "num_tokens": 397896131.0, + "step": 15517 + }, + { + "epoch": 1.7041511091587964, + "grad_norm": 2.3093769550323486, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7156342267990112, + "num_tokens": 397920923.0, + "step": 15518 + }, + { + "epoch": 1.70426092686141, + "grad_norm": 2.5066726207733154, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7265756130218506, + "num_tokens": 397941078.0, + "step": 15519 + }, + { + "epoch": 1.7043707445640237, + "grad_norm": 2.257091760635376, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7138872146606445, + "num_tokens": 397967326.0, + "step": 15520 + }, + { + "epoch": 1.7044805622666375, + "grad_norm": 2.3393123149871826, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7187108993530273, + "num_tokens": 397991115.0, + "step": 15521 + }, + { + "epoch": 1.704590379969251, + "grad_norm": 2.171738386154175, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7262884378433228, + "num_tokens": 398016144.0, + "step": 15522 + }, + { + "epoch": 1.7047001976718648, + "grad_norm": 2.1818509101867676, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.711014449596405, + "num_tokens": 398045494.0, + "step": 15523 + }, + { + "epoch": 1.7048100153744783, + "grad_norm": 2.294666290283203, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7103302478790283, + "num_tokens": 398071689.0, + "step": 15524 + }, + { + "epoch": 1.704919833077092, + "grad_norm": 2.0225725173950195, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7177575826644897, + "num_tokens": 398104118.0, + "step": 15525 + }, + { + "epoch": 1.7050296507797058, + "grad_norm": 2.5209765434265137, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7239952683448792, + "num_tokens": 398124865.0, + "step": 15526 + }, + { + "epoch": 1.7051394684823193, + "grad_norm": 2.296085834503174, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7152892351150513, + "num_tokens": 398149806.0, + "step": 15527 + }, + { + "epoch": 1.7052492861849329, + "grad_norm": 2.3057916164398193, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7286339998245239, + "num_tokens": 398172215.0, + "step": 15528 + }, + { + "epoch": 1.7053591038875466, + "grad_norm": 2.3970139026641846, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7127991914749146, + "num_tokens": 398193603.0, + "step": 15529 + }, + { + "epoch": 1.7054689215901604, + "grad_norm": 2.2452964782714844, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7171109914779663, + "num_tokens": 398219994.0, + "step": 15530 + }, + { + "epoch": 1.7055787392927741, + "grad_norm": 2.3703322410583496, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7262741923332214, + "num_tokens": 398243165.0, + "step": 15531 + }, + { + "epoch": 1.7056885569953877, + "grad_norm": 1.9457584619522095, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7123615741729736, + "num_tokens": 398277235.0, + "step": 15532 + }, + { + "epoch": 1.7057983746980012, + "grad_norm": 2.2906856536865234, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7183617353439331, + "num_tokens": 398302260.0, + "step": 15533 + }, + { + "epoch": 1.705908192400615, + "grad_norm": 2.19994854927063, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7135952711105347, + "num_tokens": 398328771.0, + "step": 15534 + }, + { + "epoch": 1.7060180101032287, + "grad_norm": 2.3612592220306396, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7340030670166016, + "num_tokens": 398353404.0, + "step": 15535 + }, + { + "epoch": 1.7061278278058423, + "grad_norm": 2.0780186653137207, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6922401785850525, + "num_tokens": 398385376.0, + "step": 15536 + }, + { + "epoch": 1.7062376455084558, + "grad_norm": 2.5435376167297363, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7307142019271851, + "num_tokens": 398405253.0, + "step": 15537 + }, + { + "epoch": 1.7063474632110696, + "grad_norm": 2.3317015171051025, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7182753086090088, + "num_tokens": 398429412.0, + "step": 15538 + }, + { + "epoch": 1.7064572809136833, + "grad_norm": 2.267082691192627, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7208863496780396, + "num_tokens": 398453340.0, + "step": 15539 + }, + { + "epoch": 1.706567098616297, + "grad_norm": 2.475654363632202, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7116696834564209, + "num_tokens": 398475483.0, + "step": 15540 + }, + { + "epoch": 1.7066769163189106, + "grad_norm": 2.044520378112793, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7347227334976196, + "num_tokens": 398507094.0, + "step": 15541 + }, + { + "epoch": 1.7067867340215241, + "grad_norm": 2.1084649562835693, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7229611277580261, + "num_tokens": 398537026.0, + "step": 15542 + }, + { + "epoch": 1.706896551724138, + "grad_norm": 2.4008660316467285, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7283650040626526, + "num_tokens": 398559495.0, + "step": 15543 + }, + { + "epoch": 1.7070063694267517, + "grad_norm": 2.3379666805267334, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7058268189430237, + "num_tokens": 398584952.0, + "step": 15544 + }, + { + "epoch": 1.7071161871293654, + "grad_norm": 2.4748167991638184, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7195088863372803, + "num_tokens": 398607714.0, + "step": 15545 + }, + { + "epoch": 1.707226004831979, + "grad_norm": 2.406367301940918, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7231239080429077, + "num_tokens": 398629602.0, + "step": 15546 + }, + { + "epoch": 1.7073358225345925, + "grad_norm": 2.017652750015259, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7425711750984192, + "num_tokens": 398655454.0, + "step": 15547 + }, + { + "epoch": 1.7074456402372062, + "grad_norm": 2.3131566047668457, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7297381162643433, + "num_tokens": 398679702.0, + "step": 15548 + }, + { + "epoch": 1.70755545793982, + "grad_norm": 2.1633737087249756, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7013548612594604, + "num_tokens": 398707502.0, + "step": 15549 + }, + { + "epoch": 1.7076652756424335, + "grad_norm": 2.4121806621551514, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7387746572494507, + "num_tokens": 398727220.0, + "step": 15550 + }, + { + "epoch": 1.707775093345047, + "grad_norm": 2.5378470420837402, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7347162961959839, + "num_tokens": 398747954.0, + "step": 15551 + }, + { + "epoch": 1.7078849110476608, + "grad_norm": 2.135427951812744, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7185211777687073, + "num_tokens": 398773675.0, + "step": 15552 + }, + { + "epoch": 1.7079947287502746, + "grad_norm": 2.160048246383667, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7117595076560974, + "num_tokens": 398800422.0, + "step": 15553 + }, + { + "epoch": 1.7081045464528883, + "grad_norm": 2.478973865509033, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7357056140899658, + "num_tokens": 398822529.0, + "step": 15554 + }, + { + "epoch": 1.7082143641555019, + "grad_norm": 2.438138961791992, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7360608577728271, + "num_tokens": 398846684.0, + "step": 15555 + }, + { + "epoch": 1.7083241818581154, + "grad_norm": 2.43815279006958, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7269555926322937, + "num_tokens": 398868114.0, + "step": 15556 + }, + { + "epoch": 1.7084339995607292, + "grad_norm": 2.580979347229004, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7355775833129883, + "num_tokens": 398886940.0, + "step": 15557 + }, + { + "epoch": 1.708543817263343, + "grad_norm": 2.257471799850464, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7043759822845459, + "num_tokens": 398911864.0, + "step": 15558 + }, + { + "epoch": 1.7086536349659565, + "grad_norm": 2.24943470954895, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7182559967041016, + "num_tokens": 398938423.0, + "step": 15559 + }, + { + "epoch": 1.7087634526685702, + "grad_norm": 2.3367950916290283, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7336764335632324, + "num_tokens": 398962305.0, + "step": 15560 + }, + { + "epoch": 1.7088732703711838, + "grad_norm": 2.2730605602264404, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7218749523162842, + "num_tokens": 398986737.0, + "step": 15561 + }, + { + "epoch": 1.7089830880737975, + "grad_norm": 2.3837039470672607, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7181344032287598, + "num_tokens": 399009966.0, + "step": 15562 + }, + { + "epoch": 1.7090929057764113, + "grad_norm": 2.1588823795318604, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7158828377723694, + "num_tokens": 399035355.0, + "step": 15563 + }, + { + "epoch": 1.7092027234790248, + "grad_norm": 1.9423471689224243, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7122107148170471, + "num_tokens": 399065493.0, + "step": 15564 + }, + { + "epoch": 1.7093125411816383, + "grad_norm": 2.1774611473083496, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7407764792442322, + "num_tokens": 399090569.0, + "step": 15565 + }, + { + "epoch": 1.709422358884252, + "grad_norm": 2.35691499710083, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7316863536834717, + "num_tokens": 399114806.0, + "step": 15566 + }, + { + "epoch": 1.7095321765868658, + "grad_norm": 1.9572458267211914, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.712610125541687, + "num_tokens": 399147173.0, + "step": 15567 + }, + { + "epoch": 1.7096419942894796, + "grad_norm": 2.2343215942382812, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7023595571517944, + "num_tokens": 399174498.0, + "step": 15568 + }, + { + "epoch": 1.7097518119920931, + "grad_norm": 2.180100679397583, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6874501705169678, + "num_tokens": 399200928.0, + "step": 15569 + }, + { + "epoch": 1.7098616296947067, + "grad_norm": 2.4909889698028564, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7291936278343201, + "num_tokens": 399223118.0, + "step": 15570 + }, + { + "epoch": 1.7099714473973204, + "grad_norm": 2.594642400741577, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7269306182861328, + "num_tokens": 399243109.0, + "step": 15571 + }, + { + "epoch": 1.7100812650999342, + "grad_norm": 2.01061749458313, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.700857400894165, + "num_tokens": 399274398.0, + "step": 15572 + }, + { + "epoch": 1.7101910828025477, + "grad_norm": 2.370493173599243, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7409021854400635, + "num_tokens": 399295350.0, + "step": 15573 + }, + { + "epoch": 1.7103009005051615, + "grad_norm": 2.2798256874084473, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7142120003700256, + "num_tokens": 399320807.0, + "step": 15574 + }, + { + "epoch": 1.710410718207775, + "grad_norm": 2.582524538040161, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7271530032157898, + "num_tokens": 399341491.0, + "step": 15575 + }, + { + "epoch": 1.7105205359103888, + "grad_norm": 2.5109574794769287, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7395870685577393, + "num_tokens": 399363620.0, + "step": 15576 + }, + { + "epoch": 1.7106303536130025, + "grad_norm": 2.160250186920166, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7262665033340454, + "num_tokens": 399390976.0, + "step": 15577 + }, + { + "epoch": 1.710740171315616, + "grad_norm": 2.0874900817871094, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7245914936065674, + "num_tokens": 399419703.0, + "step": 15578 + }, + { + "epoch": 1.7108499890182296, + "grad_norm": 2.0809664726257324, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7359164953231812, + "num_tokens": 399446425.0, + "step": 15579 + }, + { + "epoch": 1.7109598067208434, + "grad_norm": 2.3210432529449463, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7202092409133911, + "num_tokens": 399471949.0, + "step": 15580 + }, + { + "epoch": 1.7110696244234571, + "grad_norm": 2.333920955657959, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7274587154388428, + "num_tokens": 399496972.0, + "step": 15581 + }, + { + "epoch": 1.7111794421260709, + "grad_norm": 2.45804762840271, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.732085645198822, + "num_tokens": 399518928.0, + "step": 15582 + }, + { + "epoch": 1.7112892598286844, + "grad_norm": 2.1933705806732178, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7293342351913452, + "num_tokens": 399544960.0, + "step": 15583 + }, + { + "epoch": 1.711399077531298, + "grad_norm": 2.171520233154297, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7189404964447021, + "num_tokens": 399572062.0, + "step": 15584 + }, + { + "epoch": 1.7115088952339117, + "grad_norm": 2.3378489017486572, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7092059850692749, + "num_tokens": 399598832.0, + "step": 15585 + }, + { + "epoch": 1.7116187129365255, + "grad_norm": 2.0215461254119873, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7166932821273804, + "num_tokens": 399628403.0, + "step": 15586 + }, + { + "epoch": 1.711728530639139, + "grad_norm": 2.104550361633301, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7373543381690979, + "num_tokens": 399654681.0, + "step": 15587 + }, + { + "epoch": 1.7118383483417525, + "grad_norm": 2.064737319946289, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7514920234680176, + "num_tokens": 399680773.0, + "step": 15588 + }, + { + "epoch": 1.7119481660443663, + "grad_norm": 2.377603054046631, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6887847185134888, + "num_tokens": 399706710.0, + "step": 15589 + }, + { + "epoch": 1.71205798374698, + "grad_norm": 2.084407329559326, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7203905582427979, + "num_tokens": 399738002.0, + "step": 15590 + }, + { + "epoch": 1.7121678014495938, + "grad_norm": 2.176224708557129, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7448216676712036, + "num_tokens": 399763552.0, + "step": 15591 + }, + { + "epoch": 1.7122776191522073, + "grad_norm": 2.094630002975464, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.718298614025116, + "num_tokens": 399790217.0, + "step": 15592 + }, + { + "epoch": 1.7123874368548209, + "grad_norm": 2.433053731918335, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.712765097618103, + "num_tokens": 399813192.0, + "step": 15593 + }, + { + "epoch": 1.7124972545574346, + "grad_norm": 2.2049832344055176, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7397757768630981, + "num_tokens": 399837789.0, + "step": 15594 + }, + { + "epoch": 1.7126070722600484, + "grad_norm": 2.3788905143737793, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7340252995491028, + "num_tokens": 399858747.0, + "step": 15595 + }, + { + "epoch": 1.7127168899626621, + "grad_norm": 2.046936273574829, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7135709524154663, + "num_tokens": 399887058.0, + "step": 15596 + }, + { + "epoch": 1.7128267076652757, + "grad_norm": 2.031769037246704, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7265649437904358, + "num_tokens": 399916625.0, + "step": 15597 + }, + { + "epoch": 1.7129365253678892, + "grad_norm": 2.1067519187927246, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.697428822517395, + "num_tokens": 399947527.0, + "step": 15598 + }, + { + "epoch": 1.713046343070503, + "grad_norm": 2.360065221786499, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.6973827481269836, + "num_tokens": 399971753.0, + "step": 15599 + }, + { + "epoch": 1.7131561607731167, + "grad_norm": 2.1599550247192383, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7378244996070862, + "num_tokens": 399999738.0, + "step": 15600 + }, + { + "epoch": 1.7132659784757303, + "grad_norm": 2.3659281730651855, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7091877460479736, + "num_tokens": 400024993.0, + "step": 15601 + }, + { + "epoch": 1.7133757961783438, + "grad_norm": 2.2476627826690674, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.718073844909668, + "num_tokens": 400051237.0, + "step": 15602 + }, + { + "epoch": 1.7134856138809575, + "grad_norm": 2.2067065238952637, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7173096537590027, + "num_tokens": 400077366.0, + "step": 15603 + }, + { + "epoch": 1.7135954315835713, + "grad_norm": 2.1220755577087402, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7258993983268738, + "num_tokens": 400108865.0, + "step": 15604 + }, + { + "epoch": 1.713705249286185, + "grad_norm": 1.8044097423553467, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7000448107719421, + "num_tokens": 400148105.0, + "step": 15605 + }, + { + "epoch": 1.7138150669887986, + "grad_norm": 2.346083641052246, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7088546752929688, + "num_tokens": 400173305.0, + "step": 15606 + }, + { + "epoch": 1.7139248846914121, + "grad_norm": 2.146085739135742, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7256502509117126, + "num_tokens": 400199887.0, + "step": 15607 + }, + { + "epoch": 1.7140347023940259, + "grad_norm": 2.0872225761413574, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7268064022064209, + "num_tokens": 400226020.0, + "step": 15608 + }, + { + "epoch": 1.7141445200966396, + "grad_norm": 2.235686779022217, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6809109449386597, + "num_tokens": 400253530.0, + "step": 15609 + }, + { + "epoch": 1.7142543377992534, + "grad_norm": 2.243784189224243, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7111640572547913, + "num_tokens": 400279932.0, + "step": 15610 + }, + { + "epoch": 1.714364155501867, + "grad_norm": 2.3759238719940186, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7223902344703674, + "num_tokens": 400302423.0, + "step": 15611 + }, + { + "epoch": 1.7144739732044805, + "grad_norm": 2.5160653591156006, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6948761343955994, + "num_tokens": 400324374.0, + "step": 15612 + }, + { + "epoch": 1.7145837909070942, + "grad_norm": 2.2375969886779785, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7113972902297974, + "num_tokens": 400351095.0, + "step": 15613 + }, + { + "epoch": 1.714693608609708, + "grad_norm": 2.407715320587158, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7126575708389282, + "num_tokens": 400377439.0, + "step": 15614 + }, + { + "epoch": 1.7148034263123215, + "grad_norm": 2.249028444290161, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7153253555297852, + "num_tokens": 400405144.0, + "step": 15615 + }, + { + "epoch": 1.714913244014935, + "grad_norm": 2.362565517425537, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7421780824661255, + "num_tokens": 400427267.0, + "step": 15616 + }, + { + "epoch": 1.7150230617175488, + "grad_norm": 2.1102867126464844, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.717035710811615, + "num_tokens": 400456835.0, + "step": 15617 + }, + { + "epoch": 1.7151328794201626, + "grad_norm": 2.0360167026519775, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7035694718360901, + "num_tokens": 400487249.0, + "step": 15618 + }, + { + "epoch": 1.7152426971227763, + "grad_norm": 2.279445171356201, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7340049743652344, + "num_tokens": 400511945.0, + "step": 15619 + }, + { + "epoch": 1.7153525148253899, + "grad_norm": 2.250391960144043, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7362982034683228, + "num_tokens": 400536386.0, + "step": 15620 + }, + { + "epoch": 1.7154623325280034, + "grad_norm": 2.4903485774993896, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7254210710525513, + "num_tokens": 400557395.0, + "step": 15621 + }, + { + "epoch": 1.7155721502306172, + "grad_norm": 1.961302638053894, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7033431529998779, + "num_tokens": 400591101.0, + "step": 15622 + }, + { + "epoch": 1.715681967933231, + "grad_norm": 2.145664691925049, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.729321300983429, + "num_tokens": 400617539.0, + "step": 15623 + }, + { + "epoch": 1.7157917856358444, + "grad_norm": 2.119586706161499, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7207273244857788, + "num_tokens": 400645665.0, + "step": 15624 + }, + { + "epoch": 1.7159016033384582, + "grad_norm": 2.0354702472686768, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7325092554092407, + "num_tokens": 400674503.0, + "step": 15625 + }, + { + "epoch": 1.7160114210410717, + "grad_norm": 2.276413917541504, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7122749090194702, + "num_tokens": 400699000.0, + "step": 15626 + }, + { + "epoch": 1.7161212387436855, + "grad_norm": 2.6829705238342285, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7236330509185791, + "num_tokens": 400720041.0, + "step": 15627 + }, + { + "epoch": 1.7162310564462993, + "grad_norm": 2.5252275466918945, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7216372489929199, + "num_tokens": 400742660.0, + "step": 15628 + }, + { + "epoch": 1.7163408741489128, + "grad_norm": 2.4979608058929443, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7298815250396729, + "num_tokens": 400765056.0, + "step": 15629 + }, + { + "epoch": 1.7164506918515263, + "grad_norm": 2.2378618717193604, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7150101065635681, + "num_tokens": 400791990.0, + "step": 15630 + }, + { + "epoch": 1.71656050955414, + "grad_norm": 2.6273365020751953, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7330238819122314, + "num_tokens": 400812540.0, + "step": 15631 + }, + { + "epoch": 1.7166703272567538, + "grad_norm": 2.021667718887329, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7177484035491943, + "num_tokens": 400841069.0, + "step": 15632 + }, + { + "epoch": 1.7167801449593676, + "grad_norm": 2.113765239715576, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7064212560653687, + "num_tokens": 400870040.0, + "step": 15633 + }, + { + "epoch": 1.7168899626619811, + "grad_norm": 2.196187973022461, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7315501570701599, + "num_tokens": 400895065.0, + "step": 15634 + }, + { + "epoch": 1.7169997803645947, + "grad_norm": 2.165112257003784, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7137784361839294, + "num_tokens": 400921589.0, + "step": 15635 + }, + { + "epoch": 1.7171095980672084, + "grad_norm": 2.1131579875946045, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.6980659365653992, + "num_tokens": 400949994.0, + "step": 15636 + }, + { + "epoch": 1.7172194157698222, + "grad_norm": 2.171579122543335, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7202401161193848, + "num_tokens": 400977095.0, + "step": 15637 + }, + { + "epoch": 1.7173292334724357, + "grad_norm": 2.620126724243164, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.734679639339447, + "num_tokens": 400995740.0, + "step": 15638 + }, + { + "epoch": 1.7174390511750495, + "grad_norm": 2.1808979511260986, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.699629545211792, + "num_tokens": 401022509.0, + "step": 15639 + }, + { + "epoch": 1.717548868877663, + "grad_norm": 2.016817092895508, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6908558011054993, + "num_tokens": 401052399.0, + "step": 15640 + }, + { + "epoch": 1.7176586865802768, + "grad_norm": 2.3541738986968994, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7230631113052368, + "num_tokens": 401076807.0, + "step": 15641 + }, + { + "epoch": 1.7177685042828905, + "grad_norm": 2.0220017433166504, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7123746275901794, + "num_tokens": 401107508.0, + "step": 15642 + }, + { + "epoch": 1.717878321985504, + "grad_norm": 2.162639617919922, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7275140285491943, + "num_tokens": 401135035.0, + "step": 15643 + }, + { + "epoch": 1.7179881396881176, + "grad_norm": 2.509399652481079, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7173606157302856, + "num_tokens": 401156002.0, + "step": 15644 + }, + { + "epoch": 1.7180979573907313, + "grad_norm": 2.0103251934051514, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.6998014450073242, + "num_tokens": 401186692.0, + "step": 15645 + }, + { + "epoch": 1.718207775093345, + "grad_norm": 1.9117810726165771, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7175440788269043, + "num_tokens": 401220304.0, + "step": 15646 + }, + { + "epoch": 1.7183175927959589, + "grad_norm": 2.5582056045532227, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7365605235099792, + "num_tokens": 401241331.0, + "step": 15647 + }, + { + "epoch": 1.7184274104985724, + "grad_norm": 2.326568841934204, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7151898145675659, + "num_tokens": 401264544.0, + "step": 15648 + }, + { + "epoch": 1.718537228201186, + "grad_norm": 2.5178701877593994, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7515000104904175, + "num_tokens": 401284405.0, + "step": 15649 + }, + { + "epoch": 1.7186470459037997, + "grad_norm": 2.183161735534668, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7357990741729736, + "num_tokens": 401310174.0, + "step": 15650 + }, + { + "epoch": 1.7187568636064134, + "grad_norm": 2.35160493850708, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.709700345993042, + "num_tokens": 401335236.0, + "step": 15651 + }, + { + "epoch": 1.718866681309027, + "grad_norm": 2.3665971755981445, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7177404761314392, + "num_tokens": 401357412.0, + "step": 15652 + }, + { + "epoch": 1.7189764990116405, + "grad_norm": 2.24393630027771, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7060830593109131, + "num_tokens": 401382366.0, + "step": 15653 + }, + { + "epoch": 1.7190863167142543, + "grad_norm": 2.4051547050476074, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.722016453742981, + "num_tokens": 401404206.0, + "step": 15654 + }, + { + "epoch": 1.719196134416868, + "grad_norm": 2.0922093391418457, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7175614833831787, + "num_tokens": 401431032.0, + "step": 15655 + }, + { + "epoch": 1.7193059521194818, + "grad_norm": 2.1501193046569824, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7020800113677979, + "num_tokens": 401458998.0, + "step": 15656 + }, + { + "epoch": 1.7194157698220953, + "grad_norm": 2.3189401626586914, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7217110991477966, + "num_tokens": 401483272.0, + "step": 15657 + }, + { + "epoch": 1.7195255875247089, + "grad_norm": 2.2425355911254883, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7191675901412964, + "num_tokens": 401509617.0, + "step": 15658 + }, + { + "epoch": 1.7196354052273226, + "grad_norm": 2.13163423538208, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.711943507194519, + "num_tokens": 401538708.0, + "step": 15659 + }, + { + "epoch": 1.7197452229299364, + "grad_norm": 2.100914716720581, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7056553363800049, + "num_tokens": 401567196.0, + "step": 15660 + }, + { + "epoch": 1.7198550406325501, + "grad_norm": 2.393200397491455, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7222511172294617, + "num_tokens": 401590298.0, + "step": 15661 + }, + { + "epoch": 1.7199648583351637, + "grad_norm": 2.5823638439178467, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7370266914367676, + "num_tokens": 401610070.0, + "step": 15662 + }, + { + "epoch": 1.7200746760377772, + "grad_norm": 2.167572021484375, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.717992901802063, + "num_tokens": 401636981.0, + "step": 15663 + }, + { + "epoch": 1.720184493740391, + "grad_norm": 2.082725763320923, + "learning_rate": 1e-06, + "loss": 1.08, + "mean_token_accuracy": 0.6888428926467896, + "num_tokens": 401668188.0, + "step": 15664 + }, + { + "epoch": 1.7202943114430047, + "grad_norm": 2.380868434906006, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7160570621490479, + "num_tokens": 401691512.0, + "step": 15665 + }, + { + "epoch": 1.7204041291456182, + "grad_norm": 2.2949016094207764, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7041550874710083, + "num_tokens": 401716195.0, + "step": 15666 + }, + { + "epoch": 1.7205139468482318, + "grad_norm": 1.8859809637069702, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6924452781677246, + "num_tokens": 401749982.0, + "step": 15667 + }, + { + "epoch": 1.7206237645508455, + "grad_norm": 2.5637145042419434, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7264308929443359, + "num_tokens": 401771087.0, + "step": 15668 + }, + { + "epoch": 1.7207335822534593, + "grad_norm": 1.9558194875717163, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7123068571090698, + "num_tokens": 401804765.0, + "step": 15669 + }, + { + "epoch": 1.720843399956073, + "grad_norm": 2.266937494277954, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7261091470718384, + "num_tokens": 401829100.0, + "step": 15670 + }, + { + "epoch": 1.7209532176586866, + "grad_norm": 2.2473673820495605, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7183083295822144, + "num_tokens": 401856272.0, + "step": 15671 + }, + { + "epoch": 1.7210630353613001, + "grad_norm": 1.9777249097824097, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7102886438369751, + "num_tokens": 401888542.0, + "step": 15672 + }, + { + "epoch": 1.7211728530639139, + "grad_norm": 2.1419081687927246, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7144163846969604, + "num_tokens": 401918436.0, + "step": 15673 + }, + { + "epoch": 1.7212826707665276, + "grad_norm": 2.23801326751709, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.708793044090271, + "num_tokens": 401944816.0, + "step": 15674 + }, + { + "epoch": 1.7213924884691414, + "grad_norm": 2.1764657497406006, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.6988391280174255, + "num_tokens": 401974279.0, + "step": 15675 + }, + { + "epoch": 1.721502306171755, + "grad_norm": 2.186138153076172, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7306511998176575, + "num_tokens": 402002253.0, + "step": 15676 + }, + { + "epoch": 1.7216121238743685, + "grad_norm": 2.2380588054656982, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7420895099639893, + "num_tokens": 402027286.0, + "step": 15677 + }, + { + "epoch": 1.7217219415769822, + "grad_norm": 2.4469261169433594, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7218829989433289, + "num_tokens": 402047979.0, + "step": 15678 + }, + { + "epoch": 1.721831759279596, + "grad_norm": 2.489906072616577, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.735183596611023, + "num_tokens": 402068165.0, + "step": 15679 + }, + { + "epoch": 1.7219415769822095, + "grad_norm": 2.224461555480957, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7216192483901978, + "num_tokens": 402093938.0, + "step": 15680 + }, + { + "epoch": 1.722051394684823, + "grad_norm": 1.9819728136062622, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7011328935623169, + "num_tokens": 402126337.0, + "step": 15681 + }, + { + "epoch": 1.7221612123874368, + "grad_norm": 2.4154930114746094, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7516635060310364, + "num_tokens": 402147082.0, + "step": 15682 + }, + { + "epoch": 1.7222710300900506, + "grad_norm": 2.0804266929626465, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7133522033691406, + "num_tokens": 402175754.0, + "step": 15683 + }, + { + "epoch": 1.7223808477926643, + "grad_norm": 2.1760449409484863, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7146521210670471, + "num_tokens": 402202609.0, + "step": 15684 + }, + { + "epoch": 1.7224906654952779, + "grad_norm": 2.3990976810455322, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7158012986183167, + "num_tokens": 402225501.0, + "step": 15685 + }, + { + "epoch": 1.7226004831978914, + "grad_norm": 2.406789541244507, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.697113037109375, + "num_tokens": 402249474.0, + "step": 15686 + }, + { + "epoch": 1.7227103009005051, + "grad_norm": 2.461552858352661, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.733879566192627, + "num_tokens": 402271803.0, + "step": 15687 + }, + { + "epoch": 1.722820118603119, + "grad_norm": 2.1622300148010254, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7157223224639893, + "num_tokens": 402297577.0, + "step": 15688 + }, + { + "epoch": 1.7229299363057324, + "grad_norm": 2.0787642002105713, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7096228003501892, + "num_tokens": 402326488.0, + "step": 15689 + }, + { + "epoch": 1.7230397540083462, + "grad_norm": 2.185178518295288, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7179238796234131, + "num_tokens": 402353504.0, + "step": 15690 + }, + { + "epoch": 1.7231495717109597, + "grad_norm": 1.9472591876983643, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7191023826599121, + "num_tokens": 402383385.0, + "step": 15691 + }, + { + "epoch": 1.7232593894135735, + "grad_norm": 2.419537305831909, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7379265427589417, + "num_tokens": 402407241.0, + "step": 15692 + }, + { + "epoch": 1.7233692071161872, + "grad_norm": 2.3472747802734375, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7335690259933472, + "num_tokens": 402431413.0, + "step": 15693 + }, + { + "epoch": 1.7234790248188008, + "grad_norm": 2.22189998626709, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7085052728652954, + "num_tokens": 402458720.0, + "step": 15694 + }, + { + "epoch": 1.7235888425214143, + "grad_norm": 1.9363434314727783, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6938092708587646, + "num_tokens": 402492097.0, + "step": 15695 + }, + { + "epoch": 1.723698660224028, + "grad_norm": 2.1990761756896973, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7209280729293823, + "num_tokens": 402518070.0, + "step": 15696 + }, + { + "epoch": 1.7238084779266418, + "grad_norm": 2.1802611351013184, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7106766700744629, + "num_tokens": 402546401.0, + "step": 15697 + }, + { + "epoch": 1.7239182956292556, + "grad_norm": 2.312329053878784, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7221214771270752, + "num_tokens": 402571715.0, + "step": 15698 + }, + { + "epoch": 1.7240281133318691, + "grad_norm": 2.1297006607055664, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7307153344154358, + "num_tokens": 402598489.0, + "step": 15699 + }, + { + "epoch": 1.7241379310344827, + "grad_norm": 2.166700601577759, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7171899676322937, + "num_tokens": 402625210.0, + "step": 15700 + }, + { + "epoch": 1.7242477487370964, + "grad_norm": 2.0816335678100586, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7113937139511108, + "num_tokens": 402654265.0, + "step": 15701 + }, + { + "epoch": 1.7243575664397102, + "grad_norm": 2.0186705589294434, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7009350061416626, + "num_tokens": 402684676.0, + "step": 15702 + }, + { + "epoch": 1.7244673841423237, + "grad_norm": 2.6202564239501953, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7437541484832764, + "num_tokens": 402703680.0, + "step": 15703 + }, + { + "epoch": 1.7245772018449375, + "grad_norm": 2.5105397701263428, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7397401332855225, + "num_tokens": 402724378.0, + "step": 15704 + }, + { + "epoch": 1.724687019547551, + "grad_norm": 2.507721185684204, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7387917041778564, + "num_tokens": 402744470.0, + "step": 15705 + }, + { + "epoch": 1.7247968372501647, + "grad_norm": 2.324367046356201, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.7019661664962769, + "num_tokens": 402770374.0, + "step": 15706 + }, + { + "epoch": 1.7249066549527785, + "grad_norm": 2.6960127353668213, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.722877562046051, + "num_tokens": 402789882.0, + "step": 15707 + }, + { + "epoch": 1.725016472655392, + "grad_norm": 2.3026492595672607, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7456624507904053, + "num_tokens": 402814535.0, + "step": 15708 + }, + { + "epoch": 1.7251262903580056, + "grad_norm": 2.405395746231079, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7346840500831604, + "num_tokens": 402837488.0, + "step": 15709 + }, + { + "epoch": 1.7252361080606193, + "grad_norm": 2.408876895904541, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7116529941558838, + "num_tokens": 402859689.0, + "step": 15710 + }, + { + "epoch": 1.725345925763233, + "grad_norm": 2.230570077896118, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7258679866790771, + "num_tokens": 402885322.0, + "step": 15711 + }, + { + "epoch": 1.7254557434658468, + "grad_norm": 2.468766212463379, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7402195930480957, + "num_tokens": 402906063.0, + "step": 15712 + }, + { + "epoch": 1.7255655611684604, + "grad_norm": 2.6196861267089844, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7229782938957214, + "num_tokens": 402926747.0, + "step": 15713 + }, + { + "epoch": 1.725675378871074, + "grad_norm": 2.185533046722412, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.710637629032135, + "num_tokens": 402953191.0, + "step": 15714 + }, + { + "epoch": 1.7257851965736877, + "grad_norm": 2.4884939193725586, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7260857820510864, + "num_tokens": 402974204.0, + "step": 15715 + }, + { + "epoch": 1.7258950142763014, + "grad_norm": 2.3199942111968994, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.715094268321991, + "num_tokens": 402998464.0, + "step": 15716 + }, + { + "epoch": 1.726004831978915, + "grad_norm": 2.56938099861145, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7548853158950806, + "num_tokens": 403017615.0, + "step": 15717 + }, + { + "epoch": 1.7261146496815285, + "grad_norm": 2.3381614685058594, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7494077682495117, + "num_tokens": 403039334.0, + "step": 15718 + }, + { + "epoch": 1.7262244673841423, + "grad_norm": 2.2701172828674316, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.715141236782074, + "num_tokens": 403062951.0, + "step": 15719 + }, + { + "epoch": 1.726334285086756, + "grad_norm": 2.18611741065979, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7104993462562561, + "num_tokens": 403089599.0, + "step": 15720 + }, + { + "epoch": 1.7264441027893698, + "grad_norm": 2.0845658779144287, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7161980867385864, + "num_tokens": 403118858.0, + "step": 15721 + }, + { + "epoch": 1.7265539204919833, + "grad_norm": 2.290072441101074, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7126864194869995, + "num_tokens": 403144031.0, + "step": 15722 + }, + { + "epoch": 1.7266637381945968, + "grad_norm": 2.1909737586975098, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7126936316490173, + "num_tokens": 403172854.0, + "step": 15723 + }, + { + "epoch": 1.7267735558972106, + "grad_norm": 2.348785400390625, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7401390075683594, + "num_tokens": 403195159.0, + "step": 15724 + }, + { + "epoch": 1.7268833735998244, + "grad_norm": 2.3648297786712646, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7127910852432251, + "num_tokens": 403219435.0, + "step": 15725 + }, + { + "epoch": 1.7269931913024381, + "grad_norm": 2.083024263381958, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7302905321121216, + "num_tokens": 403246598.0, + "step": 15726 + }, + { + "epoch": 1.7271030090050516, + "grad_norm": 2.4020230770111084, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7367299795150757, + "num_tokens": 403269178.0, + "step": 15727 + }, + { + "epoch": 1.7272128267076652, + "grad_norm": 2.113162040710449, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7194581031799316, + "num_tokens": 403297657.0, + "step": 15728 + }, + { + "epoch": 1.727322644410279, + "grad_norm": 2.3041574954986572, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7202878594398499, + "num_tokens": 403321810.0, + "step": 15729 + }, + { + "epoch": 1.7274324621128927, + "grad_norm": 2.38409161567688, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7194170355796814, + "num_tokens": 403345612.0, + "step": 15730 + }, + { + "epoch": 1.7275422798155062, + "grad_norm": 2.586155891418457, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7142788171768188, + "num_tokens": 403366017.0, + "step": 15731 + }, + { + "epoch": 1.7276520975181198, + "grad_norm": 2.2171285152435303, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7005444765090942, + "num_tokens": 403394067.0, + "step": 15732 + }, + { + "epoch": 1.7277619152207335, + "grad_norm": 2.0654900074005127, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6952397227287292, + "num_tokens": 403423762.0, + "step": 15733 + }, + { + "epoch": 1.7278717329233473, + "grad_norm": 1.9705328941345215, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7317230701446533, + "num_tokens": 403454463.0, + "step": 15734 + }, + { + "epoch": 1.727981550625961, + "grad_norm": 2.2811238765716553, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7246649861335754, + "num_tokens": 403478324.0, + "step": 15735 + }, + { + "epoch": 1.7280913683285746, + "grad_norm": 2.042022466659546, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7084743976593018, + "num_tokens": 403508214.0, + "step": 15736 + }, + { + "epoch": 1.728201186031188, + "grad_norm": 2.176090955734253, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6868942379951477, + "num_tokens": 403539218.0, + "step": 15737 + }, + { + "epoch": 1.7283110037338019, + "grad_norm": 2.126680612564087, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.6997340321540833, + "num_tokens": 403569666.0, + "step": 15738 + }, + { + "epoch": 1.7284208214364156, + "grad_norm": 2.1861789226531982, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6950994729995728, + "num_tokens": 403596348.0, + "step": 15739 + }, + { + "epoch": 1.7285306391390292, + "grad_norm": 2.268325090408325, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7105408906936646, + "num_tokens": 403621473.0, + "step": 15740 + }, + { + "epoch": 1.728640456841643, + "grad_norm": 2.1196889877319336, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7141481637954712, + "num_tokens": 403649741.0, + "step": 15741 + }, + { + "epoch": 1.7287502745442564, + "grad_norm": 2.385615348815918, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6966061592102051, + "num_tokens": 403673397.0, + "step": 15742 + }, + { + "epoch": 1.7288600922468702, + "grad_norm": 1.912608027458191, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6981930136680603, + "num_tokens": 403706460.0, + "step": 15743 + }, + { + "epoch": 1.728969909949484, + "grad_norm": 2.0036275386810303, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6946061849594116, + "num_tokens": 403740205.0, + "step": 15744 + }, + { + "epoch": 1.7290797276520975, + "grad_norm": 2.1203010082244873, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7157598733901978, + "num_tokens": 403770532.0, + "step": 15745 + }, + { + "epoch": 1.729189545354711, + "grad_norm": 2.2034788131713867, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7213565111160278, + "num_tokens": 403796336.0, + "step": 15746 + }, + { + "epoch": 1.7292993630573248, + "grad_norm": 2.216952085494995, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7328380942344666, + "num_tokens": 403820862.0, + "step": 15747 + }, + { + "epoch": 1.7294091807599385, + "grad_norm": 2.0695321559906006, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7036596536636353, + "num_tokens": 403848588.0, + "step": 15748 + }, + { + "epoch": 1.7295189984625523, + "grad_norm": 1.9393190145492554, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7059240341186523, + "num_tokens": 403880270.0, + "step": 15749 + }, + { + "epoch": 1.7296288161651658, + "grad_norm": 2.135774850845337, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.718749463558197, + "num_tokens": 403907796.0, + "step": 15750 + }, + { + "epoch": 1.7297386338677794, + "grad_norm": 2.103166103363037, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7116835117340088, + "num_tokens": 403936775.0, + "step": 15751 + }, + { + "epoch": 1.7298484515703931, + "grad_norm": 1.9919438362121582, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6932096481323242, + "num_tokens": 403969626.0, + "step": 15752 + }, + { + "epoch": 1.7299582692730069, + "grad_norm": 2.0612282752990723, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.6994996070861816, + "num_tokens": 403999556.0, + "step": 15753 + }, + { + "epoch": 1.7300680869756204, + "grad_norm": 2.103053331375122, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7329608201980591, + "num_tokens": 404028141.0, + "step": 15754 + }, + { + "epoch": 1.7301779046782342, + "grad_norm": 1.894437313079834, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6905796527862549, + "num_tokens": 404061388.0, + "step": 15755 + }, + { + "epoch": 1.7302877223808477, + "grad_norm": 2.1299307346343994, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7295856475830078, + "num_tokens": 404087875.0, + "step": 15756 + }, + { + "epoch": 1.7303975400834615, + "grad_norm": 2.263211250305176, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.724075973033905, + "num_tokens": 404112700.0, + "step": 15757 + }, + { + "epoch": 1.7305073577860752, + "grad_norm": 2.2516162395477295, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7200144529342651, + "num_tokens": 404139639.0, + "step": 15758 + }, + { + "epoch": 1.7306171754886888, + "grad_norm": 2.032005548477173, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6885499358177185, + "num_tokens": 404172199.0, + "step": 15759 + }, + { + "epoch": 1.7307269931913023, + "grad_norm": 2.352046251296997, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7271294593811035, + "num_tokens": 404195050.0, + "step": 15760 + }, + { + "epoch": 1.730836810893916, + "grad_norm": 2.173187255859375, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7086322903633118, + "num_tokens": 404224637.0, + "step": 15761 + }, + { + "epoch": 1.7309466285965298, + "grad_norm": 2.202195882797241, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7217464447021484, + "num_tokens": 404252679.0, + "step": 15762 + }, + { + "epoch": 1.7310564462991436, + "grad_norm": 2.6357905864715576, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7476084232330322, + "num_tokens": 404271786.0, + "step": 15763 + }, + { + "epoch": 1.731166264001757, + "grad_norm": 2.1004481315612793, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7121680974960327, + "num_tokens": 404301454.0, + "step": 15764 + }, + { + "epoch": 1.7312760817043706, + "grad_norm": 2.0695385932922363, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.6995878219604492, + "num_tokens": 404331956.0, + "step": 15765 + }, + { + "epoch": 1.7313858994069844, + "grad_norm": 2.374497890472412, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7415815591812134, + "num_tokens": 404355771.0, + "step": 15766 + }, + { + "epoch": 1.7314957171095982, + "grad_norm": 2.0893049240112305, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7380443811416626, + "num_tokens": 404381898.0, + "step": 15767 + }, + { + "epoch": 1.7316055348122117, + "grad_norm": 2.064802646636963, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7172380685806274, + "num_tokens": 404415553.0, + "step": 15768 + }, + { + "epoch": 1.7317153525148252, + "grad_norm": 1.9566447734832764, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7041700482368469, + "num_tokens": 404449707.0, + "step": 15769 + }, + { + "epoch": 1.731825170217439, + "grad_norm": 2.4645705223083496, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7089595794677734, + "num_tokens": 404472739.0, + "step": 15770 + }, + { + "epoch": 1.7319349879200527, + "grad_norm": 2.3146169185638428, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7335002422332764, + "num_tokens": 404496632.0, + "step": 15771 + }, + { + "epoch": 1.7320448056226665, + "grad_norm": 2.284893274307251, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7526782155036926, + "num_tokens": 404521099.0, + "step": 15772 + }, + { + "epoch": 1.73215462332528, + "grad_norm": 2.0513675212860107, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7128133773803711, + "num_tokens": 404550297.0, + "step": 15773 + }, + { + "epoch": 1.7322644410278936, + "grad_norm": 2.0205280780792236, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7210309505462646, + "num_tokens": 404582335.0, + "step": 15774 + }, + { + "epoch": 1.7323742587305073, + "grad_norm": 2.3270106315612793, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7046271562576294, + "num_tokens": 404606802.0, + "step": 15775 + }, + { + "epoch": 1.732484076433121, + "grad_norm": 2.2191474437713623, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7434230446815491, + "num_tokens": 404632106.0, + "step": 15776 + }, + { + "epoch": 1.7325938941357348, + "grad_norm": 2.0161054134368896, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.731134295463562, + "num_tokens": 404661229.0, + "step": 15777 + }, + { + "epoch": 1.7327037118383484, + "grad_norm": 2.3801088333129883, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7042787075042725, + "num_tokens": 404685114.0, + "step": 15778 + }, + { + "epoch": 1.732813529540962, + "grad_norm": 2.594045877456665, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7238751649856567, + "num_tokens": 404705485.0, + "step": 15779 + }, + { + "epoch": 1.7329233472435757, + "grad_norm": 2.4374582767486572, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7130680084228516, + "num_tokens": 404728318.0, + "step": 15780 + }, + { + "epoch": 1.7330331649461894, + "grad_norm": 2.1210670471191406, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7205750942230225, + "num_tokens": 404754963.0, + "step": 15781 + }, + { + "epoch": 1.733142982648803, + "grad_norm": 2.22483229637146, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.723439633846283, + "num_tokens": 404780446.0, + "step": 15782 + }, + { + "epoch": 1.7332528003514165, + "grad_norm": 2.2803757190704346, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7370097637176514, + "num_tokens": 404804017.0, + "step": 15783 + }, + { + "epoch": 1.7333626180540302, + "grad_norm": 2.6114604473114014, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7432781457901001, + "num_tokens": 404822521.0, + "step": 15784 + }, + { + "epoch": 1.733472435756644, + "grad_norm": 2.144418239593506, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7321856617927551, + "num_tokens": 404848432.0, + "step": 15785 + }, + { + "epoch": 1.7335822534592578, + "grad_norm": 2.2398486137390137, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7586496472358704, + "num_tokens": 404871532.0, + "step": 15786 + }, + { + "epoch": 1.7336920711618713, + "grad_norm": 2.32023549079895, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7124197483062744, + "num_tokens": 404895359.0, + "step": 15787 + }, + { + "epoch": 1.7338018888644848, + "grad_norm": 1.9967796802520752, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6900536417961121, + "num_tokens": 404927653.0, + "step": 15788 + }, + { + "epoch": 1.7339117065670986, + "grad_norm": 2.429828405380249, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7310340404510498, + "num_tokens": 404949480.0, + "step": 15789 + }, + { + "epoch": 1.7340215242697123, + "grad_norm": 2.0084307193756104, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6966699957847595, + "num_tokens": 404982305.0, + "step": 15790 + }, + { + "epoch": 1.734131341972326, + "grad_norm": 2.3838396072387695, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.717005729675293, + "num_tokens": 405004619.0, + "step": 15791 + }, + { + "epoch": 1.7342411596749396, + "grad_norm": 2.196460247039795, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7093857526779175, + "num_tokens": 405030881.0, + "step": 15792 + }, + { + "epoch": 1.7343509773775532, + "grad_norm": 2.1109461784362793, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7251347303390503, + "num_tokens": 405059272.0, + "step": 15793 + }, + { + "epoch": 1.734460795080167, + "grad_norm": 2.343660354614258, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7126840353012085, + "num_tokens": 405083754.0, + "step": 15794 + }, + { + "epoch": 1.7345706127827807, + "grad_norm": 1.903327465057373, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7012302875518799, + "num_tokens": 405118779.0, + "step": 15795 + }, + { + "epoch": 1.7346804304853942, + "grad_norm": 2.273238182067871, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7201969623565674, + "num_tokens": 405144688.0, + "step": 15796 + }, + { + "epoch": 1.7347902481880078, + "grad_norm": 2.5353024005889893, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7163965106010437, + "num_tokens": 405165481.0, + "step": 15797 + }, + { + "epoch": 1.7349000658906215, + "grad_norm": 2.15450119972229, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7311651706695557, + "num_tokens": 405192449.0, + "step": 15798 + }, + { + "epoch": 1.7350098835932353, + "grad_norm": 2.7438507080078125, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.712256669998169, + "num_tokens": 405211477.0, + "step": 15799 + }, + { + "epoch": 1.735119701295849, + "grad_norm": 2.452500581741333, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.746081531047821, + "num_tokens": 405231777.0, + "step": 15800 + }, + { + "epoch": 1.7352295189984626, + "grad_norm": 1.9927016496658325, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7320161461830139, + "num_tokens": 405261022.0, + "step": 15801 + }, + { + "epoch": 1.735339336701076, + "grad_norm": 2.072384834289551, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7241638898849487, + "num_tokens": 405290107.0, + "step": 15802 + }, + { + "epoch": 1.7354491544036899, + "grad_norm": 2.1140096187591553, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7315812706947327, + "num_tokens": 405318055.0, + "step": 15803 + }, + { + "epoch": 1.7355589721063036, + "grad_norm": 2.1486494541168213, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7263500094413757, + "num_tokens": 405345186.0, + "step": 15804 + }, + { + "epoch": 1.7356687898089171, + "grad_norm": 2.1583147048950195, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7039801478385925, + "num_tokens": 405372750.0, + "step": 15805 + }, + { + "epoch": 1.735778607511531, + "grad_norm": 2.3943517208099365, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7155013084411621, + "num_tokens": 405396194.0, + "step": 15806 + }, + { + "epoch": 1.7358884252141444, + "grad_norm": 2.8607351779937744, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7403579950332642, + "num_tokens": 405412604.0, + "step": 15807 + }, + { + "epoch": 1.7359982429167582, + "grad_norm": 2.469179630279541, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7316135168075562, + "num_tokens": 405434900.0, + "step": 15808 + }, + { + "epoch": 1.736108060619372, + "grad_norm": 2.2198915481567383, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7127012014389038, + "num_tokens": 405460479.0, + "step": 15809 + }, + { + "epoch": 1.7362178783219855, + "grad_norm": 2.0881876945495605, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7247045040130615, + "num_tokens": 405488498.0, + "step": 15810 + }, + { + "epoch": 1.736327696024599, + "grad_norm": 2.0027711391448975, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7225922346115112, + "num_tokens": 405520008.0, + "step": 15811 + }, + { + "epoch": 1.7364375137272128, + "grad_norm": 2.104562997817993, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7184871435165405, + "num_tokens": 405548685.0, + "step": 15812 + }, + { + "epoch": 1.7365473314298265, + "grad_norm": 2.170543909072876, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.71246337890625, + "num_tokens": 405577740.0, + "step": 15813 + }, + { + "epoch": 1.7366571491324403, + "grad_norm": 2.0561907291412354, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.698067307472229, + "num_tokens": 405607661.0, + "step": 15814 + }, + { + "epoch": 1.7367669668350538, + "grad_norm": 2.051610231399536, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7267897129058838, + "num_tokens": 405636250.0, + "step": 15815 + }, + { + "epoch": 1.7368767845376674, + "grad_norm": 2.394953489303589, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7178705930709839, + "num_tokens": 405660292.0, + "step": 15816 + }, + { + "epoch": 1.7369866022402811, + "grad_norm": 2.2833781242370605, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7230994701385498, + "num_tokens": 405685004.0, + "step": 15817 + }, + { + "epoch": 1.7370964199428949, + "grad_norm": 2.333717107772827, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7062843441963196, + "num_tokens": 405707929.0, + "step": 15818 + }, + { + "epoch": 1.7372062376455084, + "grad_norm": 2.5523853302001953, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7294843196868896, + "num_tokens": 405727615.0, + "step": 15819 + }, + { + "epoch": 1.7373160553481222, + "grad_norm": 2.0424489974975586, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6953638792037964, + "num_tokens": 405760584.0, + "step": 15820 + }, + { + "epoch": 1.7374258730507357, + "grad_norm": 2.469322681427002, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7160593867301941, + "num_tokens": 405781691.0, + "step": 15821 + }, + { + "epoch": 1.7375356907533495, + "grad_norm": 2.005138397216797, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7208898067474365, + "num_tokens": 405814090.0, + "step": 15822 + }, + { + "epoch": 1.7376455084559632, + "grad_norm": 2.244189739227295, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7425706386566162, + "num_tokens": 405839366.0, + "step": 15823 + }, + { + "epoch": 1.7377553261585768, + "grad_norm": 2.280465602874756, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7131287455558777, + "num_tokens": 405864876.0, + "step": 15824 + }, + { + "epoch": 1.7378651438611903, + "grad_norm": 2.45499849319458, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7038262486457825, + "num_tokens": 405890326.0, + "step": 15825 + }, + { + "epoch": 1.737974961563804, + "grad_norm": 2.0329551696777344, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7064136862754822, + "num_tokens": 405922544.0, + "step": 15826 + }, + { + "epoch": 1.7380847792664178, + "grad_norm": 2.1066372394561768, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.725804328918457, + "num_tokens": 405949814.0, + "step": 15827 + }, + { + "epoch": 1.7381945969690316, + "grad_norm": 2.322039842605591, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7200765609741211, + "num_tokens": 405975661.0, + "step": 15828 + }, + { + "epoch": 1.738304414671645, + "grad_norm": 2.1283681392669678, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7293179035186768, + "num_tokens": 406002859.0, + "step": 15829 + }, + { + "epoch": 1.7384142323742586, + "grad_norm": 1.979429006576538, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7205480337142944, + "num_tokens": 406032877.0, + "step": 15830 + }, + { + "epoch": 1.7385240500768724, + "grad_norm": 2.293025255203247, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7254067659378052, + "num_tokens": 406059252.0, + "step": 15831 + }, + { + "epoch": 1.7386338677794861, + "grad_norm": 2.257894277572632, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7156689167022705, + "num_tokens": 406085484.0, + "step": 15832 + }, + { + "epoch": 1.7387436854820997, + "grad_norm": 2.44468092918396, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7181365489959717, + "num_tokens": 406108535.0, + "step": 15833 + }, + { + "epoch": 1.7388535031847132, + "grad_norm": 2.523085117340088, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7359631657600403, + "num_tokens": 406129852.0, + "step": 15834 + }, + { + "epoch": 1.738963320887327, + "grad_norm": 2.2430851459503174, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7039929628372192, + "num_tokens": 406156597.0, + "step": 15835 + }, + { + "epoch": 1.7390731385899407, + "grad_norm": 2.1503396034240723, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7119315266609192, + "num_tokens": 406185337.0, + "step": 15836 + }, + { + "epoch": 1.7391829562925545, + "grad_norm": 2.164801597595215, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.709945023059845, + "num_tokens": 406213371.0, + "step": 15837 + }, + { + "epoch": 1.739292773995168, + "grad_norm": 2.3346259593963623, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7264569401741028, + "num_tokens": 406238496.0, + "step": 15838 + }, + { + "epoch": 1.7394025916977816, + "grad_norm": 2.4240260124206543, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7234470844268799, + "num_tokens": 406261180.0, + "step": 15839 + }, + { + "epoch": 1.7395124094003953, + "grad_norm": 2.1169676780700684, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6927894949913025, + "num_tokens": 406289001.0, + "step": 15840 + }, + { + "epoch": 1.739622227103009, + "grad_norm": 2.4342591762542725, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7317510843276978, + "num_tokens": 406311594.0, + "step": 15841 + }, + { + "epoch": 1.7397320448056228, + "grad_norm": 2.1845178604125977, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7254294753074646, + "num_tokens": 406337357.0, + "step": 15842 + }, + { + "epoch": 1.7398418625082364, + "grad_norm": 2.1308724880218506, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7142894268035889, + "num_tokens": 406365945.0, + "step": 15843 + }, + { + "epoch": 1.73995168021085, + "grad_norm": 2.3630623817443848, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7159859538078308, + "num_tokens": 406393094.0, + "step": 15844 + }, + { + "epoch": 1.7400614979134637, + "grad_norm": 1.9253580570220947, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7087920904159546, + "num_tokens": 406424438.0, + "step": 15845 + }, + { + "epoch": 1.7401713156160774, + "grad_norm": 2.3739540576934814, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7408790588378906, + "num_tokens": 406446739.0, + "step": 15846 + }, + { + "epoch": 1.740281133318691, + "grad_norm": 2.2488183975219727, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6928202509880066, + "num_tokens": 406472307.0, + "step": 15847 + }, + { + "epoch": 1.7403909510213045, + "grad_norm": 2.3054752349853516, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7192450165748596, + "num_tokens": 406497693.0, + "step": 15848 + }, + { + "epoch": 1.7405007687239182, + "grad_norm": 2.0156402587890625, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7616373896598816, + "num_tokens": 406525377.0, + "step": 15849 + }, + { + "epoch": 1.740610586426532, + "grad_norm": 2.1679553985595703, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7243359684944153, + "num_tokens": 406551672.0, + "step": 15850 + }, + { + "epoch": 1.7407204041291457, + "grad_norm": 2.044062614440918, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7180934548377991, + "num_tokens": 406582053.0, + "step": 15851 + }, + { + "epoch": 1.7408302218317593, + "grad_norm": 2.106292486190796, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7120428085327148, + "num_tokens": 406608176.0, + "step": 15852 + }, + { + "epoch": 1.7409400395343728, + "grad_norm": 2.5806117057800293, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7200555801391602, + "num_tokens": 406629708.0, + "step": 15853 + }, + { + "epoch": 1.7410498572369866, + "grad_norm": 2.0458576679229736, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7004496455192566, + "num_tokens": 406658198.0, + "step": 15854 + }, + { + "epoch": 1.7411596749396003, + "grad_norm": 2.25065016746521, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7336546182632446, + "num_tokens": 406682398.0, + "step": 15855 + }, + { + "epoch": 1.741269492642214, + "grad_norm": 2.4226834774017334, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.738277792930603, + "num_tokens": 406704047.0, + "step": 15856 + }, + { + "epoch": 1.7413793103448276, + "grad_norm": 2.186915636062622, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7152098417282104, + "num_tokens": 406730405.0, + "step": 15857 + }, + { + "epoch": 1.7414891280474412, + "grad_norm": 2.1250619888305664, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7033922672271729, + "num_tokens": 406759683.0, + "step": 15858 + }, + { + "epoch": 1.741598945750055, + "grad_norm": 2.4671225547790527, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7266131043434143, + "num_tokens": 406781512.0, + "step": 15859 + }, + { + "epoch": 1.7417087634526687, + "grad_norm": 2.1528947353363037, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7343003749847412, + "num_tokens": 406807741.0, + "step": 15860 + }, + { + "epoch": 1.7418185811552822, + "grad_norm": 2.3470566272735596, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7211179733276367, + "num_tokens": 406830709.0, + "step": 15861 + }, + { + "epoch": 1.7419283988578957, + "grad_norm": 2.9759347438812256, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7378036975860596, + "num_tokens": 406846971.0, + "step": 15862 + }, + { + "epoch": 1.7420382165605095, + "grad_norm": 2.123682737350464, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7211719751358032, + "num_tokens": 406874026.0, + "step": 15863 + }, + { + "epoch": 1.7421480342631233, + "grad_norm": 2.0019047260284424, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.715618371963501, + "num_tokens": 406905976.0, + "step": 15864 + }, + { + "epoch": 1.742257851965737, + "grad_norm": 2.3425662517547607, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7181276082992554, + "num_tokens": 406929274.0, + "step": 15865 + }, + { + "epoch": 1.7423676696683505, + "grad_norm": 2.2477598190307617, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7174212336540222, + "num_tokens": 406954611.0, + "step": 15866 + }, + { + "epoch": 1.742477487370964, + "grad_norm": 1.9365984201431274, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7056841254234314, + "num_tokens": 406988711.0, + "step": 15867 + }, + { + "epoch": 1.7425873050735778, + "grad_norm": 2.2191624641418457, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7276358008384705, + "num_tokens": 407012869.0, + "step": 15868 + }, + { + "epoch": 1.7426971227761916, + "grad_norm": 2.1287779808044434, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6761974096298218, + "num_tokens": 407043547.0, + "step": 15869 + }, + { + "epoch": 1.7428069404788051, + "grad_norm": 2.5188989639282227, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7302895784378052, + "num_tokens": 407065289.0, + "step": 15870 + }, + { + "epoch": 1.742916758181419, + "grad_norm": 1.990665316581726, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7320568561553955, + "num_tokens": 407094800.0, + "step": 15871 + }, + { + "epoch": 1.7430265758840324, + "grad_norm": 2.2567074298858643, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7228806614875793, + "num_tokens": 407119875.0, + "step": 15872 + }, + { + "epoch": 1.7431363935866462, + "grad_norm": 2.5011322498321533, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7096412777900696, + "num_tokens": 407142007.0, + "step": 15873 + }, + { + "epoch": 1.74324621128926, + "grad_norm": 2.441030740737915, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7340892553329468, + "num_tokens": 407164219.0, + "step": 15874 + }, + { + "epoch": 1.7433560289918735, + "grad_norm": 2.0291783809661865, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7466331720352173, + "num_tokens": 407192411.0, + "step": 15875 + }, + { + "epoch": 1.743465846694487, + "grad_norm": 2.3648242950439453, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7292182445526123, + "num_tokens": 407213720.0, + "step": 15876 + }, + { + "epoch": 1.7435756643971008, + "grad_norm": 2.196558952331543, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7050874829292297, + "num_tokens": 407242267.0, + "step": 15877 + }, + { + "epoch": 1.7436854820997145, + "grad_norm": 2.367488145828247, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7206841111183167, + "num_tokens": 407265494.0, + "step": 15878 + }, + { + "epoch": 1.7437952998023283, + "grad_norm": 2.379671573638916, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7185573577880859, + "num_tokens": 407289358.0, + "step": 15879 + }, + { + "epoch": 1.7439051175049418, + "grad_norm": 2.2192161083221436, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7132129073143005, + "num_tokens": 407315852.0, + "step": 15880 + }, + { + "epoch": 1.7440149352075554, + "grad_norm": 2.079749345779419, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7125310897827148, + "num_tokens": 407345935.0, + "step": 15881 + }, + { + "epoch": 1.744124752910169, + "grad_norm": 2.3260364532470703, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7097499370574951, + "num_tokens": 407371035.0, + "step": 15882 + }, + { + "epoch": 1.7442345706127829, + "grad_norm": 2.3854358196258545, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7091461420059204, + "num_tokens": 407394590.0, + "step": 15883 + }, + { + "epoch": 1.7443443883153964, + "grad_norm": 2.130275011062622, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7095335721969604, + "num_tokens": 407422131.0, + "step": 15884 + }, + { + "epoch": 1.7444542060180102, + "grad_norm": 2.634273052215576, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.743142306804657, + "num_tokens": 407440302.0, + "step": 15885 + }, + { + "epoch": 1.7445640237206237, + "grad_norm": 2.529266119003296, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7213941812515259, + "num_tokens": 407462388.0, + "step": 15886 + }, + { + "epoch": 1.7446738414232374, + "grad_norm": 2.3130362033843994, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7104072570800781, + "num_tokens": 407487566.0, + "step": 15887 + }, + { + "epoch": 1.7447836591258512, + "grad_norm": 2.184195041656494, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7340081930160522, + "num_tokens": 407515504.0, + "step": 15888 + }, + { + "epoch": 1.7448934768284647, + "grad_norm": 2.3575379848480225, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7361825704574585, + "num_tokens": 407538387.0, + "step": 15889 + }, + { + "epoch": 1.7450032945310783, + "grad_norm": 2.2416388988494873, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7457629442214966, + "num_tokens": 407562615.0, + "step": 15890 + }, + { + "epoch": 1.745113112233692, + "grad_norm": 2.2199480533599854, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7146767377853394, + "num_tokens": 407590985.0, + "step": 15891 + }, + { + "epoch": 1.7452229299363058, + "grad_norm": 2.396613597869873, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7113972306251526, + "num_tokens": 407614939.0, + "step": 15892 + }, + { + "epoch": 1.7453327476389195, + "grad_norm": 2.086822032928467, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6977578401565552, + "num_tokens": 407645404.0, + "step": 15893 + }, + { + "epoch": 1.745442565341533, + "grad_norm": 2.3181869983673096, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.709357738494873, + "num_tokens": 407671005.0, + "step": 15894 + }, + { + "epoch": 1.7455523830441466, + "grad_norm": 2.1864819526672363, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7220057249069214, + "num_tokens": 407697492.0, + "step": 15895 + }, + { + "epoch": 1.7456622007467604, + "grad_norm": 2.163893699645996, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7053521871566772, + "num_tokens": 407725995.0, + "step": 15896 + }, + { + "epoch": 1.7457720184493741, + "grad_norm": 2.1447653770446777, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7460367679595947, + "num_tokens": 407751530.0, + "step": 15897 + }, + { + "epoch": 1.7458818361519877, + "grad_norm": 2.3154335021972656, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7026278972625732, + "num_tokens": 407775095.0, + "step": 15898 + }, + { + "epoch": 1.7459916538546012, + "grad_norm": 2.1142916679382324, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7135576009750366, + "num_tokens": 407805143.0, + "step": 15899 + }, + { + "epoch": 1.746101471557215, + "grad_norm": 2.0857741832733154, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7022262811660767, + "num_tokens": 407838162.0, + "step": 15900 + }, + { + "epoch": 1.7462112892598287, + "grad_norm": 2.313826560974121, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.733235776424408, + "num_tokens": 407860905.0, + "step": 15901 + }, + { + "epoch": 1.7463211069624425, + "grad_norm": 2.186441659927368, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7108659744262695, + "num_tokens": 407887652.0, + "step": 15902 + }, + { + "epoch": 1.746430924665056, + "grad_norm": 2.0563104152679443, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7081639766693115, + "num_tokens": 407917232.0, + "step": 15903 + }, + { + "epoch": 1.7465407423676695, + "grad_norm": 2.170419216156006, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7229197025299072, + "num_tokens": 407944580.0, + "step": 15904 + }, + { + "epoch": 1.7466505600702833, + "grad_norm": 2.4832398891448975, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.735640823841095, + "num_tokens": 407965308.0, + "step": 15905 + }, + { + "epoch": 1.746760377772897, + "grad_norm": 2.2182669639587402, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7109633684158325, + "num_tokens": 407993373.0, + "step": 15906 + }, + { + "epoch": 1.7468701954755108, + "grad_norm": 2.3978841304779053, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7131447792053223, + "num_tokens": 408018262.0, + "step": 15907 + }, + { + "epoch": 1.7469800131781243, + "grad_norm": 2.298740863800049, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7268123626708984, + "num_tokens": 408041238.0, + "step": 15908 + }, + { + "epoch": 1.7470898308807379, + "grad_norm": 2.719557046890259, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7282467484474182, + "num_tokens": 408059157.0, + "step": 15909 + }, + { + "epoch": 1.7471996485833516, + "grad_norm": 2.59773588180542, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7360971570014954, + "num_tokens": 408077922.0, + "step": 15910 + }, + { + "epoch": 1.7473094662859654, + "grad_norm": 1.9575457572937012, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7102575898170471, + "num_tokens": 408110424.0, + "step": 15911 + }, + { + "epoch": 1.747419283988579, + "grad_norm": 2.1112022399902344, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7272236347198486, + "num_tokens": 408135966.0, + "step": 15912 + }, + { + "epoch": 1.7475291016911925, + "grad_norm": 2.2248616218566895, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7003591656684875, + "num_tokens": 408162063.0, + "step": 15913 + }, + { + "epoch": 1.7476389193938062, + "grad_norm": 2.3099846839904785, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7001163959503174, + "num_tokens": 408187759.0, + "step": 15914 + }, + { + "epoch": 1.74774873709642, + "grad_norm": 2.010711908340454, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.70087069272995, + "num_tokens": 408221080.0, + "step": 15915 + }, + { + "epoch": 1.7478585547990337, + "grad_norm": 2.044365882873535, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7047407031059265, + "num_tokens": 408249989.0, + "step": 15916 + }, + { + "epoch": 1.7479683725016473, + "grad_norm": 2.2406952381134033, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7254106402397156, + "num_tokens": 408274591.0, + "step": 15917 + }, + { + "epoch": 1.7480781902042608, + "grad_norm": 2.2988829612731934, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7230141162872314, + "num_tokens": 408300614.0, + "step": 15918 + }, + { + "epoch": 1.7481880079068746, + "grad_norm": 2.038313388824463, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7262750864028931, + "num_tokens": 408330719.0, + "step": 15919 + }, + { + "epoch": 1.7482978256094883, + "grad_norm": 2.0222949981689453, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7168912291526794, + "num_tokens": 408358774.0, + "step": 15920 + }, + { + "epoch": 1.7484076433121019, + "grad_norm": 2.2918643951416016, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7360948324203491, + "num_tokens": 408383302.0, + "step": 15921 + }, + { + "epoch": 1.7485174610147156, + "grad_norm": 2.1255128383636475, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7235894203186035, + "num_tokens": 408409477.0, + "step": 15922 + }, + { + "epoch": 1.7486272787173291, + "grad_norm": 2.579540967941284, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7117787599563599, + "num_tokens": 408430569.0, + "step": 15923 + }, + { + "epoch": 1.748737096419943, + "grad_norm": 2.579000234603882, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7036479711532593, + "num_tokens": 408450771.0, + "step": 15924 + }, + { + "epoch": 1.7488469141225567, + "grad_norm": 2.213857889175415, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6964535713195801, + "num_tokens": 408479019.0, + "step": 15925 + }, + { + "epoch": 1.7489567318251702, + "grad_norm": 2.0537660121917725, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7150080800056458, + "num_tokens": 408508085.0, + "step": 15926 + }, + { + "epoch": 1.7490665495277837, + "grad_norm": 2.5144903659820557, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7239183187484741, + "num_tokens": 408530205.0, + "step": 15927 + }, + { + "epoch": 1.7491763672303975, + "grad_norm": 2.228425979614258, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7097735404968262, + "num_tokens": 408555297.0, + "step": 15928 + }, + { + "epoch": 1.7492861849330112, + "grad_norm": 2.156755208969116, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7197046279907227, + "num_tokens": 408584593.0, + "step": 15929 + }, + { + "epoch": 1.749396002635625, + "grad_norm": 2.278669595718384, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7075827121734619, + "num_tokens": 408610259.0, + "step": 15930 + }, + { + "epoch": 1.7495058203382385, + "grad_norm": 2.2701330184936523, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7179416418075562, + "num_tokens": 408635561.0, + "step": 15931 + }, + { + "epoch": 1.749615638040852, + "grad_norm": 2.197880983352661, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7126947641372681, + "num_tokens": 408662008.0, + "step": 15932 + }, + { + "epoch": 1.7497254557434658, + "grad_norm": 2.1247975826263428, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.703456461429596, + "num_tokens": 408691333.0, + "step": 15933 + }, + { + "epoch": 1.7498352734460796, + "grad_norm": 2.0431771278381348, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7032697200775146, + "num_tokens": 408722895.0, + "step": 15934 + }, + { + "epoch": 1.7499450911486931, + "grad_norm": 2.110550880432129, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7538577318191528, + "num_tokens": 408749347.0, + "step": 15935 + }, + { + "epoch": 1.7500549088513069, + "grad_norm": 2.10841703414917, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7150885462760925, + "num_tokens": 408777894.0, + "step": 15936 + }, + { + "epoch": 1.7501647265539204, + "grad_norm": 2.3513457775115967, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7056065797805786, + "num_tokens": 408801689.0, + "step": 15937 + }, + { + "epoch": 1.7502745442565342, + "grad_norm": 2.0176661014556885, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7029712796211243, + "num_tokens": 408831883.0, + "step": 15938 + }, + { + "epoch": 1.750384361959148, + "grad_norm": 2.037219762802124, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7480192184448242, + "num_tokens": 408859253.0, + "step": 15939 + }, + { + "epoch": 1.7504941796617615, + "grad_norm": 2.2672817707061768, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7284469604492188, + "num_tokens": 408884258.0, + "step": 15940 + }, + { + "epoch": 1.750603997364375, + "grad_norm": 2.1820969581604004, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7253580689430237, + "num_tokens": 408910454.0, + "step": 15941 + }, + { + "epoch": 1.7507138150669888, + "grad_norm": 2.4869155883789062, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7591426372528076, + "num_tokens": 408929655.0, + "step": 15942 + }, + { + "epoch": 1.7508236327696025, + "grad_norm": 2.3442881107330322, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.704399585723877, + "num_tokens": 408952649.0, + "step": 15943 + }, + { + "epoch": 1.7509334504722163, + "grad_norm": 2.025911808013916, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7100937366485596, + "num_tokens": 408982654.0, + "step": 15944 + }, + { + "epoch": 1.7510432681748298, + "grad_norm": 1.907073974609375, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7116893529891968, + "num_tokens": 409014463.0, + "step": 15945 + }, + { + "epoch": 1.7511530858774433, + "grad_norm": 2.440798282623291, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.731542706489563, + "num_tokens": 409036213.0, + "step": 15946 + }, + { + "epoch": 1.751262903580057, + "grad_norm": 2.541865825653076, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.742114782333374, + "num_tokens": 409056055.0, + "step": 15947 + }, + { + "epoch": 1.7513727212826709, + "grad_norm": 2.019808530807495, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7003157138824463, + "num_tokens": 409084678.0, + "step": 15948 + }, + { + "epoch": 1.7514825389852844, + "grad_norm": 2.2584147453308105, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7163133025169373, + "num_tokens": 409111074.0, + "step": 15949 + }, + { + "epoch": 1.7515923566878981, + "grad_norm": 2.292022943496704, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.702242374420166, + "num_tokens": 409138127.0, + "step": 15950 + }, + { + "epoch": 1.7517021743905117, + "grad_norm": 2.2532894611358643, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7109428644180298, + "num_tokens": 409164083.0, + "step": 15951 + }, + { + "epoch": 1.7518119920931254, + "grad_norm": 2.1955766677856445, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7341765761375427, + "num_tokens": 409188596.0, + "step": 15952 + }, + { + "epoch": 1.7519218097957392, + "grad_norm": 2.240302085876465, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6995435953140259, + "num_tokens": 409215698.0, + "step": 15953 + }, + { + "epoch": 1.7520316274983527, + "grad_norm": 2.392760992050171, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7257034778594971, + "num_tokens": 409238200.0, + "step": 15954 + }, + { + "epoch": 1.7521414452009663, + "grad_norm": 2.401048183441162, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7182924747467041, + "num_tokens": 409260602.0, + "step": 15955 + }, + { + "epoch": 1.75225126290358, + "grad_norm": 2.322324275970459, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.724206268787384, + "num_tokens": 409284624.0, + "step": 15956 + }, + { + "epoch": 1.7523610806061938, + "grad_norm": 2.0647377967834473, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7297090291976929, + "num_tokens": 409311870.0, + "step": 15957 + }, + { + "epoch": 1.7524708983088075, + "grad_norm": 2.2720344066619873, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7299544811248779, + "num_tokens": 409336835.0, + "step": 15958 + }, + { + "epoch": 1.752580716011421, + "grad_norm": 2.057464599609375, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6878281831741333, + "num_tokens": 409369198.0, + "step": 15959 + }, + { + "epoch": 1.7526905337140346, + "grad_norm": 2.2366957664489746, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7254922986030579, + "num_tokens": 409395392.0, + "step": 15960 + }, + { + "epoch": 1.7528003514166484, + "grad_norm": 2.419356346130371, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7257740497589111, + "num_tokens": 409416434.0, + "step": 15961 + }, + { + "epoch": 1.7529101691192621, + "grad_norm": 2.378082275390625, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6981748342514038, + "num_tokens": 409440257.0, + "step": 15962 + }, + { + "epoch": 1.7530199868218757, + "grad_norm": 1.930444359779358, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6867592334747314, + "num_tokens": 409472527.0, + "step": 15963 + }, + { + "epoch": 1.7531298045244892, + "grad_norm": 1.9738000631332397, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7004892230033875, + "num_tokens": 409506127.0, + "step": 15964 + }, + { + "epoch": 1.753239622227103, + "grad_norm": 2.5167133808135986, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7172287106513977, + "num_tokens": 409527131.0, + "step": 15965 + }, + { + "epoch": 1.7533494399297167, + "grad_norm": 2.2535040378570557, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.718768835067749, + "num_tokens": 409551861.0, + "step": 15966 + }, + { + "epoch": 1.7534592576323305, + "grad_norm": 2.0454180240631104, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7225342988967896, + "num_tokens": 409582397.0, + "step": 15967 + }, + { + "epoch": 1.753569075334944, + "grad_norm": 2.271512508392334, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7203638553619385, + "num_tokens": 409608585.0, + "step": 15968 + }, + { + "epoch": 1.7536788930375575, + "grad_norm": 2.1015360355377197, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7045970559120178, + "num_tokens": 409637585.0, + "step": 15969 + }, + { + "epoch": 1.7537887107401713, + "grad_norm": 2.253582239151001, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7225114107131958, + "num_tokens": 409662424.0, + "step": 15970 + }, + { + "epoch": 1.753898528442785, + "grad_norm": 2.3839902877807617, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.725526750087738, + "num_tokens": 409686593.0, + "step": 15971 + }, + { + "epoch": 1.7540083461453988, + "grad_norm": 2.472761869430542, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7193709015846252, + "num_tokens": 409709157.0, + "step": 15972 + }, + { + "epoch": 1.7541181638480123, + "grad_norm": 2.401130437850952, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7145081758499146, + "num_tokens": 409733696.0, + "step": 15973 + }, + { + "epoch": 1.7542279815506259, + "grad_norm": 2.4948489665985107, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7128708958625793, + "num_tokens": 409756521.0, + "step": 15974 + }, + { + "epoch": 1.7543377992532396, + "grad_norm": 2.0252721309661865, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7390018701553345, + "num_tokens": 409785063.0, + "step": 15975 + }, + { + "epoch": 1.7544476169558534, + "grad_norm": 2.40554141998291, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7069168090820312, + "num_tokens": 409809278.0, + "step": 15976 + }, + { + "epoch": 1.754557434658467, + "grad_norm": 2.243947744369507, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7234739065170288, + "num_tokens": 409833752.0, + "step": 15977 + }, + { + "epoch": 1.7546672523610805, + "grad_norm": 2.0022637844085693, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7011620402336121, + "num_tokens": 409864986.0, + "step": 15978 + }, + { + "epoch": 1.7547770700636942, + "grad_norm": 2.1462409496307373, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7146125435829163, + "num_tokens": 409891852.0, + "step": 15979 + }, + { + "epoch": 1.754886887766308, + "grad_norm": 2.286681652069092, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7292042374610901, + "num_tokens": 409915180.0, + "step": 15980 + }, + { + "epoch": 1.7549967054689217, + "grad_norm": 2.553748369216919, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7403042316436768, + "num_tokens": 409936079.0, + "step": 15981 + }, + { + "epoch": 1.7551065231715353, + "grad_norm": 2.36203932762146, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7262138724327087, + "num_tokens": 409960029.0, + "step": 15982 + }, + { + "epoch": 1.7552163408741488, + "grad_norm": 2.2869679927825928, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7260496616363525, + "num_tokens": 409984851.0, + "step": 15983 + }, + { + "epoch": 1.7553261585767626, + "grad_norm": 2.004945755004883, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7339403629302979, + "num_tokens": 410013915.0, + "step": 15984 + }, + { + "epoch": 1.7554359762793763, + "grad_norm": 2.089134454727173, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6967889070510864, + "num_tokens": 410046019.0, + "step": 15985 + }, + { + "epoch": 1.7555457939819898, + "grad_norm": 2.7250540256500244, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7367362380027771, + "num_tokens": 410065690.0, + "step": 15986 + }, + { + "epoch": 1.7556556116846036, + "grad_norm": 1.8542790412902832, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7059762477874756, + "num_tokens": 410098873.0, + "step": 15987 + }, + { + "epoch": 1.7557654293872171, + "grad_norm": 1.9909007549285889, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7058966159820557, + "num_tokens": 410131206.0, + "step": 15988 + }, + { + "epoch": 1.755875247089831, + "grad_norm": 2.005697250366211, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7311680316925049, + "num_tokens": 410160889.0, + "step": 15989 + }, + { + "epoch": 1.7559850647924446, + "grad_norm": 2.244842767715454, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6894908547401428, + "num_tokens": 410188951.0, + "step": 15990 + }, + { + "epoch": 1.7560948824950582, + "grad_norm": 2.595449686050415, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7246579527854919, + "num_tokens": 410210121.0, + "step": 15991 + }, + { + "epoch": 1.7562047001976717, + "grad_norm": 2.1943154335021973, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7111767530441284, + "num_tokens": 410236950.0, + "step": 15992 + }, + { + "epoch": 1.7563145179002855, + "grad_norm": 2.3364474773406982, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7285406589508057, + "num_tokens": 410259763.0, + "step": 15993 + }, + { + "epoch": 1.7564243356028992, + "grad_norm": 2.312685966491699, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7329668998718262, + "num_tokens": 410284720.0, + "step": 15994 + }, + { + "epoch": 1.756534153305513, + "grad_norm": 2.314147472381592, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7408769130706787, + "num_tokens": 410308683.0, + "step": 15995 + }, + { + "epoch": 1.7566439710081265, + "grad_norm": 2.1838579177856445, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7106033563613892, + "num_tokens": 410333134.0, + "step": 15996 + }, + { + "epoch": 1.75675378871074, + "grad_norm": 2.460756301879883, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7289433479309082, + "num_tokens": 410354470.0, + "step": 15997 + }, + { + "epoch": 1.7568636064133538, + "grad_norm": 2.3718478679656982, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7364715337753296, + "num_tokens": 410377234.0, + "step": 15998 + }, + { + "epoch": 1.7569734241159676, + "grad_norm": 2.4829609394073486, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7382171154022217, + "num_tokens": 410397862.0, + "step": 15999 + }, + { + "epoch": 1.757083241818581, + "grad_norm": 2.38191294670105, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7380140423774719, + "num_tokens": 410420679.0, + "step": 16000 + }, + { + "epoch": 1.7571930595211949, + "grad_norm": 2.0392985343933105, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6875613927841187, + "num_tokens": 410451613.0, + "step": 16001 + }, + { + "epoch": 1.7573028772238084, + "grad_norm": 2.4795632362365723, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7224549651145935, + "num_tokens": 410473588.0, + "step": 16002 + }, + { + "epoch": 1.7574126949264222, + "grad_norm": 2.2231850624084473, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7216547727584839, + "num_tokens": 410500163.0, + "step": 16003 + }, + { + "epoch": 1.757522512629036, + "grad_norm": 2.1040902137756348, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7083241939544678, + "num_tokens": 410528282.0, + "step": 16004 + }, + { + "epoch": 1.7576323303316495, + "grad_norm": 2.085104465484619, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7202765345573425, + "num_tokens": 410556550.0, + "step": 16005 + }, + { + "epoch": 1.757742148034263, + "grad_norm": 2.197124719619751, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7237746715545654, + "num_tokens": 410582850.0, + "step": 16006 + }, + { + "epoch": 1.7578519657368767, + "grad_norm": 2.53694748878479, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7230172157287598, + "num_tokens": 410605753.0, + "step": 16007 + }, + { + "epoch": 1.7579617834394905, + "grad_norm": 2.3168299198150635, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7119618654251099, + "num_tokens": 410629444.0, + "step": 16008 + }, + { + "epoch": 1.7580716011421043, + "grad_norm": 2.326988935470581, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7032845616340637, + "num_tokens": 410654403.0, + "step": 16009 + }, + { + "epoch": 1.7581814188447178, + "grad_norm": 2.5324902534484863, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7079156637191772, + "num_tokens": 410675495.0, + "step": 16010 + }, + { + "epoch": 1.7582912365473313, + "grad_norm": 2.4357028007507324, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7317796945571899, + "num_tokens": 410697449.0, + "step": 16011 + }, + { + "epoch": 1.758401054249945, + "grad_norm": 2.1131343841552734, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6964734196662903, + "num_tokens": 410727602.0, + "step": 16012 + }, + { + "epoch": 1.7585108719525588, + "grad_norm": 2.4090490341186523, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7288854122161865, + "num_tokens": 410749732.0, + "step": 16013 + }, + { + "epoch": 1.7586206896551724, + "grad_norm": 2.0578930377960205, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.745514452457428, + "num_tokens": 410777852.0, + "step": 16014 + }, + { + "epoch": 1.758730507357786, + "grad_norm": 2.0538296699523926, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7088174819946289, + "num_tokens": 410806882.0, + "step": 16015 + }, + { + "epoch": 1.7588403250603997, + "grad_norm": 2.146211862564087, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7073272466659546, + "num_tokens": 410834154.0, + "step": 16016 + }, + { + "epoch": 1.7589501427630134, + "grad_norm": 2.0373339653015137, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7151761054992676, + "num_tokens": 410865102.0, + "step": 16017 + }, + { + "epoch": 1.7590599604656272, + "grad_norm": 2.159123420715332, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7002894878387451, + "num_tokens": 410893430.0, + "step": 16018 + }, + { + "epoch": 1.7591697781682407, + "grad_norm": 2.607465982437134, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.740296483039856, + "num_tokens": 410913343.0, + "step": 16019 + }, + { + "epoch": 1.7592795958708543, + "grad_norm": 2.512167453765869, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7275794148445129, + "num_tokens": 410934924.0, + "step": 16020 + }, + { + "epoch": 1.759389413573468, + "grad_norm": 2.512430191040039, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7358905076980591, + "num_tokens": 410953279.0, + "step": 16021 + }, + { + "epoch": 1.7594992312760818, + "grad_norm": 2.2378628253936768, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7118539214134216, + "num_tokens": 410977051.0, + "step": 16022 + }, + { + "epoch": 1.7596090489786955, + "grad_norm": 2.064481258392334, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.712852954864502, + "num_tokens": 411007202.0, + "step": 16023 + }, + { + "epoch": 1.759718866681309, + "grad_norm": 2.1258840560913086, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7433992624282837, + "num_tokens": 411033279.0, + "step": 16024 + }, + { + "epoch": 1.7598286843839226, + "grad_norm": 2.2314765453338623, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7029027938842773, + "num_tokens": 411061091.0, + "step": 16025 + }, + { + "epoch": 1.7599385020865363, + "grad_norm": 2.1839654445648193, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7113693952560425, + "num_tokens": 411087352.0, + "step": 16026 + }, + { + "epoch": 1.76004831978915, + "grad_norm": 2.679718017578125, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7198008298873901, + "num_tokens": 411105716.0, + "step": 16027 + }, + { + "epoch": 1.7601581374917636, + "grad_norm": 1.8672548532485962, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.6975652575492859, + "num_tokens": 411140530.0, + "step": 16028 + }, + { + "epoch": 1.7602679551943772, + "grad_norm": 2.136291980743408, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7150442004203796, + "num_tokens": 411170925.0, + "step": 16029 + }, + { + "epoch": 1.760377772896991, + "grad_norm": 2.1304566860198975, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7168857455253601, + "num_tokens": 411199180.0, + "step": 16030 + }, + { + "epoch": 1.7604875905996047, + "grad_norm": 2.0200858116149902, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7272955775260925, + "num_tokens": 411228600.0, + "step": 16031 + }, + { + "epoch": 1.7605974083022184, + "grad_norm": 2.2605676651000977, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7281308174133301, + "num_tokens": 411254083.0, + "step": 16032 + }, + { + "epoch": 1.760707226004832, + "grad_norm": 2.2430872917175293, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7243514060974121, + "num_tokens": 411279750.0, + "step": 16033 + }, + { + "epoch": 1.7608170437074455, + "grad_norm": 2.1511387825012207, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.714634120464325, + "num_tokens": 411308494.0, + "step": 16034 + }, + { + "epoch": 1.7609268614100593, + "grad_norm": 2.0849106311798096, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7385241985321045, + "num_tokens": 411336524.0, + "step": 16035 + }, + { + "epoch": 1.761036679112673, + "grad_norm": 2.2608144283294678, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7220062017440796, + "num_tokens": 411362490.0, + "step": 16036 + }, + { + "epoch": 1.7611464968152868, + "grad_norm": 2.117582082748413, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.701479971408844, + "num_tokens": 411392609.0, + "step": 16037 + }, + { + "epoch": 1.7612563145179003, + "grad_norm": 2.1740849018096924, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7178271412849426, + "num_tokens": 411420423.0, + "step": 16038 + }, + { + "epoch": 1.7613661322205139, + "grad_norm": 2.1235129833221436, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7085494995117188, + "num_tokens": 411447480.0, + "step": 16039 + }, + { + "epoch": 1.7614759499231276, + "grad_norm": 2.305128335952759, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7008990049362183, + "num_tokens": 411471134.0, + "step": 16040 + }, + { + "epoch": 1.7615857676257414, + "grad_norm": 2.1392037868499756, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7214595079421997, + "num_tokens": 411498035.0, + "step": 16041 + }, + { + "epoch": 1.761695585328355, + "grad_norm": 2.2216269969940186, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7344551682472229, + "num_tokens": 411521725.0, + "step": 16042 + }, + { + "epoch": 1.7618054030309684, + "grad_norm": 2.283568859100342, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7178336977958679, + "num_tokens": 411549021.0, + "step": 16043 + }, + { + "epoch": 1.7619152207335822, + "grad_norm": 2.09366512298584, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7342022657394409, + "num_tokens": 411578029.0, + "step": 16044 + }, + { + "epoch": 1.762025038436196, + "grad_norm": 2.1934547424316406, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7060443758964539, + "num_tokens": 411604485.0, + "step": 16045 + }, + { + "epoch": 1.7621348561388097, + "grad_norm": 2.4938042163848877, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7397003769874573, + "num_tokens": 411625972.0, + "step": 16046 + }, + { + "epoch": 1.7622446738414232, + "grad_norm": 2.270663261413574, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7180695533752441, + "num_tokens": 411650766.0, + "step": 16047 + }, + { + "epoch": 1.7623544915440368, + "grad_norm": 2.343541145324707, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7023407816886902, + "num_tokens": 411672082.0, + "step": 16048 + }, + { + "epoch": 1.7624643092466505, + "grad_norm": 2.528395652770996, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7268123626708984, + "num_tokens": 411693646.0, + "step": 16049 + }, + { + "epoch": 1.7625741269492643, + "grad_norm": 2.7912962436676025, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7342402935028076, + "num_tokens": 411711412.0, + "step": 16050 + }, + { + "epoch": 1.7626839446518778, + "grad_norm": 2.3520267009735107, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.722760021686554, + "num_tokens": 411733572.0, + "step": 16051 + }, + { + "epoch": 1.7627937623544916, + "grad_norm": 2.3482582569122314, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7231814861297607, + "num_tokens": 411757125.0, + "step": 16052 + }, + { + "epoch": 1.7629035800571051, + "grad_norm": 1.902840495109558, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7473727464675903, + "num_tokens": 411786113.0, + "step": 16053 + }, + { + "epoch": 1.7630133977597189, + "grad_norm": 2.4921233654022217, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7023552656173706, + "num_tokens": 411808299.0, + "step": 16054 + }, + { + "epoch": 1.7631232154623326, + "grad_norm": 2.3312857151031494, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7200125455856323, + "num_tokens": 411831463.0, + "step": 16055 + }, + { + "epoch": 1.7632330331649462, + "grad_norm": 2.1380181312561035, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7053191661834717, + "num_tokens": 411858942.0, + "step": 16056 + }, + { + "epoch": 1.7633428508675597, + "grad_norm": 2.2042999267578125, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7303203344345093, + "num_tokens": 411884449.0, + "step": 16057 + }, + { + "epoch": 1.7634526685701735, + "grad_norm": 2.0640158653259277, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7402682900428772, + "num_tokens": 411913697.0, + "step": 16058 + }, + { + "epoch": 1.7635624862727872, + "grad_norm": 2.147512912750244, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7169526815414429, + "num_tokens": 411940564.0, + "step": 16059 + }, + { + "epoch": 1.763672303975401, + "grad_norm": 2.2206404209136963, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7290914058685303, + "num_tokens": 411966085.0, + "step": 16060 + }, + { + "epoch": 1.7637821216780145, + "grad_norm": 2.3638477325439453, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7233746647834778, + "num_tokens": 411989746.0, + "step": 16061 + }, + { + "epoch": 1.763891939380628, + "grad_norm": 2.341973066329956, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7073454856872559, + "num_tokens": 412014964.0, + "step": 16062 + }, + { + "epoch": 1.7640017570832418, + "grad_norm": 2.290001153945923, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7299346327781677, + "num_tokens": 412039326.0, + "step": 16063 + }, + { + "epoch": 1.7641115747858556, + "grad_norm": 2.383209228515625, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7190259695053101, + "num_tokens": 412061945.0, + "step": 16064 + }, + { + "epoch": 1.764221392488469, + "grad_norm": 2.206282615661621, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7314019799232483, + "num_tokens": 412087965.0, + "step": 16065 + }, + { + "epoch": 1.7643312101910829, + "grad_norm": 2.284090757369995, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7144404053688049, + "num_tokens": 412112149.0, + "step": 16066 + }, + { + "epoch": 1.7644410278936964, + "grad_norm": 2.442667245864868, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7336968779563904, + "num_tokens": 412134645.0, + "step": 16067 + }, + { + "epoch": 1.7645508455963101, + "grad_norm": 2.3914787769317627, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7217770218849182, + "num_tokens": 412158171.0, + "step": 16068 + }, + { + "epoch": 1.764660663298924, + "grad_norm": 2.04398775100708, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.6973695158958435, + "num_tokens": 412189231.0, + "step": 16069 + }, + { + "epoch": 1.7647704810015374, + "grad_norm": 2.5586130619049072, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7383033633232117, + "num_tokens": 412209455.0, + "step": 16070 + }, + { + "epoch": 1.764880298704151, + "grad_norm": 2.267653226852417, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7217826843261719, + "num_tokens": 412233636.0, + "step": 16071 + }, + { + "epoch": 1.7649901164067647, + "grad_norm": 2.8394370079040527, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7225776314735413, + "num_tokens": 412261761.0, + "step": 16072 + }, + { + "epoch": 1.7650999341093785, + "grad_norm": 2.5460147857666016, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7315951585769653, + "num_tokens": 412280205.0, + "step": 16073 + }, + { + "epoch": 1.7652097518119922, + "grad_norm": 2.416384220123291, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7203651666641235, + "num_tokens": 412303884.0, + "step": 16074 + }, + { + "epoch": 1.7653195695146058, + "grad_norm": 2.154536247253418, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7047021389007568, + "num_tokens": 412331921.0, + "step": 16075 + }, + { + "epoch": 1.7654293872172193, + "grad_norm": 2.6077983379364014, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7260026335716248, + "num_tokens": 412352181.0, + "step": 16076 + }, + { + "epoch": 1.765539204919833, + "grad_norm": 2.200235366821289, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7337411642074585, + "num_tokens": 412379142.0, + "step": 16077 + }, + { + "epoch": 1.7656490226224468, + "grad_norm": 2.0554065704345703, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7395949959754944, + "num_tokens": 412408579.0, + "step": 16078 + }, + { + "epoch": 1.7657588403250604, + "grad_norm": 2.275407075881958, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7401522397994995, + "num_tokens": 412431528.0, + "step": 16079 + }, + { + "epoch": 1.765868658027674, + "grad_norm": 2.3718338012695312, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7193597555160522, + "num_tokens": 412455141.0, + "step": 16080 + }, + { + "epoch": 1.7659784757302877, + "grad_norm": 2.294628620147705, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7065504193305969, + "num_tokens": 412480385.0, + "step": 16081 + }, + { + "epoch": 1.7660882934329014, + "grad_norm": 2.7862045764923096, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7108514308929443, + "num_tokens": 412497947.0, + "step": 16082 + }, + { + "epoch": 1.7661981111355152, + "grad_norm": 2.0984671115875244, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.709750771522522, + "num_tokens": 412525284.0, + "step": 16083 + }, + { + "epoch": 1.7663079288381287, + "grad_norm": 2.3716561794281006, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7340826988220215, + "num_tokens": 412547334.0, + "step": 16084 + }, + { + "epoch": 1.7664177465407422, + "grad_norm": 2.2370402812957764, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7318376302719116, + "num_tokens": 412571595.0, + "step": 16085 + }, + { + "epoch": 1.766527564243356, + "grad_norm": 2.045140027999878, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.6993305087089539, + "num_tokens": 412603040.0, + "step": 16086 + }, + { + "epoch": 1.7666373819459698, + "grad_norm": 2.284914493560791, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7050671577453613, + "num_tokens": 412628756.0, + "step": 16087 + }, + { + "epoch": 1.7667471996485835, + "grad_norm": 2.0113794803619385, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6884662508964539, + "num_tokens": 412660127.0, + "step": 16088 + }, + { + "epoch": 1.766857017351197, + "grad_norm": 2.171790361404419, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7174673676490784, + "num_tokens": 412685372.0, + "step": 16089 + }, + { + "epoch": 1.7669668350538106, + "grad_norm": 2.074019432067871, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.727569580078125, + "num_tokens": 412713658.0, + "step": 16090 + }, + { + "epoch": 1.7670766527564243, + "grad_norm": 2.1689507961273193, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7105044722557068, + "num_tokens": 412740502.0, + "step": 16091 + }, + { + "epoch": 1.767186470459038, + "grad_norm": 2.117164134979248, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7129079103469849, + "num_tokens": 412767516.0, + "step": 16092 + }, + { + "epoch": 1.7672962881616516, + "grad_norm": 2.1109039783477783, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7126996517181396, + "num_tokens": 412795890.0, + "step": 16093 + }, + { + "epoch": 1.7674061058642652, + "grad_norm": 2.1655197143554688, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7276727557182312, + "num_tokens": 412821879.0, + "step": 16094 + }, + { + "epoch": 1.767515923566879, + "grad_norm": 2.2918622493743896, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7204543948173523, + "num_tokens": 412847075.0, + "step": 16095 + }, + { + "epoch": 1.7676257412694927, + "grad_norm": 2.211207866668701, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.719501256942749, + "num_tokens": 412873247.0, + "step": 16096 + }, + { + "epoch": 1.7677355589721064, + "grad_norm": 2.0615901947021484, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7164385318756104, + "num_tokens": 412901864.0, + "step": 16097 + }, + { + "epoch": 1.76784537667472, + "grad_norm": 2.0768561363220215, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7297419309616089, + "num_tokens": 412929160.0, + "step": 16098 + }, + { + "epoch": 1.7679551943773335, + "grad_norm": 2.2470314502716064, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7101155519485474, + "num_tokens": 412953831.0, + "step": 16099 + }, + { + "epoch": 1.7680650120799473, + "grad_norm": 2.25042986869812, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7334580421447754, + "num_tokens": 412979849.0, + "step": 16100 + }, + { + "epoch": 1.768174829782561, + "grad_norm": 2.486922025680542, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.720552921295166, + "num_tokens": 413000849.0, + "step": 16101 + }, + { + "epoch": 1.7682846474851748, + "grad_norm": 2.09253191947937, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7240564227104187, + "num_tokens": 413029284.0, + "step": 16102 + }, + { + "epoch": 1.7683944651877883, + "grad_norm": 2.0904858112335205, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7049990296363831, + "num_tokens": 413059029.0, + "step": 16103 + }, + { + "epoch": 1.7685042828904018, + "grad_norm": 2.2294600009918213, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7233920097351074, + "num_tokens": 413085438.0, + "step": 16104 + }, + { + "epoch": 1.7686141005930156, + "grad_norm": 2.3555378913879395, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7264270186424255, + "num_tokens": 413109537.0, + "step": 16105 + }, + { + "epoch": 1.7687239182956294, + "grad_norm": 2.1669230461120605, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7297073602676392, + "num_tokens": 413136067.0, + "step": 16106 + }, + { + "epoch": 1.768833735998243, + "grad_norm": 2.3569350242614746, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6950632333755493, + "num_tokens": 413160880.0, + "step": 16107 + }, + { + "epoch": 1.7689435537008564, + "grad_norm": 2.2331528663635254, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7275349497795105, + "num_tokens": 413186359.0, + "step": 16108 + }, + { + "epoch": 1.7690533714034702, + "grad_norm": 2.684509754180908, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7321560382843018, + "num_tokens": 413205349.0, + "step": 16109 + }, + { + "epoch": 1.769163189106084, + "grad_norm": 2.173487901687622, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7005090117454529, + "num_tokens": 413232099.0, + "step": 16110 + }, + { + "epoch": 1.7692730068086977, + "grad_norm": 2.204927682876587, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.715333878993988, + "num_tokens": 413258416.0, + "step": 16111 + }, + { + "epoch": 1.7693828245113112, + "grad_norm": 2.0781824588775635, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7233667373657227, + "num_tokens": 413285146.0, + "step": 16112 + }, + { + "epoch": 1.7694926422139248, + "grad_norm": 2.4277262687683105, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7054216861724854, + "num_tokens": 413308427.0, + "step": 16113 + }, + { + "epoch": 1.7696024599165385, + "grad_norm": 2.441211223602295, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7055650949478149, + "num_tokens": 413330376.0, + "step": 16114 + }, + { + "epoch": 1.7697122776191523, + "grad_norm": 2.288503885269165, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7075849771499634, + "num_tokens": 413356451.0, + "step": 16115 + }, + { + "epoch": 1.7698220953217658, + "grad_norm": 2.535982847213745, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7303876876831055, + "num_tokens": 413377636.0, + "step": 16116 + }, + { + "epoch": 1.7699319130243796, + "grad_norm": 2.405073404312134, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7305804491043091, + "num_tokens": 413399580.0, + "step": 16117 + }, + { + "epoch": 1.7700417307269931, + "grad_norm": 2.2110538482666016, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7134753465652466, + "num_tokens": 413424547.0, + "step": 16118 + }, + { + "epoch": 1.7701515484296069, + "grad_norm": 2.163954257965088, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7167069315910339, + "num_tokens": 413451629.0, + "step": 16119 + }, + { + "epoch": 1.7702613661322206, + "grad_norm": 2.15519642829895, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7130829095840454, + "num_tokens": 413479499.0, + "step": 16120 + }, + { + "epoch": 1.7703711838348342, + "grad_norm": 2.4065699577331543, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7232111692428589, + "num_tokens": 413502136.0, + "step": 16121 + }, + { + "epoch": 1.7704810015374477, + "grad_norm": 2.256582498550415, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7416414022445679, + "num_tokens": 413526882.0, + "step": 16122 + }, + { + "epoch": 1.7705908192400615, + "grad_norm": 2.57624888420105, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7431955337524414, + "num_tokens": 413546788.0, + "step": 16123 + }, + { + "epoch": 1.7707006369426752, + "grad_norm": 2.501833438873291, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7325886487960815, + "num_tokens": 413567183.0, + "step": 16124 + }, + { + "epoch": 1.770810454645289, + "grad_norm": 2.1821322441101074, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7196922302246094, + "num_tokens": 413595856.0, + "step": 16125 + }, + { + "epoch": 1.7709202723479025, + "grad_norm": 1.8840428590774536, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7180752754211426, + "num_tokens": 413630339.0, + "step": 16126 + }, + { + "epoch": 1.771030090050516, + "grad_norm": 2.0592751502990723, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7163233757019043, + "num_tokens": 413659480.0, + "step": 16127 + }, + { + "epoch": 1.7711399077531298, + "grad_norm": 2.0861711502075195, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6949397325515747, + "num_tokens": 413689684.0, + "step": 16128 + }, + { + "epoch": 1.7712497254557436, + "grad_norm": 2.294408082962036, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7389463186264038, + "num_tokens": 413713347.0, + "step": 16129 + }, + { + "epoch": 1.771359543158357, + "grad_norm": 2.068618059158325, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7173752784729004, + "num_tokens": 413743186.0, + "step": 16130 + }, + { + "epoch": 1.7714693608609708, + "grad_norm": 2.2964377403259277, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7293906211853027, + "num_tokens": 413767552.0, + "step": 16131 + }, + { + "epoch": 1.7715791785635844, + "grad_norm": 2.254875659942627, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7116020917892456, + "num_tokens": 413793825.0, + "step": 16132 + }, + { + "epoch": 1.7716889962661981, + "grad_norm": 2.429511785507202, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7319750785827637, + "num_tokens": 413814390.0, + "step": 16133 + }, + { + "epoch": 1.771798813968812, + "grad_norm": 2.145312547683716, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.72183758020401, + "num_tokens": 413841885.0, + "step": 16134 + }, + { + "epoch": 1.7719086316714254, + "grad_norm": 2.3198394775390625, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7386666536331177, + "num_tokens": 413866854.0, + "step": 16135 + }, + { + "epoch": 1.772018449374039, + "grad_norm": 2.374582529067993, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7170045375823975, + "num_tokens": 413889767.0, + "step": 16136 + }, + { + "epoch": 1.7721282670766527, + "grad_norm": 2.580205202102661, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7293035387992859, + "num_tokens": 413910328.0, + "step": 16137 + }, + { + "epoch": 1.7722380847792665, + "grad_norm": 2.220139503479004, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.713464617729187, + "num_tokens": 413937005.0, + "step": 16138 + }, + { + "epoch": 1.7723479024818802, + "grad_norm": 2.367877721786499, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7419305443763733, + "num_tokens": 413959765.0, + "step": 16139 + }, + { + "epoch": 1.7724577201844938, + "grad_norm": 2.261244535446167, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7221896648406982, + "num_tokens": 413985096.0, + "step": 16140 + }, + { + "epoch": 1.7725675378871073, + "grad_norm": 2.1453845500946045, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7343906164169312, + "num_tokens": 414010865.0, + "step": 16141 + }, + { + "epoch": 1.772677355589721, + "grad_norm": 2.6591033935546875, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7312585115432739, + "num_tokens": 414028156.0, + "step": 16142 + }, + { + "epoch": 1.7727871732923348, + "grad_norm": 2.033311605453491, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7034552097320557, + "num_tokens": 414056937.0, + "step": 16143 + }, + { + "epoch": 1.7728969909949484, + "grad_norm": 2.30484676361084, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7121443152427673, + "num_tokens": 414081043.0, + "step": 16144 + }, + { + "epoch": 1.7730068086975619, + "grad_norm": 2.127047300338745, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7310134768486023, + "num_tokens": 414108749.0, + "step": 16145 + }, + { + "epoch": 1.7731166264001756, + "grad_norm": 2.373563766479492, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7236112952232361, + "num_tokens": 414131428.0, + "step": 16146 + }, + { + "epoch": 1.7732264441027894, + "grad_norm": 2.2786428928375244, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7206883430480957, + "num_tokens": 414156834.0, + "step": 16147 + }, + { + "epoch": 1.7733362618054032, + "grad_norm": 1.986799955368042, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.699378490447998, + "num_tokens": 414188731.0, + "step": 16148 + }, + { + "epoch": 1.7734460795080167, + "grad_norm": 1.9636259078979492, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7076895833015442, + "num_tokens": 414222335.0, + "step": 16149 + }, + { + "epoch": 1.7735558972106302, + "grad_norm": 2.620818614959717, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7345030903816223, + "num_tokens": 414242716.0, + "step": 16150 + }, + { + "epoch": 1.773665714913244, + "grad_norm": 2.0572257041931152, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7337160706520081, + "num_tokens": 414271631.0, + "step": 16151 + }, + { + "epoch": 1.7737755326158577, + "grad_norm": 2.049137830734253, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7352732419967651, + "num_tokens": 414299916.0, + "step": 16152 + }, + { + "epoch": 1.7738853503184715, + "grad_norm": 2.4280548095703125, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7121734619140625, + "num_tokens": 414323937.0, + "step": 16153 + }, + { + "epoch": 1.773995168021085, + "grad_norm": 2.5597214698791504, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.725555419921875, + "num_tokens": 414347284.0, + "step": 16154 + }, + { + "epoch": 1.7741049857236986, + "grad_norm": 2.0663809776306152, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6879189014434814, + "num_tokens": 414380338.0, + "step": 16155 + }, + { + "epoch": 1.7742148034263123, + "grad_norm": 2.2914834022521973, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7052986025810242, + "num_tokens": 414405679.0, + "step": 16156 + }, + { + "epoch": 1.774324621128926, + "grad_norm": 2.2484962940216064, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7099729776382446, + "num_tokens": 414429838.0, + "step": 16157 + }, + { + "epoch": 1.7744344388315396, + "grad_norm": 2.000351667404175, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7234249114990234, + "num_tokens": 414461045.0, + "step": 16158 + }, + { + "epoch": 1.7745442565341532, + "grad_norm": 2.0637826919555664, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.724396288394928, + "num_tokens": 414492324.0, + "step": 16159 + }, + { + "epoch": 1.774654074236767, + "grad_norm": 2.2953176498413086, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7260147929191589, + "num_tokens": 414516371.0, + "step": 16160 + }, + { + "epoch": 1.7747638919393807, + "grad_norm": 1.943058729171753, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7172131538391113, + "num_tokens": 414549849.0, + "step": 16161 + }, + { + "epoch": 1.7748737096419944, + "grad_norm": 2.519979953765869, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7431474924087524, + "num_tokens": 414569474.0, + "step": 16162 + }, + { + "epoch": 1.774983527344608, + "grad_norm": 2.1021716594696045, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7193487286567688, + "num_tokens": 414598201.0, + "step": 16163 + }, + { + "epoch": 1.7750933450472215, + "grad_norm": 2.2616829872131348, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7272176742553711, + "num_tokens": 414623551.0, + "step": 16164 + }, + { + "epoch": 1.7752031627498353, + "grad_norm": 1.7623822689056396, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7134208679199219, + "num_tokens": 414659403.0, + "step": 16165 + }, + { + "epoch": 1.775312980452449, + "grad_norm": 2.3927512168884277, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7098097801208496, + "num_tokens": 414684340.0, + "step": 16166 + }, + { + "epoch": 1.7754227981550625, + "grad_norm": 2.5335638523101807, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7346702218055725, + "num_tokens": 414704328.0, + "step": 16167 + }, + { + "epoch": 1.7755326158576763, + "grad_norm": 2.3537862300872803, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7099248170852661, + "num_tokens": 414729212.0, + "step": 16168 + }, + { + "epoch": 1.7756424335602898, + "grad_norm": 2.271214008331299, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7467554807662964, + "num_tokens": 414752906.0, + "step": 16169 + }, + { + "epoch": 1.7757522512629036, + "grad_norm": 2.2915632724761963, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7209555506706238, + "num_tokens": 414777329.0, + "step": 16170 + }, + { + "epoch": 1.7758620689655173, + "grad_norm": 2.1457366943359375, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7007973194122314, + "num_tokens": 414807041.0, + "step": 16171 + }, + { + "epoch": 1.7759718866681309, + "grad_norm": 2.0877373218536377, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.725655198097229, + "num_tokens": 414835583.0, + "step": 16172 + }, + { + "epoch": 1.7760817043707444, + "grad_norm": 2.1796858310699463, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7173886299133301, + "num_tokens": 414861327.0, + "step": 16173 + }, + { + "epoch": 1.7761915220733582, + "grad_norm": 2.1364474296569824, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7028093338012695, + "num_tokens": 414888473.0, + "step": 16174 + }, + { + "epoch": 1.776301339775972, + "grad_norm": 2.3528060913085938, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7232526540756226, + "num_tokens": 414910947.0, + "step": 16175 + }, + { + "epoch": 1.7764111574785857, + "grad_norm": 2.2849843502044678, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7295672297477722, + "num_tokens": 414935127.0, + "step": 16176 + }, + { + "epoch": 1.7765209751811992, + "grad_norm": 2.2078754901885986, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7017046809196472, + "num_tokens": 414962765.0, + "step": 16177 + }, + { + "epoch": 1.7766307928838128, + "grad_norm": 2.3181204795837402, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7227300405502319, + "num_tokens": 414986077.0, + "step": 16178 + }, + { + "epoch": 1.7767406105864265, + "grad_norm": 2.207270622253418, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6968745589256287, + "num_tokens": 415013479.0, + "step": 16179 + }, + { + "epoch": 1.7768504282890403, + "grad_norm": 2.150217294692993, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7324869632720947, + "num_tokens": 415040385.0, + "step": 16180 + }, + { + "epoch": 1.7769602459916538, + "grad_norm": 2.291532278060913, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.716263473033905, + "num_tokens": 415065560.0, + "step": 16181 + }, + { + "epoch": 1.7770700636942676, + "grad_norm": 2.1945111751556396, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7230803966522217, + "num_tokens": 415092467.0, + "step": 16182 + }, + { + "epoch": 1.777179881396881, + "grad_norm": 2.0651798248291016, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7106367349624634, + "num_tokens": 415120448.0, + "step": 16183 + }, + { + "epoch": 1.7772896990994949, + "grad_norm": 2.219928026199341, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7171356678009033, + "num_tokens": 415145186.0, + "step": 16184 + }, + { + "epoch": 1.7773995168021086, + "grad_norm": 1.875608205795288, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7017551064491272, + "num_tokens": 415180942.0, + "step": 16185 + }, + { + "epoch": 1.7775093345047221, + "grad_norm": 2.0239181518554688, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7091392278671265, + "num_tokens": 415210386.0, + "step": 16186 + }, + { + "epoch": 1.7776191522073357, + "grad_norm": 2.0762124061584473, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7316184043884277, + "num_tokens": 415237618.0, + "step": 16187 + }, + { + "epoch": 1.7777289699099494, + "grad_norm": 2.730783224105835, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7453804016113281, + "num_tokens": 415254785.0, + "step": 16188 + }, + { + "epoch": 1.7778387876125632, + "grad_norm": 2.462228536605835, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7233888506889343, + "num_tokens": 415275206.0, + "step": 16189 + }, + { + "epoch": 1.777948605315177, + "grad_norm": 2.3356387615203857, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.722172737121582, + "num_tokens": 415297962.0, + "step": 16190 + }, + { + "epoch": 1.7780584230177905, + "grad_norm": 2.1124937534332275, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7262390851974487, + "num_tokens": 415328236.0, + "step": 16191 + }, + { + "epoch": 1.778168240720404, + "grad_norm": 2.444875717163086, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7269936203956604, + "num_tokens": 415350703.0, + "step": 16192 + }, + { + "epoch": 1.7782780584230178, + "grad_norm": 2.100451707839966, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6986545920372009, + "num_tokens": 415381127.0, + "step": 16193 + }, + { + "epoch": 1.7783878761256315, + "grad_norm": 2.2240281105041504, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7092328667640686, + "num_tokens": 415406429.0, + "step": 16194 + }, + { + "epoch": 1.778497693828245, + "grad_norm": 2.173832416534424, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7059245705604553, + "num_tokens": 415432875.0, + "step": 16195 + }, + { + "epoch": 1.7786075115308586, + "grad_norm": 2.0493276119232178, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7375461459159851, + "num_tokens": 415459749.0, + "step": 16196 + }, + { + "epoch": 1.7787173292334724, + "grad_norm": 2.0578131675720215, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.6942530870437622, + "num_tokens": 415490878.0, + "step": 16197 + }, + { + "epoch": 1.7788271469360861, + "grad_norm": 2.313746213912964, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7274047136306763, + "num_tokens": 415514722.0, + "step": 16198 + }, + { + "epoch": 1.7789369646386999, + "grad_norm": 2.208904981613159, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6992497444152832, + "num_tokens": 415543257.0, + "step": 16199 + }, + { + "epoch": 1.7790467823413134, + "grad_norm": 2.255898952484131, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7089964151382446, + "num_tokens": 415567479.0, + "step": 16200 + }, + { + "epoch": 1.779156600043927, + "grad_norm": 2.6045844554901123, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7388502359390259, + "num_tokens": 415586077.0, + "step": 16201 + }, + { + "epoch": 1.7792664177465407, + "grad_norm": 2.4385669231414795, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7330236434936523, + "num_tokens": 415607310.0, + "step": 16202 + }, + { + "epoch": 1.7793762354491545, + "grad_norm": 2.276341199874878, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7148053646087646, + "num_tokens": 415633069.0, + "step": 16203 + }, + { + "epoch": 1.7794860531517682, + "grad_norm": 2.187021493911743, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7216777801513672, + "num_tokens": 415657691.0, + "step": 16204 + }, + { + "epoch": 1.7795958708543818, + "grad_norm": 1.9988791942596436, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7265756726264954, + "num_tokens": 415688756.0, + "step": 16205 + }, + { + "epoch": 1.7797056885569953, + "grad_norm": 2.2245736122131348, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7285699844360352, + "num_tokens": 415714086.0, + "step": 16206 + }, + { + "epoch": 1.779815506259609, + "grad_norm": 2.097458600997925, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7081589102745056, + "num_tokens": 415743837.0, + "step": 16207 + }, + { + "epoch": 1.7799253239622228, + "grad_norm": 2.489004135131836, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.719342827796936, + "num_tokens": 415766287.0, + "step": 16208 + }, + { + "epoch": 1.7800351416648363, + "grad_norm": 2.357409954071045, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7572245597839355, + "num_tokens": 415788954.0, + "step": 16209 + }, + { + "epoch": 1.7801449593674499, + "grad_norm": 2.3215909004211426, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7169853448867798, + "num_tokens": 415813218.0, + "step": 16210 + }, + { + "epoch": 1.7802547770700636, + "grad_norm": 2.223212242126465, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7236391305923462, + "num_tokens": 415839874.0, + "step": 16211 + }, + { + "epoch": 1.7803645947726774, + "grad_norm": 2.250584840774536, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7383823990821838, + "num_tokens": 415863958.0, + "step": 16212 + }, + { + "epoch": 1.7804744124752911, + "grad_norm": 2.1676690578460693, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7105116844177246, + "num_tokens": 415891230.0, + "step": 16213 + }, + { + "epoch": 1.7805842301779047, + "grad_norm": 2.036491632461548, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7023831605911255, + "num_tokens": 415923139.0, + "step": 16214 + }, + { + "epoch": 1.7806940478805182, + "grad_norm": 2.3596012592315674, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7150975465774536, + "num_tokens": 415947938.0, + "step": 16215 + }, + { + "epoch": 1.780803865583132, + "grad_norm": 2.8233580589294434, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7378072738647461, + "num_tokens": 415965939.0, + "step": 16216 + }, + { + "epoch": 1.7809136832857457, + "grad_norm": 2.4072539806365967, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7012107372283936, + "num_tokens": 415988572.0, + "step": 16217 + }, + { + "epoch": 1.7810235009883595, + "grad_norm": 2.2664146423339844, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.731491208076477, + "num_tokens": 416012775.0, + "step": 16218 + }, + { + "epoch": 1.781133318690973, + "grad_norm": 2.1741340160369873, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7146382331848145, + "num_tokens": 416042288.0, + "step": 16219 + }, + { + "epoch": 1.7812431363935866, + "grad_norm": 2.0741140842437744, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.701356053352356, + "num_tokens": 416072565.0, + "step": 16220 + }, + { + "epoch": 1.7813529540962003, + "grad_norm": 2.415637731552124, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7143723964691162, + "num_tokens": 416094738.0, + "step": 16221 + }, + { + "epoch": 1.781462771798814, + "grad_norm": 2.077134847640991, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6913649439811707, + "num_tokens": 416123991.0, + "step": 16222 + }, + { + "epoch": 1.7815725895014276, + "grad_norm": 2.3332135677337646, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7146275043487549, + "num_tokens": 416146407.0, + "step": 16223 + }, + { + "epoch": 1.7816824072040411, + "grad_norm": 2.17659854888916, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.731985330581665, + "num_tokens": 416171406.0, + "step": 16224 + }, + { + "epoch": 1.781792224906655, + "grad_norm": 2.1646625995635986, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7066241502761841, + "num_tokens": 416199421.0, + "step": 16225 + }, + { + "epoch": 1.7819020426092687, + "grad_norm": 2.6695621013641357, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7302474975585938, + "num_tokens": 416218825.0, + "step": 16226 + }, + { + "epoch": 1.7820118603118824, + "grad_norm": 2.1200308799743652, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7248406410217285, + "num_tokens": 416247201.0, + "step": 16227 + }, + { + "epoch": 1.782121678014496, + "grad_norm": 2.214528799057007, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.711908221244812, + "num_tokens": 416273894.0, + "step": 16228 + }, + { + "epoch": 1.7822314957171095, + "grad_norm": 2.1306684017181396, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7355535626411438, + "num_tokens": 416300461.0, + "step": 16229 + }, + { + "epoch": 1.7823413134197232, + "grad_norm": 2.424130439758301, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7340271472930908, + "num_tokens": 416323701.0, + "step": 16230 + }, + { + "epoch": 1.782451131122337, + "grad_norm": 2.293698787689209, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7212625741958618, + "num_tokens": 416346820.0, + "step": 16231 + }, + { + "epoch": 1.7825609488249505, + "grad_norm": 2.450087308883667, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7058700323104858, + "num_tokens": 416369955.0, + "step": 16232 + }, + { + "epoch": 1.7826707665275643, + "grad_norm": 2.188188076019287, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7190290689468384, + "num_tokens": 416395718.0, + "step": 16233 + }, + { + "epoch": 1.7827805842301778, + "grad_norm": 2.4578819274902344, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7303379774093628, + "num_tokens": 416418729.0, + "step": 16234 + }, + { + "epoch": 1.7828904019327916, + "grad_norm": 2.3335957527160645, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7214995622634888, + "num_tokens": 416445060.0, + "step": 16235 + }, + { + "epoch": 1.7830002196354053, + "grad_norm": 2.137366771697998, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.699927568435669, + "num_tokens": 416473168.0, + "step": 16236 + }, + { + "epoch": 1.7831100373380189, + "grad_norm": 2.1104838848114014, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7132077813148499, + "num_tokens": 416500534.0, + "step": 16237 + }, + { + "epoch": 1.7832198550406324, + "grad_norm": 2.140495777130127, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.724068284034729, + "num_tokens": 416526953.0, + "step": 16238 + }, + { + "epoch": 1.7833296727432462, + "grad_norm": 2.2865841388702393, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7046287059783936, + "num_tokens": 416551786.0, + "step": 16239 + }, + { + "epoch": 1.78343949044586, + "grad_norm": 2.1755218505859375, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.7008192539215088, + "num_tokens": 416579361.0, + "step": 16240 + }, + { + "epoch": 1.7835493081484737, + "grad_norm": 2.547451972961426, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7164485454559326, + "num_tokens": 416602147.0, + "step": 16241 + }, + { + "epoch": 1.7836591258510872, + "grad_norm": 2.262779474258423, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7297982573509216, + "num_tokens": 416628758.0, + "step": 16242 + }, + { + "epoch": 1.7837689435537007, + "grad_norm": 2.289052963256836, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7369113564491272, + "num_tokens": 416654183.0, + "step": 16243 + }, + { + "epoch": 1.7838787612563145, + "grad_norm": 2.4479353427886963, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7190492153167725, + "num_tokens": 416676651.0, + "step": 16244 + }, + { + "epoch": 1.7839885789589283, + "grad_norm": 2.2651636600494385, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7367368340492249, + "num_tokens": 416700337.0, + "step": 16245 + }, + { + "epoch": 1.7840983966615418, + "grad_norm": 1.9559849500656128, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6906068325042725, + "num_tokens": 416732700.0, + "step": 16246 + }, + { + "epoch": 1.7842082143641556, + "grad_norm": 2.0719614028930664, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7167688608169556, + "num_tokens": 416761423.0, + "step": 16247 + }, + { + "epoch": 1.784318032066769, + "grad_norm": 2.0769054889678955, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7091261148452759, + "num_tokens": 416792691.0, + "step": 16248 + }, + { + "epoch": 1.7844278497693828, + "grad_norm": 2.1741883754730225, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7318264245986938, + "num_tokens": 416821601.0, + "step": 16249 + }, + { + "epoch": 1.7845376674719966, + "grad_norm": 2.4691364765167236, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7277427911758423, + "num_tokens": 416843154.0, + "step": 16250 + }, + { + "epoch": 1.7846474851746101, + "grad_norm": 2.1649696826934814, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7346999049186707, + "num_tokens": 416869586.0, + "step": 16251 + }, + { + "epoch": 1.7847573028772237, + "grad_norm": 2.2498373985290527, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7234969139099121, + "num_tokens": 416895010.0, + "step": 16252 + }, + { + "epoch": 1.7848671205798374, + "grad_norm": 2.4880924224853516, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7404463291168213, + "num_tokens": 416916101.0, + "step": 16253 + }, + { + "epoch": 1.7849769382824512, + "grad_norm": 2.3871703147888184, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7360060214996338, + "num_tokens": 416937534.0, + "step": 16254 + }, + { + "epoch": 1.785086755985065, + "grad_norm": 2.32047700881958, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7522862553596497, + "num_tokens": 416959199.0, + "step": 16255 + }, + { + "epoch": 1.7851965736876785, + "grad_norm": 2.341350555419922, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7070688009262085, + "num_tokens": 416984522.0, + "step": 16256 + }, + { + "epoch": 1.785306391390292, + "grad_norm": 2.1922013759613037, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.730268120765686, + "num_tokens": 417009671.0, + "step": 16257 + }, + { + "epoch": 1.7854162090929058, + "grad_norm": 2.079023838043213, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7336630821228027, + "num_tokens": 417038346.0, + "step": 16258 + }, + { + "epoch": 1.7855260267955195, + "grad_norm": 2.20558500289917, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7219979763031006, + "num_tokens": 417066306.0, + "step": 16259 + }, + { + "epoch": 1.785635844498133, + "grad_norm": 2.205207347869873, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7242444157600403, + "num_tokens": 417091606.0, + "step": 16260 + }, + { + "epoch": 1.7857456622007466, + "grad_norm": 2.236586093902588, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7285258769989014, + "num_tokens": 417117611.0, + "step": 16261 + }, + { + "epoch": 1.7858554799033604, + "grad_norm": 2.3106741905212402, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7109793424606323, + "num_tokens": 417143831.0, + "step": 16262 + }, + { + "epoch": 1.7859652976059741, + "grad_norm": 2.3132681846618652, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.725355327129364, + "num_tokens": 417168478.0, + "step": 16263 + }, + { + "epoch": 1.7860751153085879, + "grad_norm": 2.3645060062408447, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7524622678756714, + "num_tokens": 417191324.0, + "step": 16264 + }, + { + "epoch": 1.7861849330112014, + "grad_norm": 2.0597476959228516, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7352169752120972, + "num_tokens": 417220036.0, + "step": 16265 + }, + { + "epoch": 1.786294750713815, + "grad_norm": 2.5153024196624756, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7344752550125122, + "num_tokens": 417242143.0, + "step": 16266 + }, + { + "epoch": 1.7864045684164287, + "grad_norm": 2.404736042022705, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7273472547531128, + "num_tokens": 417265527.0, + "step": 16267 + }, + { + "epoch": 1.7865143861190425, + "grad_norm": 2.261918067932129, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7119383215904236, + "num_tokens": 417291030.0, + "step": 16268 + }, + { + "epoch": 1.7866242038216562, + "grad_norm": 2.173994541168213, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.711435854434967, + "num_tokens": 417318780.0, + "step": 16269 + }, + { + "epoch": 1.7867340215242697, + "grad_norm": 2.657595634460449, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7409263253211975, + "num_tokens": 417336429.0, + "step": 16270 + }, + { + "epoch": 1.7868438392268833, + "grad_norm": 2.846924066543579, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7284644246101379, + "num_tokens": 417355648.0, + "step": 16271 + }, + { + "epoch": 1.786953656929497, + "grad_norm": 2.196941375732422, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6981047987937927, + "num_tokens": 417380536.0, + "step": 16272 + }, + { + "epoch": 1.7870634746321108, + "grad_norm": 2.1515424251556396, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7169966697692871, + "num_tokens": 417407016.0, + "step": 16273 + }, + { + "epoch": 1.7871732923347243, + "grad_norm": 2.347970724105835, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7277055978775024, + "num_tokens": 417429060.0, + "step": 16274 + }, + { + "epoch": 1.7872831100373379, + "grad_norm": 2.2656421661376953, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7050644159317017, + "num_tokens": 417455216.0, + "step": 16275 + }, + { + "epoch": 1.7873929277399516, + "grad_norm": 2.125812292098999, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7187500596046448, + "num_tokens": 417483268.0, + "step": 16276 + }, + { + "epoch": 1.7875027454425654, + "grad_norm": 2.3059494495391846, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7103831768035889, + "num_tokens": 417506535.0, + "step": 16277 + }, + { + "epoch": 1.7876125631451791, + "grad_norm": 2.286588191986084, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7283907532691956, + "num_tokens": 417527774.0, + "step": 16278 + }, + { + "epoch": 1.7877223808477927, + "grad_norm": 2.007420063018799, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7279754877090454, + "num_tokens": 417556872.0, + "step": 16279 + }, + { + "epoch": 1.7878321985504062, + "grad_norm": 2.0934412479400635, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6949647665023804, + "num_tokens": 417588210.0, + "step": 16280 + }, + { + "epoch": 1.78794201625302, + "grad_norm": 2.4522905349731445, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6985533237457275, + "num_tokens": 417612293.0, + "step": 16281 + }, + { + "epoch": 1.7880518339556337, + "grad_norm": 2.3467812538146973, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7547268867492676, + "num_tokens": 417633933.0, + "step": 16282 + }, + { + "epoch": 1.7881616516582475, + "grad_norm": 2.1038832664489746, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7168117761611938, + "num_tokens": 417662109.0, + "step": 16283 + }, + { + "epoch": 1.788271469360861, + "grad_norm": 2.386827230453491, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.735217273235321, + "num_tokens": 417682818.0, + "step": 16284 + }, + { + "epoch": 1.7883812870634745, + "grad_norm": 2.3022842407226562, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7008017301559448, + "num_tokens": 417709932.0, + "step": 16285 + }, + { + "epoch": 1.7884911047660883, + "grad_norm": 2.278733968734741, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7236997485160828, + "num_tokens": 417735790.0, + "step": 16286 + }, + { + "epoch": 1.788600922468702, + "grad_norm": 1.8835831880569458, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7161632180213928, + "num_tokens": 417769815.0, + "step": 16287 + }, + { + "epoch": 1.7887107401713156, + "grad_norm": 2.3965542316436768, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7317731976509094, + "num_tokens": 417792832.0, + "step": 16288 + }, + { + "epoch": 1.7888205578739291, + "grad_norm": 2.1127707958221436, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7130248546600342, + "num_tokens": 417819624.0, + "step": 16289 + }, + { + "epoch": 1.7889303755765429, + "grad_norm": 2.2639267444610596, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7124133706092834, + "num_tokens": 417844554.0, + "step": 16290 + }, + { + "epoch": 1.7890401932791566, + "grad_norm": 2.0325727462768555, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6948957443237305, + "num_tokens": 417877707.0, + "step": 16291 + }, + { + "epoch": 1.7891500109817704, + "grad_norm": 2.3919222354888916, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7146685123443604, + "num_tokens": 417899997.0, + "step": 16292 + }, + { + "epoch": 1.789259828684384, + "grad_norm": 2.391465663909912, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7086026668548584, + "num_tokens": 417924097.0, + "step": 16293 + }, + { + "epoch": 1.7893696463869975, + "grad_norm": 2.4656872749328613, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7276926040649414, + "num_tokens": 417944716.0, + "step": 16294 + }, + { + "epoch": 1.7894794640896112, + "grad_norm": 2.628933906555176, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7062810659408569, + "num_tokens": 417965966.0, + "step": 16295 + }, + { + "epoch": 1.789589281792225, + "grad_norm": 2.0493314266204834, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7033862471580505, + "num_tokens": 417994709.0, + "step": 16296 + }, + { + "epoch": 1.7896990994948385, + "grad_norm": 2.339341878890991, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7296327352523804, + "num_tokens": 418019185.0, + "step": 16297 + }, + { + "epoch": 1.7898089171974523, + "grad_norm": 2.223184585571289, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6923044919967651, + "num_tokens": 418048392.0, + "step": 16298 + }, + { + "epoch": 1.7899187349000658, + "grad_norm": 2.113326072692871, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7084998488426208, + "num_tokens": 418076345.0, + "step": 16299 + }, + { + "epoch": 1.7900285526026796, + "grad_norm": 2.304861545562744, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7106314897537231, + "num_tokens": 418100952.0, + "step": 16300 + }, + { + "epoch": 1.7901383703052933, + "grad_norm": 2.1411728858947754, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7268987894058228, + "num_tokens": 418127565.0, + "step": 16301 + }, + { + "epoch": 1.7902481880079069, + "grad_norm": 2.1081061363220215, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.727921724319458, + "num_tokens": 418156279.0, + "step": 16302 + }, + { + "epoch": 1.7903580057105204, + "grad_norm": 2.7139925956726074, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.736619234085083, + "num_tokens": 418173952.0, + "step": 16303 + }, + { + "epoch": 1.7904678234131342, + "grad_norm": 2.129612684249878, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7039248943328857, + "num_tokens": 418203849.0, + "step": 16304 + }, + { + "epoch": 1.790577641115748, + "grad_norm": 2.169267416000366, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7244696021080017, + "num_tokens": 418231037.0, + "step": 16305 + }, + { + "epoch": 1.7906874588183617, + "grad_norm": 2.3956422805786133, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7368320226669312, + "num_tokens": 418253603.0, + "step": 16306 + }, + { + "epoch": 1.7907972765209752, + "grad_norm": 2.0565946102142334, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7145589590072632, + "num_tokens": 418282939.0, + "step": 16307 + }, + { + "epoch": 1.7909070942235887, + "grad_norm": 2.192990303039551, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7105557322502136, + "num_tokens": 418313158.0, + "step": 16308 + }, + { + "epoch": 1.7910169119262025, + "grad_norm": 2.294823169708252, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7236031293869019, + "num_tokens": 418337284.0, + "step": 16309 + }, + { + "epoch": 1.7911267296288162, + "grad_norm": 2.311418056488037, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7097541093826294, + "num_tokens": 418361907.0, + "step": 16310 + }, + { + "epoch": 1.7912365473314298, + "grad_norm": 2.6267154216766357, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7347106337547302, + "num_tokens": 418381307.0, + "step": 16311 + }, + { + "epoch": 1.7913463650340435, + "grad_norm": 2.009716749191284, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7132756114006042, + "num_tokens": 418411947.0, + "step": 16312 + }, + { + "epoch": 1.791456182736657, + "grad_norm": 2.270260810852051, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7042150497436523, + "num_tokens": 418435471.0, + "step": 16313 + }, + { + "epoch": 1.7915660004392708, + "grad_norm": 2.451268434524536, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.727420449256897, + "num_tokens": 418457993.0, + "step": 16314 + }, + { + "epoch": 1.7916758181418846, + "grad_norm": 2.3241608142852783, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7278668880462646, + "num_tokens": 418481095.0, + "step": 16315 + }, + { + "epoch": 1.7917856358444981, + "grad_norm": 2.888345241546631, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7342044115066528, + "num_tokens": 418497757.0, + "step": 16316 + }, + { + "epoch": 1.7918954535471117, + "grad_norm": 2.466595411300659, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7214512825012207, + "num_tokens": 418519193.0, + "step": 16317 + }, + { + "epoch": 1.7920052712497254, + "grad_norm": 2.1583306789398193, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.6963636875152588, + "num_tokens": 418546160.0, + "step": 16318 + }, + { + "epoch": 1.7921150889523392, + "grad_norm": 1.957108497619629, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7164857387542725, + "num_tokens": 418577821.0, + "step": 16319 + }, + { + "epoch": 1.792224906654953, + "grad_norm": 2.227276563644409, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7236551642417908, + "num_tokens": 418605655.0, + "step": 16320 + }, + { + "epoch": 1.7923347243575665, + "grad_norm": 2.0415589809417725, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7131644487380981, + "num_tokens": 418633760.0, + "step": 16321 + }, + { + "epoch": 1.79244454206018, + "grad_norm": 2.1876156330108643, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6992030143737793, + "num_tokens": 418662643.0, + "step": 16322 + }, + { + "epoch": 1.7925543597627938, + "grad_norm": 2.4242281913757324, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7300910353660583, + "num_tokens": 418685648.0, + "step": 16323 + }, + { + "epoch": 1.7926641774654075, + "grad_norm": 2.1181352138519287, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7364289164543152, + "num_tokens": 418711500.0, + "step": 16324 + }, + { + "epoch": 1.792773995168021, + "grad_norm": 2.231433153152466, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7256453037261963, + "num_tokens": 418738180.0, + "step": 16325 + }, + { + "epoch": 1.7928838128706346, + "grad_norm": 2.359583854675293, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6953049302101135, + "num_tokens": 418765084.0, + "step": 16326 + }, + { + "epoch": 1.7929936305732483, + "grad_norm": 2.2834577560424805, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7122697830200195, + "num_tokens": 418790682.0, + "step": 16327 + }, + { + "epoch": 1.793103448275862, + "grad_norm": 2.0448367595672607, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7336590886116028, + "num_tokens": 418818711.0, + "step": 16328 + }, + { + "epoch": 1.7932132659784759, + "grad_norm": 2.6414740085601807, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7407383918762207, + "num_tokens": 418837454.0, + "step": 16329 + }, + { + "epoch": 1.7933230836810894, + "grad_norm": 2.2731451988220215, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7148532867431641, + "num_tokens": 418861987.0, + "step": 16330 + }, + { + "epoch": 1.793432901383703, + "grad_norm": 1.8733782768249512, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7015029191970825, + "num_tokens": 418897875.0, + "step": 16331 + }, + { + "epoch": 1.7935427190863167, + "grad_norm": 2.6067936420440674, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7244385480880737, + "num_tokens": 418920425.0, + "step": 16332 + }, + { + "epoch": 1.7936525367889304, + "grad_norm": 2.419811248779297, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7313551306724548, + "num_tokens": 418943622.0, + "step": 16333 + }, + { + "epoch": 1.7937623544915442, + "grad_norm": 2.300813674926758, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7250417470932007, + "num_tokens": 418969671.0, + "step": 16334 + }, + { + "epoch": 1.7938721721941577, + "grad_norm": 2.2812106609344482, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7287085056304932, + "num_tokens": 418992840.0, + "step": 16335 + }, + { + "epoch": 1.7939819898967713, + "grad_norm": 2.2732954025268555, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7494021654129028, + "num_tokens": 419016085.0, + "step": 16336 + }, + { + "epoch": 1.794091807599385, + "grad_norm": 2.387284994125366, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7318915724754333, + "num_tokens": 419038203.0, + "step": 16337 + }, + { + "epoch": 1.7942016253019988, + "grad_norm": 2.0404930114746094, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7234296798706055, + "num_tokens": 419065816.0, + "step": 16338 + }, + { + "epoch": 1.7943114430046123, + "grad_norm": 1.9918543100357056, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6981445550918579, + "num_tokens": 419098252.0, + "step": 16339 + }, + { + "epoch": 1.7944212607072259, + "grad_norm": 2.3902924060821533, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.727832555770874, + "num_tokens": 419121264.0, + "step": 16340 + }, + { + "epoch": 1.7945310784098396, + "grad_norm": 2.600273609161377, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7184778451919556, + "num_tokens": 419141564.0, + "step": 16341 + }, + { + "epoch": 1.7946408961124534, + "grad_norm": 2.2437093257904053, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7169796228408813, + "num_tokens": 419168311.0, + "step": 16342 + }, + { + "epoch": 1.7947507138150671, + "grad_norm": 2.0934832096099854, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7253972887992859, + "num_tokens": 419197709.0, + "step": 16343 + }, + { + "epoch": 1.7948605315176807, + "grad_norm": 2.086604118347168, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7254838943481445, + "num_tokens": 419227089.0, + "step": 16344 + }, + { + "epoch": 1.7949703492202942, + "grad_norm": 2.090282440185547, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7144060134887695, + "num_tokens": 419259459.0, + "step": 16345 + }, + { + "epoch": 1.795080166922908, + "grad_norm": 3.0341787338256836, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7314288020133972, + "num_tokens": 419274752.0, + "step": 16346 + }, + { + "epoch": 1.7951899846255217, + "grad_norm": 2.297593355178833, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7344815135002136, + "num_tokens": 419299408.0, + "step": 16347 + }, + { + "epoch": 1.7952998023281352, + "grad_norm": 2.1552023887634277, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6938025951385498, + "num_tokens": 419325768.0, + "step": 16348 + }, + { + "epoch": 1.795409620030749, + "grad_norm": 2.3087868690490723, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7247732281684875, + "num_tokens": 419350560.0, + "step": 16349 + }, + { + "epoch": 1.7955194377333625, + "grad_norm": 2.439970016479492, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.713894784450531, + "num_tokens": 419374287.0, + "step": 16350 + }, + { + "epoch": 1.7956292554359763, + "grad_norm": 2.3383560180664062, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6957415342330933, + "num_tokens": 419400305.0, + "step": 16351 + }, + { + "epoch": 1.79573907313859, + "grad_norm": 2.1970434188842773, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7504095435142517, + "num_tokens": 419425740.0, + "step": 16352 + }, + { + "epoch": 1.7958488908412036, + "grad_norm": 2.423825740814209, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7363262176513672, + "num_tokens": 419450824.0, + "step": 16353 + }, + { + "epoch": 1.7959587085438171, + "grad_norm": 2.496455669403076, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7163869142532349, + "num_tokens": 419474158.0, + "step": 16354 + }, + { + "epoch": 1.7960685262464309, + "grad_norm": 2.106714963912964, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7370470762252808, + "num_tokens": 419503481.0, + "step": 16355 + }, + { + "epoch": 1.7961783439490446, + "grad_norm": 1.9846467971801758, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7140026688575745, + "num_tokens": 419535782.0, + "step": 16356 + }, + { + "epoch": 1.7962881616516584, + "grad_norm": 2.5142409801483154, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7149451375007629, + "num_tokens": 419557914.0, + "step": 16357 + }, + { + "epoch": 1.796397979354272, + "grad_norm": 1.8645981550216675, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7154480218887329, + "num_tokens": 419592614.0, + "step": 16358 + }, + { + "epoch": 1.7965077970568855, + "grad_norm": 2.103774070739746, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6963856816291809, + "num_tokens": 419622048.0, + "step": 16359 + }, + { + "epoch": 1.7966176147594992, + "grad_norm": 2.1946606636047363, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7107460498809814, + "num_tokens": 419647380.0, + "step": 16360 + }, + { + "epoch": 1.796727432462113, + "grad_norm": 2.291072130203247, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.746096134185791, + "num_tokens": 419670934.0, + "step": 16361 + }, + { + "epoch": 1.7968372501647265, + "grad_norm": 2.3982512950897217, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.712173342704773, + "num_tokens": 419693451.0, + "step": 16362 + }, + { + "epoch": 1.7969470678673403, + "grad_norm": 2.4524388313293457, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7191075086593628, + "num_tokens": 419716639.0, + "step": 16363 + }, + { + "epoch": 1.7970568855699538, + "grad_norm": 2.21169114112854, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.735054612159729, + "num_tokens": 419741861.0, + "step": 16364 + }, + { + "epoch": 1.7971667032725676, + "grad_norm": 1.8552751541137695, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7089834809303284, + "num_tokens": 419776953.0, + "step": 16365 + }, + { + "epoch": 1.7972765209751813, + "grad_norm": 2.261793375015259, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.730050802230835, + "num_tokens": 419802699.0, + "step": 16366 + }, + { + "epoch": 1.7973863386777948, + "grad_norm": 2.207552671432495, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7197373509407043, + "num_tokens": 419827701.0, + "step": 16367 + }, + { + "epoch": 1.7974961563804084, + "grad_norm": 2.2134039402008057, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7085791826248169, + "num_tokens": 419858571.0, + "step": 16368 + }, + { + "epoch": 1.7976059740830221, + "grad_norm": 2.1588222980499268, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7303493022918701, + "num_tokens": 419885302.0, + "step": 16369 + }, + { + "epoch": 1.797715791785636, + "grad_norm": 2.3286352157592773, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.732756495475769, + "num_tokens": 419908638.0, + "step": 16370 + }, + { + "epoch": 1.7978256094882497, + "grad_norm": 2.018207550048828, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7253649234771729, + "num_tokens": 419937714.0, + "step": 16371 + }, + { + "epoch": 1.7979354271908632, + "grad_norm": 2.5088274478912354, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.712272584438324, + "num_tokens": 419958598.0, + "step": 16372 + }, + { + "epoch": 1.7980452448934767, + "grad_norm": 1.9231983423233032, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7270081639289856, + "num_tokens": 419990505.0, + "step": 16373 + }, + { + "epoch": 1.7981550625960905, + "grad_norm": 2.6880950927734375, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.71805340051651, + "num_tokens": 420010901.0, + "step": 16374 + }, + { + "epoch": 1.7982648802987042, + "grad_norm": 2.1921660900115967, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7530694007873535, + "num_tokens": 420035117.0, + "step": 16375 + }, + { + "epoch": 1.7983746980013178, + "grad_norm": 2.1579415798187256, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7406335473060608, + "num_tokens": 420061314.0, + "step": 16376 + }, + { + "epoch": 1.7984845157039313, + "grad_norm": 2.377697229385376, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7239209413528442, + "num_tokens": 420084817.0, + "step": 16377 + }, + { + "epoch": 1.798594333406545, + "grad_norm": 2.329303741455078, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7270251512527466, + "num_tokens": 420108755.0, + "step": 16378 + }, + { + "epoch": 1.7987041511091588, + "grad_norm": 1.99944269657135, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6976867914199829, + "num_tokens": 420139657.0, + "step": 16379 + }, + { + "epoch": 1.7988139688117726, + "grad_norm": 2.2061209678649902, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7115335464477539, + "num_tokens": 420167472.0, + "step": 16380 + }, + { + "epoch": 1.7989237865143861, + "grad_norm": 1.9915070533752441, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7252914309501648, + "num_tokens": 420200308.0, + "step": 16381 + }, + { + "epoch": 1.7990336042169996, + "grad_norm": 2.131873846054077, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7297372817993164, + "num_tokens": 420227500.0, + "step": 16382 + }, + { + "epoch": 1.7991434219196134, + "grad_norm": 2.063913106918335, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7102750539779663, + "num_tokens": 420258562.0, + "step": 16383 + }, + { + "epoch": 1.7992532396222272, + "grad_norm": 2.089834690093994, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7266854643821716, + "num_tokens": 420285884.0, + "step": 16384 + }, + { + "epoch": 1.799363057324841, + "grad_norm": 2.320089101791382, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7037270069122314, + "num_tokens": 420309707.0, + "step": 16385 + }, + { + "epoch": 1.7994728750274545, + "grad_norm": 1.9781055450439453, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7079358696937561, + "num_tokens": 420341864.0, + "step": 16386 + }, + { + "epoch": 1.799582692730068, + "grad_norm": 2.489738941192627, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6979765295982361, + "num_tokens": 420365730.0, + "step": 16387 + }, + { + "epoch": 1.7996925104326817, + "grad_norm": 2.53717303276062, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7468583583831787, + "num_tokens": 420385284.0, + "step": 16388 + }, + { + "epoch": 1.7998023281352955, + "grad_norm": 2.49285626411438, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.727037787437439, + "num_tokens": 420405437.0, + "step": 16389 + }, + { + "epoch": 1.799912145837909, + "grad_norm": 2.1604835987091064, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7078681588172913, + "num_tokens": 420431640.0, + "step": 16390 + }, + { + "epoch": 1.8000219635405226, + "grad_norm": 2.1136116981506348, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7430794835090637, + "num_tokens": 420456909.0, + "step": 16391 + }, + { + "epoch": 1.8001317812431363, + "grad_norm": 2.2000961303710938, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7137771844863892, + "num_tokens": 420482949.0, + "step": 16392 + }, + { + "epoch": 1.80024159894575, + "grad_norm": 2.203054666519165, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6956924796104431, + "num_tokens": 420508791.0, + "step": 16393 + }, + { + "epoch": 1.8003514166483638, + "grad_norm": 1.9358289241790771, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.6994175910949707, + "num_tokens": 420540523.0, + "step": 16394 + }, + { + "epoch": 1.8004612343509774, + "grad_norm": 2.0402424335479736, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7198878526687622, + "num_tokens": 420569943.0, + "step": 16395 + }, + { + "epoch": 1.800571052053591, + "grad_norm": 2.3949530124664307, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7187293171882629, + "num_tokens": 420595030.0, + "step": 16396 + }, + { + "epoch": 1.8006808697562047, + "grad_norm": 2.2371926307678223, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7205744385719299, + "num_tokens": 420621837.0, + "step": 16397 + }, + { + "epoch": 1.8007906874588184, + "grad_norm": 2.2844178676605225, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.732059121131897, + "num_tokens": 420646849.0, + "step": 16398 + }, + { + "epoch": 1.8009005051614322, + "grad_norm": 2.2730700969696045, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7393744587898254, + "num_tokens": 420670635.0, + "step": 16399 + }, + { + "epoch": 1.8010103228640457, + "grad_norm": 2.1196320056915283, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7099512815475464, + "num_tokens": 420697324.0, + "step": 16400 + }, + { + "epoch": 1.8011201405666593, + "grad_norm": 2.5988359451293945, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7470218539237976, + "num_tokens": 420716939.0, + "step": 16401 + }, + { + "epoch": 1.801229958269273, + "grad_norm": 2.143017053604126, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7210803031921387, + "num_tokens": 420745527.0, + "step": 16402 + }, + { + "epoch": 1.8013397759718868, + "grad_norm": 2.2517917156219482, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.736782431602478, + "num_tokens": 420769874.0, + "step": 16403 + }, + { + "epoch": 1.8014495936745003, + "grad_norm": 2.542019844055176, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7242516279220581, + "num_tokens": 420789995.0, + "step": 16404 + }, + { + "epoch": 1.8015594113771138, + "grad_norm": 2.4216904640197754, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7201519012451172, + "num_tokens": 420813080.0, + "step": 16405 + }, + { + "epoch": 1.8016692290797276, + "grad_norm": 2.0308279991149902, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.72425377368927, + "num_tokens": 420842081.0, + "step": 16406 + }, + { + "epoch": 1.8017790467823414, + "grad_norm": 2.2553913593292236, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7083016633987427, + "num_tokens": 420870346.0, + "step": 16407 + }, + { + "epoch": 1.801888864484955, + "grad_norm": 2.3832600116729736, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.717271089553833, + "num_tokens": 420894966.0, + "step": 16408 + }, + { + "epoch": 1.8019986821875686, + "grad_norm": 2.2009050846099854, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7044116258621216, + "num_tokens": 420924698.0, + "step": 16409 + }, + { + "epoch": 1.8021084998901822, + "grad_norm": 2.1108243465423584, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7283622026443481, + "num_tokens": 420953232.0, + "step": 16410 + }, + { + "epoch": 1.802218317592796, + "grad_norm": 2.209527015686035, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7254625558853149, + "num_tokens": 420981942.0, + "step": 16411 + }, + { + "epoch": 1.8023281352954097, + "grad_norm": 2.027737855911255, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7305135726928711, + "num_tokens": 421010559.0, + "step": 16412 + }, + { + "epoch": 1.8024379529980232, + "grad_norm": 2.1008307933807373, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7257264256477356, + "num_tokens": 421039101.0, + "step": 16413 + }, + { + "epoch": 1.802547770700637, + "grad_norm": 2.3512325286865234, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6973637938499451, + "num_tokens": 421064055.0, + "step": 16414 + }, + { + "epoch": 1.8026575884032505, + "grad_norm": 1.9045331478118896, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6857253313064575, + "num_tokens": 421102315.0, + "step": 16415 + }, + { + "epoch": 1.8027674061058643, + "grad_norm": 2.034148693084717, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.721016526222229, + "num_tokens": 421131338.0, + "step": 16416 + }, + { + "epoch": 1.802877223808478, + "grad_norm": 1.9506688117980957, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7129003405570984, + "num_tokens": 421165463.0, + "step": 16417 + }, + { + "epoch": 1.8029870415110916, + "grad_norm": 2.259397506713867, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7083283066749573, + "num_tokens": 421190354.0, + "step": 16418 + }, + { + "epoch": 1.803096859213705, + "grad_norm": 2.1022777557373047, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7361626029014587, + "num_tokens": 421216438.0, + "step": 16419 + }, + { + "epoch": 1.8032066769163189, + "grad_norm": 2.2804088592529297, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6987581849098206, + "num_tokens": 421240869.0, + "step": 16420 + }, + { + "epoch": 1.8033164946189326, + "grad_norm": 2.559736490249634, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7000466585159302, + "num_tokens": 421262568.0, + "step": 16421 + }, + { + "epoch": 1.8034263123215464, + "grad_norm": 2.223588466644287, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7120206952095032, + "num_tokens": 421287574.0, + "step": 16422 + }, + { + "epoch": 1.80353613002416, + "grad_norm": 2.3201985359191895, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.719459056854248, + "num_tokens": 421310807.0, + "step": 16423 + }, + { + "epoch": 1.8036459477267734, + "grad_norm": 2.03171706199646, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6839001178741455, + "num_tokens": 421340895.0, + "step": 16424 + }, + { + "epoch": 1.8037557654293872, + "grad_norm": 2.2253973484039307, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6990801692008972, + "num_tokens": 421368995.0, + "step": 16425 + }, + { + "epoch": 1.803865583132001, + "grad_norm": 2.4055521488189697, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7113739252090454, + "num_tokens": 421392039.0, + "step": 16426 + }, + { + "epoch": 1.8039754008346145, + "grad_norm": 2.1264491081237793, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.732735276222229, + "num_tokens": 421418802.0, + "step": 16427 + }, + { + "epoch": 1.8040852185372283, + "grad_norm": 2.0019724369049072, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7416888475418091, + "num_tokens": 421446090.0, + "step": 16428 + }, + { + "epoch": 1.8041950362398418, + "grad_norm": 2.5550734996795654, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7665007710456848, + "num_tokens": 421464913.0, + "step": 16429 + }, + { + "epoch": 1.8043048539424555, + "grad_norm": 2.2122344970703125, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7196664810180664, + "num_tokens": 421491269.0, + "step": 16430 + }, + { + "epoch": 1.8044146716450693, + "grad_norm": 1.9521429538726807, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7020902037620544, + "num_tokens": 421524227.0, + "step": 16431 + }, + { + "epoch": 1.8045244893476828, + "grad_norm": 2.343742847442627, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7280375957489014, + "num_tokens": 421547393.0, + "step": 16432 + }, + { + "epoch": 1.8046343070502964, + "grad_norm": 2.01147198677063, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7336586117744446, + "num_tokens": 421576751.0, + "step": 16433 + }, + { + "epoch": 1.8047441247529101, + "grad_norm": 2.1693572998046875, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7213640213012695, + "num_tokens": 421603720.0, + "step": 16434 + }, + { + "epoch": 1.8048539424555239, + "grad_norm": 2.040947914123535, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7139667272567749, + "num_tokens": 421634102.0, + "step": 16435 + }, + { + "epoch": 1.8049637601581376, + "grad_norm": 2.708728075027466, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7506174445152283, + "num_tokens": 421652657.0, + "step": 16436 + }, + { + "epoch": 1.8050735778607512, + "grad_norm": 2.240565538406372, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.720658540725708, + "num_tokens": 421676781.0, + "step": 16437 + }, + { + "epoch": 1.8051833955633647, + "grad_norm": 2.019498825073242, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.69862961769104, + "num_tokens": 421708648.0, + "step": 16438 + }, + { + "epoch": 1.8052932132659785, + "grad_norm": 2.3140716552734375, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7277582287788391, + "num_tokens": 421732429.0, + "step": 16439 + }, + { + "epoch": 1.8054030309685922, + "grad_norm": 2.242293357849121, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6809265613555908, + "num_tokens": 421760909.0, + "step": 16440 + }, + { + "epoch": 1.8055128486712058, + "grad_norm": 2.3487040996551514, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6970253586769104, + "num_tokens": 421785745.0, + "step": 16441 + }, + { + "epoch": 1.8056226663738193, + "grad_norm": 2.3451087474823, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7256988883018494, + "num_tokens": 421810852.0, + "step": 16442 + }, + { + "epoch": 1.805732484076433, + "grad_norm": 2.0451431274414062, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7039408683776855, + "num_tokens": 421843310.0, + "step": 16443 + }, + { + "epoch": 1.8058423017790468, + "grad_norm": 2.259190797805786, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7137407660484314, + "num_tokens": 421870765.0, + "step": 16444 + }, + { + "epoch": 1.8059521194816606, + "grad_norm": 2.646476984024048, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7285709381103516, + "num_tokens": 421891397.0, + "step": 16445 + }, + { + "epoch": 1.806061937184274, + "grad_norm": 2.2279350757598877, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7278339862823486, + "num_tokens": 421917250.0, + "step": 16446 + }, + { + "epoch": 1.8061717548868876, + "grad_norm": 2.198500394821167, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7266311049461365, + "num_tokens": 421943945.0, + "step": 16447 + }, + { + "epoch": 1.8062815725895014, + "grad_norm": 2.3883275985717773, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7393394708633423, + "num_tokens": 421966975.0, + "step": 16448 + }, + { + "epoch": 1.8063913902921152, + "grad_norm": 2.0231614112854004, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7079266309738159, + "num_tokens": 421996936.0, + "step": 16449 + }, + { + "epoch": 1.806501207994729, + "grad_norm": 2.3794236183166504, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7397760152816772, + "num_tokens": 422020548.0, + "step": 16450 + }, + { + "epoch": 1.8066110256973424, + "grad_norm": 2.226794481277466, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.704586386680603, + "num_tokens": 422046073.0, + "step": 16451 + }, + { + "epoch": 1.806720843399956, + "grad_norm": 2.281167984008789, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7209746837615967, + "num_tokens": 422070775.0, + "step": 16452 + }, + { + "epoch": 1.8068306611025697, + "grad_norm": 2.0406951904296875, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7390832901000977, + "num_tokens": 422099091.0, + "step": 16453 + }, + { + "epoch": 1.8069404788051835, + "grad_norm": 2.2088820934295654, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7269478440284729, + "num_tokens": 422124162.0, + "step": 16454 + }, + { + "epoch": 1.807050296507797, + "grad_norm": 2.5462539196014404, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7283006310462952, + "num_tokens": 422145364.0, + "step": 16455 + }, + { + "epoch": 1.8071601142104106, + "grad_norm": 2.0536599159240723, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7143969535827637, + "num_tokens": 422174902.0, + "step": 16456 + }, + { + "epoch": 1.8072699319130243, + "grad_norm": 2.2411282062530518, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7139902114868164, + "num_tokens": 422200043.0, + "step": 16457 + }, + { + "epoch": 1.807379749615638, + "grad_norm": 2.1010632514953613, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7039686441421509, + "num_tokens": 422228628.0, + "step": 16458 + }, + { + "epoch": 1.8074895673182518, + "grad_norm": 1.947143793106079, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7423601746559143, + "num_tokens": 422258531.0, + "step": 16459 + }, + { + "epoch": 1.8075993850208654, + "grad_norm": 2.5889828205108643, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7522842288017273, + "num_tokens": 422277803.0, + "step": 16460 + }, + { + "epoch": 1.807709202723479, + "grad_norm": 2.2659623622894287, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7189385890960693, + "num_tokens": 422304933.0, + "step": 16461 + }, + { + "epoch": 1.8078190204260927, + "grad_norm": 2.369946241378784, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7458797693252563, + "num_tokens": 422327542.0, + "step": 16462 + }, + { + "epoch": 1.8079288381287064, + "grad_norm": 1.8766648769378662, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7177765369415283, + "num_tokens": 422361320.0, + "step": 16463 + }, + { + "epoch": 1.8080386558313202, + "grad_norm": 2.295565366744995, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7144132256507874, + "num_tokens": 422384132.0, + "step": 16464 + }, + { + "epoch": 1.8081484735339337, + "grad_norm": 2.105433940887451, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7254669666290283, + "num_tokens": 422410773.0, + "step": 16465 + }, + { + "epoch": 1.8082582912365472, + "grad_norm": 2.169846534729004, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7057425379753113, + "num_tokens": 422438036.0, + "step": 16466 + }, + { + "epoch": 1.808368108939161, + "grad_norm": 1.8917464017868042, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7167474031448364, + "num_tokens": 422472644.0, + "step": 16467 + }, + { + "epoch": 1.8084779266417748, + "grad_norm": 2.068939447402954, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.70273756980896, + "num_tokens": 422504222.0, + "step": 16468 + }, + { + "epoch": 1.8085877443443883, + "grad_norm": 2.1008572578430176, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7199573516845703, + "num_tokens": 422533636.0, + "step": 16469 + }, + { + "epoch": 1.8086975620470018, + "grad_norm": 2.358222007751465, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7251049876213074, + "num_tokens": 422558628.0, + "step": 16470 + }, + { + "epoch": 1.8088073797496156, + "grad_norm": 1.9251277446746826, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.710695207118988, + "num_tokens": 422591619.0, + "step": 16471 + }, + { + "epoch": 1.8089171974522293, + "grad_norm": 2.3297393321990967, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7022014260292053, + "num_tokens": 422616061.0, + "step": 16472 + }, + { + "epoch": 1.809027015154843, + "grad_norm": 2.3905153274536133, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7237663865089417, + "num_tokens": 422640816.0, + "step": 16473 + }, + { + "epoch": 1.8091368328574566, + "grad_norm": 2.4076457023620605, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6977020502090454, + "num_tokens": 422665329.0, + "step": 16474 + }, + { + "epoch": 1.8092466505600702, + "grad_norm": 2.2553038597106934, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7200229167938232, + "num_tokens": 422689339.0, + "step": 16475 + }, + { + "epoch": 1.809356468262684, + "grad_norm": 2.250075101852417, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7044192552566528, + "num_tokens": 422713686.0, + "step": 16476 + }, + { + "epoch": 1.8094662859652977, + "grad_norm": 1.8977605104446411, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6894989013671875, + "num_tokens": 422747644.0, + "step": 16477 + }, + { + "epoch": 1.8095761036679112, + "grad_norm": 2.4828782081604004, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7210039496421814, + "num_tokens": 422769536.0, + "step": 16478 + }, + { + "epoch": 1.809685921370525, + "grad_norm": 2.3482158184051514, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7237058281898499, + "num_tokens": 422793544.0, + "step": 16479 + }, + { + "epoch": 1.8097957390731385, + "grad_norm": 2.4785380363464355, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7392610311508179, + "num_tokens": 422816107.0, + "step": 16480 + }, + { + "epoch": 1.8099055567757523, + "grad_norm": 2.442345142364502, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7197384834289551, + "num_tokens": 422838947.0, + "step": 16481 + }, + { + "epoch": 1.810015374478366, + "grad_norm": 2.582965612411499, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7227970957756042, + "num_tokens": 422859421.0, + "step": 16482 + }, + { + "epoch": 1.8101251921809796, + "grad_norm": 2.0702760219573975, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7296534180641174, + "num_tokens": 422888951.0, + "step": 16483 + }, + { + "epoch": 1.810235009883593, + "grad_norm": 2.434980630874634, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7318279147148132, + "num_tokens": 422911904.0, + "step": 16484 + }, + { + "epoch": 1.8103448275862069, + "grad_norm": 2.2595837116241455, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7176442742347717, + "num_tokens": 422937451.0, + "step": 16485 + }, + { + "epoch": 1.8104546452888206, + "grad_norm": 2.4089019298553467, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.733497142791748, + "num_tokens": 422960467.0, + "step": 16486 + }, + { + "epoch": 1.8105644629914344, + "grad_norm": 2.3098227977752686, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7094658613204956, + "num_tokens": 422985091.0, + "step": 16487 + }, + { + "epoch": 1.810674280694048, + "grad_norm": 2.5552315711975098, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7246633172035217, + "num_tokens": 423005467.0, + "step": 16488 + }, + { + "epoch": 1.8107840983966614, + "grad_norm": 1.9014524221420288, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7176865339279175, + "num_tokens": 423040333.0, + "step": 16489 + }, + { + "epoch": 1.8108939160992752, + "grad_norm": 2.109187126159668, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7104930877685547, + "num_tokens": 423068575.0, + "step": 16490 + }, + { + "epoch": 1.811003733801889, + "grad_norm": 2.0080018043518066, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7152525782585144, + "num_tokens": 423096981.0, + "step": 16491 + }, + { + "epoch": 1.8111135515045025, + "grad_norm": 2.288853168487549, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7114905118942261, + "num_tokens": 423121061.0, + "step": 16492 + }, + { + "epoch": 1.8112233692071162, + "grad_norm": 2.208554744720459, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7254654765129089, + "num_tokens": 423147203.0, + "step": 16493 + }, + { + "epoch": 1.8113331869097298, + "grad_norm": 2.735832929611206, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7306984663009644, + "num_tokens": 423165478.0, + "step": 16494 + }, + { + "epoch": 1.8114430046123435, + "grad_norm": 2.4076545238494873, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7110191583633423, + "num_tokens": 423188730.0, + "step": 16495 + }, + { + "epoch": 1.8115528223149573, + "grad_norm": 2.058790445327759, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.730614185333252, + "num_tokens": 423218788.0, + "step": 16496 + }, + { + "epoch": 1.8116626400175708, + "grad_norm": 2.5314652919769287, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7222540378570557, + "num_tokens": 423238964.0, + "step": 16497 + }, + { + "epoch": 1.8117724577201844, + "grad_norm": 2.4117820262908936, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7365351915359497, + "num_tokens": 423260139.0, + "step": 16498 + }, + { + "epoch": 1.8118822754227981, + "grad_norm": 2.004014730453491, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7232488393783569, + "num_tokens": 423290963.0, + "step": 16499 + }, + { + "epoch": 1.8119920931254119, + "grad_norm": 2.241100549697876, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7289538979530334, + "num_tokens": 423315510.0, + "step": 16500 + }, + { + "epoch": 1.8121019108280256, + "grad_norm": 2.1119840145111084, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7183496952056885, + "num_tokens": 423343029.0, + "step": 16501 + }, + { + "epoch": 1.8122117285306392, + "grad_norm": 2.481484889984131, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7182639837265015, + "num_tokens": 423367421.0, + "step": 16502 + }, + { + "epoch": 1.8123215462332527, + "grad_norm": 1.941343069076538, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7086578607559204, + "num_tokens": 423402228.0, + "step": 16503 + }, + { + "epoch": 1.8124313639358665, + "grad_norm": 2.3977394104003906, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7243375778198242, + "num_tokens": 423425804.0, + "step": 16504 + }, + { + "epoch": 1.8125411816384802, + "grad_norm": 2.1483235359191895, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7001011967658997, + "num_tokens": 423456005.0, + "step": 16505 + }, + { + "epoch": 1.8126509993410937, + "grad_norm": 2.358271598815918, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6998850107192993, + "num_tokens": 423482604.0, + "step": 16506 + }, + { + "epoch": 1.8127608170437073, + "grad_norm": 2.3190762996673584, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7042969465255737, + "num_tokens": 423507654.0, + "step": 16507 + }, + { + "epoch": 1.812870634746321, + "grad_norm": 2.6591570377349854, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7198469638824463, + "num_tokens": 423527820.0, + "step": 16508 + }, + { + "epoch": 1.8129804524489348, + "grad_norm": 2.16573429107666, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7150415182113647, + "num_tokens": 423554864.0, + "step": 16509 + }, + { + "epoch": 1.8130902701515486, + "grad_norm": 1.9443081617355347, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7199581861495972, + "num_tokens": 423589435.0, + "step": 16510 + }, + { + "epoch": 1.813200087854162, + "grad_norm": 2.4694626331329346, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7171725034713745, + "num_tokens": 423611848.0, + "step": 16511 + }, + { + "epoch": 1.8133099055567756, + "grad_norm": 2.408066511154175, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.773284375667572, + "num_tokens": 423632163.0, + "step": 16512 + }, + { + "epoch": 1.8134197232593894, + "grad_norm": 2.3629212379455566, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7161611318588257, + "num_tokens": 423656473.0, + "step": 16513 + }, + { + "epoch": 1.8135295409620031, + "grad_norm": 2.1716578006744385, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7124115824699402, + "num_tokens": 423683966.0, + "step": 16514 + }, + { + "epoch": 1.813639358664617, + "grad_norm": 2.0389456748962402, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.733391523361206, + "num_tokens": 423715165.0, + "step": 16515 + }, + { + "epoch": 1.8137491763672304, + "grad_norm": 2.296926498413086, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7223057746887207, + "num_tokens": 423739160.0, + "step": 16516 + }, + { + "epoch": 1.813858994069844, + "grad_norm": 2.6585731506347656, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7226768732070923, + "num_tokens": 423759270.0, + "step": 16517 + }, + { + "epoch": 1.8139688117724577, + "grad_norm": 2.1565423011779785, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7287280559539795, + "num_tokens": 423785242.0, + "step": 16518 + }, + { + "epoch": 1.8140786294750715, + "grad_norm": 2.205101490020752, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7091568112373352, + "num_tokens": 423811798.0, + "step": 16519 + }, + { + "epoch": 1.814188447177685, + "grad_norm": 2.35260272026062, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7290350198745728, + "num_tokens": 423834985.0, + "step": 16520 + }, + { + "epoch": 1.8142982648802986, + "grad_norm": 2.220306634902954, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7448972463607788, + "num_tokens": 423858546.0, + "step": 16521 + }, + { + "epoch": 1.8144080825829123, + "grad_norm": 2.0619969367980957, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7103791236877441, + "num_tokens": 423887823.0, + "step": 16522 + }, + { + "epoch": 1.814517900285526, + "grad_norm": 2.405245780944824, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.74043869972229, + "num_tokens": 423909267.0, + "step": 16523 + }, + { + "epoch": 1.8146277179881398, + "grad_norm": 2.5767881870269775, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7226231098175049, + "num_tokens": 423930490.0, + "step": 16524 + }, + { + "epoch": 1.8147375356907534, + "grad_norm": 2.488980531692505, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.728557288646698, + "num_tokens": 423950526.0, + "step": 16525 + }, + { + "epoch": 1.814847353393367, + "grad_norm": 2.5677480697631836, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7280033826828003, + "num_tokens": 423971352.0, + "step": 16526 + }, + { + "epoch": 1.8149571710959806, + "grad_norm": 2.2147467136383057, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7017597556114197, + "num_tokens": 423999398.0, + "step": 16527 + }, + { + "epoch": 1.8150669887985944, + "grad_norm": 2.571800470352173, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7212085723876953, + "num_tokens": 424019670.0, + "step": 16528 + }, + { + "epoch": 1.815176806501208, + "grad_norm": 2.0878663063049316, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7138707041740417, + "num_tokens": 424047994.0, + "step": 16529 + }, + { + "epoch": 1.8152866242038217, + "grad_norm": 2.1620829105377197, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7347510457038879, + "num_tokens": 424075569.0, + "step": 16530 + }, + { + "epoch": 1.8153964419064352, + "grad_norm": 2.564373731613159, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7102618217468262, + "num_tokens": 424097191.0, + "step": 16531 + }, + { + "epoch": 1.815506259609049, + "grad_norm": 1.9765559434890747, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7113158106803894, + "num_tokens": 424128559.0, + "step": 16532 + }, + { + "epoch": 1.8156160773116627, + "grad_norm": 2.2285919189453125, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7234069108963013, + "num_tokens": 424153693.0, + "step": 16533 + }, + { + "epoch": 1.8157258950142763, + "grad_norm": 2.055230140686035, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7066115140914917, + "num_tokens": 424182264.0, + "step": 16534 + }, + { + "epoch": 1.8158357127168898, + "grad_norm": 2.1241354942321777, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7194889783859253, + "num_tokens": 424210134.0, + "step": 16535 + }, + { + "epoch": 1.8159455304195036, + "grad_norm": 2.362985372543335, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6856261491775513, + "num_tokens": 424235003.0, + "step": 16536 + }, + { + "epoch": 1.8160553481221173, + "grad_norm": 2.091920852661133, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7069368958473206, + "num_tokens": 424266445.0, + "step": 16537 + }, + { + "epoch": 1.816165165824731, + "grad_norm": 2.319444179534912, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7206745743751526, + "num_tokens": 424290614.0, + "step": 16538 + }, + { + "epoch": 1.8162749835273446, + "grad_norm": 2.082601308822632, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7134189605712891, + "num_tokens": 424320728.0, + "step": 16539 + }, + { + "epoch": 1.8163848012299582, + "grad_norm": 2.4029383659362793, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7279561758041382, + "num_tokens": 424342377.0, + "step": 16540 + }, + { + "epoch": 1.816494618932572, + "grad_norm": 1.9220579862594604, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7127962112426758, + "num_tokens": 424377933.0, + "step": 16541 + }, + { + "epoch": 1.8166044366351857, + "grad_norm": 2.408499002456665, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.729265034198761, + "num_tokens": 424400131.0, + "step": 16542 + }, + { + "epoch": 1.8167142543377992, + "grad_norm": 2.3166720867156982, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7198190689086914, + "num_tokens": 424425412.0, + "step": 16543 + }, + { + "epoch": 1.816824072040413, + "grad_norm": 1.9204779863357544, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6946443319320679, + "num_tokens": 424459700.0, + "step": 16544 + }, + { + "epoch": 1.8169338897430265, + "grad_norm": 2.0344297885894775, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7047411203384399, + "num_tokens": 424490569.0, + "step": 16545 + }, + { + "epoch": 1.8170437074456403, + "grad_norm": 2.38395094871521, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7254201769828796, + "num_tokens": 424513158.0, + "step": 16546 + }, + { + "epoch": 1.817153525148254, + "grad_norm": 2.09798002243042, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7154905796051025, + "num_tokens": 424540974.0, + "step": 16547 + }, + { + "epoch": 1.8172633428508675, + "grad_norm": 2.5259547233581543, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7057396173477173, + "num_tokens": 424564118.0, + "step": 16548 + }, + { + "epoch": 1.817373160553481, + "grad_norm": 2.1858673095703125, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7183966636657715, + "num_tokens": 424590379.0, + "step": 16549 + }, + { + "epoch": 1.8174829782560948, + "grad_norm": 2.500290632247925, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7244560718536377, + "num_tokens": 424612719.0, + "step": 16550 + }, + { + "epoch": 1.8175927959587086, + "grad_norm": 2.0481693744659424, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7093492150306702, + "num_tokens": 424643049.0, + "step": 16551 + }, + { + "epoch": 1.8177026136613224, + "grad_norm": 2.400103807449341, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7297097444534302, + "num_tokens": 424666332.0, + "step": 16552 + }, + { + "epoch": 1.8178124313639359, + "grad_norm": 2.0353143215179443, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7308389544487, + "num_tokens": 424696789.0, + "step": 16553 + }, + { + "epoch": 1.8179222490665494, + "grad_norm": 2.2006213665008545, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7535065412521362, + "num_tokens": 424719887.0, + "step": 16554 + }, + { + "epoch": 1.8180320667691632, + "grad_norm": 2.5143682956695557, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7013418674468994, + "num_tokens": 424745144.0, + "step": 16555 + }, + { + "epoch": 1.818141884471777, + "grad_norm": 2.3168201446533203, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7062450051307678, + "num_tokens": 424770160.0, + "step": 16556 + }, + { + "epoch": 1.8182517021743905, + "grad_norm": 2.4662888050079346, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7277028560638428, + "num_tokens": 424790355.0, + "step": 16557 + }, + { + "epoch": 1.818361519877004, + "grad_norm": 2.02135968208313, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6904539465904236, + "num_tokens": 424822425.0, + "step": 16558 + }, + { + "epoch": 1.8184713375796178, + "grad_norm": 2.251075267791748, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.728165328502655, + "num_tokens": 424847984.0, + "step": 16559 + }, + { + "epoch": 1.8185811552822315, + "grad_norm": 2.2042417526245117, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7055725455284119, + "num_tokens": 424874256.0, + "step": 16560 + }, + { + "epoch": 1.8186909729848453, + "grad_norm": 2.2674999237060547, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7232615947723389, + "num_tokens": 424897639.0, + "step": 16561 + }, + { + "epoch": 1.8188007906874588, + "grad_norm": 2.3535144329071045, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7128689885139465, + "num_tokens": 424921013.0, + "step": 16562 + }, + { + "epoch": 1.8189106083900723, + "grad_norm": 2.4415953159332275, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7076294422149658, + "num_tokens": 424945417.0, + "step": 16563 + }, + { + "epoch": 1.819020426092686, + "grad_norm": 2.0720317363739014, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7148104906082153, + "num_tokens": 424974828.0, + "step": 16564 + }, + { + "epoch": 1.8191302437952999, + "grad_norm": 2.2910735607147217, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7268001437187195, + "num_tokens": 424998304.0, + "step": 16565 + }, + { + "epoch": 1.8192400614979136, + "grad_norm": 2.4818458557128906, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7504207491874695, + "num_tokens": 425019760.0, + "step": 16566 + }, + { + "epoch": 1.8193498792005272, + "grad_norm": 2.068455219268799, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7282265424728394, + "num_tokens": 425049021.0, + "step": 16567 + }, + { + "epoch": 1.8194596969031407, + "grad_norm": 2.2473342418670654, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7209256887435913, + "num_tokens": 425071945.0, + "step": 16568 + }, + { + "epoch": 1.8195695146057544, + "grad_norm": 2.34892201423645, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7226893901824951, + "num_tokens": 425095184.0, + "step": 16569 + }, + { + "epoch": 1.8196793323083682, + "grad_norm": 2.5441958904266357, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7031022310256958, + "num_tokens": 425116835.0, + "step": 16570 + }, + { + "epoch": 1.8197891500109817, + "grad_norm": 2.7870194911956787, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7202345132827759, + "num_tokens": 425134663.0, + "step": 16571 + }, + { + "epoch": 1.8198989677135953, + "grad_norm": 2.3046152591705322, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6944753527641296, + "num_tokens": 425159851.0, + "step": 16572 + }, + { + "epoch": 1.820008785416209, + "grad_norm": 2.4579966068267822, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7315465211868286, + "num_tokens": 425181504.0, + "step": 16573 + }, + { + "epoch": 1.8201186031188228, + "grad_norm": 2.23929500579834, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7085164785385132, + "num_tokens": 425207740.0, + "step": 16574 + }, + { + "epoch": 1.8202284208214365, + "grad_norm": 2.159287452697754, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7277535796165466, + "num_tokens": 425233192.0, + "step": 16575 + }, + { + "epoch": 1.82033823852405, + "grad_norm": 2.5007102489471436, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7168530225753784, + "num_tokens": 425253818.0, + "step": 16576 + }, + { + "epoch": 1.8204480562266636, + "grad_norm": 2.411081552505493, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7194383144378662, + "num_tokens": 425278868.0, + "step": 16577 + }, + { + "epoch": 1.8205578739292774, + "grad_norm": 2.415231466293335, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7037716507911682, + "num_tokens": 425301867.0, + "step": 16578 + }, + { + "epoch": 1.8206676916318911, + "grad_norm": 2.0766584873199463, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7027794122695923, + "num_tokens": 425330670.0, + "step": 16579 + }, + { + "epoch": 1.8207775093345049, + "grad_norm": 2.303344964981079, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7338212728500366, + "num_tokens": 425353921.0, + "step": 16580 + }, + { + "epoch": 1.8208873270371184, + "grad_norm": 2.520456552505493, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7501219511032104, + "num_tokens": 425373731.0, + "step": 16581 + }, + { + "epoch": 1.820997144739732, + "grad_norm": 2.4467244148254395, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7245000600814819, + "num_tokens": 425396012.0, + "step": 16582 + }, + { + "epoch": 1.8211069624423457, + "grad_norm": 2.1376712322235107, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7334447503089905, + "num_tokens": 425423782.0, + "step": 16583 + }, + { + "epoch": 1.8212167801449595, + "grad_norm": 2.067244291305542, + "learning_rate": 1e-06, + "loss": 1.0645, + "mean_token_accuracy": 0.6829420924186707, + "num_tokens": 425453426.0, + "step": 16584 + }, + { + "epoch": 1.821326597847573, + "grad_norm": 2.2913858890533447, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7328263521194458, + "num_tokens": 425477337.0, + "step": 16585 + }, + { + "epoch": 1.8214364155501865, + "grad_norm": 2.111151933670044, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7188380360603333, + "num_tokens": 425505988.0, + "step": 16586 + }, + { + "epoch": 1.8215462332528003, + "grad_norm": 2.272292375564575, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7396447062492371, + "num_tokens": 425530615.0, + "step": 16587 + }, + { + "epoch": 1.821656050955414, + "grad_norm": 1.9292346239089966, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7072617411613464, + "num_tokens": 425565068.0, + "step": 16588 + }, + { + "epoch": 1.8217658686580278, + "grad_norm": 2.0768587589263916, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7198696136474609, + "num_tokens": 425593869.0, + "step": 16589 + }, + { + "epoch": 1.8218756863606413, + "grad_norm": 2.313474416732788, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7257397770881653, + "num_tokens": 425618298.0, + "step": 16590 + }, + { + "epoch": 1.8219855040632549, + "grad_norm": 1.9641540050506592, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7088534832000732, + "num_tokens": 425647367.0, + "step": 16591 + }, + { + "epoch": 1.8220953217658686, + "grad_norm": 2.1750268936157227, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7232494354248047, + "num_tokens": 425674261.0, + "step": 16592 + }, + { + "epoch": 1.8222051394684824, + "grad_norm": 1.9331892728805542, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7193987369537354, + "num_tokens": 425707203.0, + "step": 16593 + }, + { + "epoch": 1.822314957171096, + "grad_norm": 2.033287763595581, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7139559984207153, + "num_tokens": 425738622.0, + "step": 16594 + }, + { + "epoch": 1.8224247748737097, + "grad_norm": 2.342128038406372, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7300941944122314, + "num_tokens": 425761889.0, + "step": 16595 + }, + { + "epoch": 1.8225345925763232, + "grad_norm": 2.245581865310669, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7150927782058716, + "num_tokens": 425788184.0, + "step": 16596 + }, + { + "epoch": 1.822644410278937, + "grad_norm": 2.2073633670806885, + "learning_rate": 1e-06, + "loss": 1.0943, + "mean_token_accuracy": 0.6840673685073853, + "num_tokens": 425816980.0, + "step": 16597 + }, + { + "epoch": 1.8227542279815507, + "grad_norm": 2.199826240539551, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.713442325592041, + "num_tokens": 425842890.0, + "step": 16598 + }, + { + "epoch": 1.8228640456841643, + "grad_norm": 2.0696442127227783, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7238768935203552, + "num_tokens": 425872417.0, + "step": 16599 + }, + { + "epoch": 1.8229738633867778, + "grad_norm": 2.19598388671875, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7163224220275879, + "num_tokens": 425897548.0, + "step": 16600 + }, + { + "epoch": 1.8230836810893916, + "grad_norm": 2.169639825820923, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7528513669967651, + "num_tokens": 425925507.0, + "step": 16601 + }, + { + "epoch": 1.8231934987920053, + "grad_norm": 2.634068012237549, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7145023345947266, + "num_tokens": 425946157.0, + "step": 16602 + }, + { + "epoch": 1.823303316494619, + "grad_norm": 2.1077499389648438, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7267104387283325, + "num_tokens": 425973725.0, + "step": 16603 + }, + { + "epoch": 1.8234131341972326, + "grad_norm": 2.1889708042144775, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7259988188743591, + "num_tokens": 425999221.0, + "step": 16604 + }, + { + "epoch": 1.8235229518998461, + "grad_norm": 2.118436098098755, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.706373929977417, + "num_tokens": 426026633.0, + "step": 16605 + }, + { + "epoch": 1.82363276960246, + "grad_norm": 2.2545278072357178, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7285611629486084, + "num_tokens": 426051192.0, + "step": 16606 + }, + { + "epoch": 1.8237425873050737, + "grad_norm": 2.390171527862549, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7114025354385376, + "num_tokens": 426074241.0, + "step": 16607 + }, + { + "epoch": 1.8238524050076872, + "grad_norm": 2.1189210414886475, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.706631600856781, + "num_tokens": 426100844.0, + "step": 16608 + }, + { + "epoch": 1.823962222710301, + "grad_norm": 2.4804329872131348, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7153007388114929, + "num_tokens": 426122169.0, + "step": 16609 + }, + { + "epoch": 1.8240720404129145, + "grad_norm": 2.2169551849365234, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7207698822021484, + "num_tokens": 426147580.0, + "step": 16610 + }, + { + "epoch": 1.8241818581155282, + "grad_norm": 2.5885488986968994, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6949650645256042, + "num_tokens": 426168726.0, + "step": 16611 + }, + { + "epoch": 1.824291675818142, + "grad_norm": 2.2746739387512207, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7263954877853394, + "num_tokens": 426192474.0, + "step": 16612 + }, + { + "epoch": 1.8244014935207555, + "grad_norm": 2.137284517288208, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7272175550460815, + "num_tokens": 426220536.0, + "step": 16613 + }, + { + "epoch": 1.824511311223369, + "grad_norm": 2.0785999298095703, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7198998928070068, + "num_tokens": 426248346.0, + "step": 16614 + }, + { + "epoch": 1.8246211289259828, + "grad_norm": 2.129163980484009, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.744659960269928, + "num_tokens": 426275203.0, + "step": 16615 + }, + { + "epoch": 1.8247309466285966, + "grad_norm": 2.1447181701660156, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7328507304191589, + "num_tokens": 426300711.0, + "step": 16616 + }, + { + "epoch": 1.8248407643312103, + "grad_norm": 2.1651253700256348, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7172774076461792, + "num_tokens": 426328366.0, + "step": 16617 + }, + { + "epoch": 1.8249505820338239, + "grad_norm": 2.0534510612487793, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7018341422080994, + "num_tokens": 426360437.0, + "step": 16618 + }, + { + "epoch": 1.8250603997364374, + "grad_norm": 1.98622465133667, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6998929977416992, + "num_tokens": 426391572.0, + "step": 16619 + }, + { + "epoch": 1.8251702174390512, + "grad_norm": 2.2142045497894287, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7252906560897827, + "num_tokens": 426419049.0, + "step": 16620 + }, + { + "epoch": 1.825280035141665, + "grad_norm": 2.319791316986084, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7095124125480652, + "num_tokens": 426442505.0, + "step": 16621 + }, + { + "epoch": 1.8253898528442785, + "grad_norm": 2.2157628536224365, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7326358556747437, + "num_tokens": 426468310.0, + "step": 16622 + }, + { + "epoch": 1.825499670546892, + "grad_norm": 2.4796841144561768, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7181406617164612, + "num_tokens": 426491658.0, + "step": 16623 + }, + { + "epoch": 1.8256094882495058, + "grad_norm": 2.5133931636810303, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7596191763877869, + "num_tokens": 426510963.0, + "step": 16624 + }, + { + "epoch": 1.8257193059521195, + "grad_norm": 2.2417757511138916, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7232412099838257, + "num_tokens": 426537680.0, + "step": 16625 + }, + { + "epoch": 1.8258291236547333, + "grad_norm": 2.355100631713867, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7061286568641663, + "num_tokens": 426563309.0, + "step": 16626 + }, + { + "epoch": 1.8259389413573468, + "grad_norm": 2.1495866775512695, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.711056113243103, + "num_tokens": 426591072.0, + "step": 16627 + }, + { + "epoch": 1.8260487590599603, + "grad_norm": 2.167017936706543, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7335003614425659, + "num_tokens": 426617886.0, + "step": 16628 + }, + { + "epoch": 1.826158576762574, + "grad_norm": 2.1517012119293213, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7112212181091309, + "num_tokens": 426644934.0, + "step": 16629 + }, + { + "epoch": 1.8262683944651878, + "grad_norm": 2.1606838703155518, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7129189968109131, + "num_tokens": 426670837.0, + "step": 16630 + }, + { + "epoch": 1.8263782121678016, + "grad_norm": 2.2013089656829834, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7041467428207397, + "num_tokens": 426700076.0, + "step": 16631 + }, + { + "epoch": 1.8264880298704151, + "grad_norm": 2.351104497909546, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7049775719642639, + "num_tokens": 426724314.0, + "step": 16632 + }, + { + "epoch": 1.8265978475730287, + "grad_norm": 2.0892016887664795, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7173482775688171, + "num_tokens": 426752225.0, + "step": 16633 + }, + { + "epoch": 1.8267076652756424, + "grad_norm": 2.0011134147644043, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7215203046798706, + "num_tokens": 426783661.0, + "step": 16634 + }, + { + "epoch": 1.8268174829782562, + "grad_norm": 2.4320058822631836, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.717853307723999, + "num_tokens": 426807379.0, + "step": 16635 + }, + { + "epoch": 1.8269273006808697, + "grad_norm": 2.4183270931243896, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.714495062828064, + "num_tokens": 426830164.0, + "step": 16636 + }, + { + "epoch": 1.8270371183834833, + "grad_norm": 2.3069684505462646, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7028102874755859, + "num_tokens": 426854431.0, + "step": 16637 + }, + { + "epoch": 1.827146936086097, + "grad_norm": 1.9618850946426392, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7101188898086548, + "num_tokens": 426885546.0, + "step": 16638 + }, + { + "epoch": 1.8272567537887108, + "grad_norm": 2.0497348308563232, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7306009531021118, + "num_tokens": 426913069.0, + "step": 16639 + }, + { + "epoch": 1.8273665714913245, + "grad_norm": 2.0067036151885986, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7065501809120178, + "num_tokens": 426944400.0, + "step": 16640 + }, + { + "epoch": 1.827476389193938, + "grad_norm": 2.5583338737487793, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7442139983177185, + "num_tokens": 426962536.0, + "step": 16641 + }, + { + "epoch": 1.8275862068965516, + "grad_norm": 2.262920379638672, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7356323599815369, + "num_tokens": 426986650.0, + "step": 16642 + }, + { + "epoch": 1.8276960245991654, + "grad_norm": 2.312174081802368, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7045786380767822, + "num_tokens": 427012215.0, + "step": 16643 + }, + { + "epoch": 1.8278058423017791, + "grad_norm": 1.952963948249817, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6911922693252563, + "num_tokens": 427045657.0, + "step": 16644 + }, + { + "epoch": 1.8279156600043929, + "grad_norm": 2.4888792037963867, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7591026425361633, + "num_tokens": 427066563.0, + "step": 16645 + }, + { + "epoch": 1.8280254777070064, + "grad_norm": 2.09161639213562, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7391431927680969, + "num_tokens": 427093701.0, + "step": 16646 + }, + { + "epoch": 1.82813529540962, + "grad_norm": 2.412768840789795, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7116215825080872, + "num_tokens": 427115595.0, + "step": 16647 + }, + { + "epoch": 1.8282451131122337, + "grad_norm": 2.3088603019714355, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7102785706520081, + "num_tokens": 427136615.0, + "step": 16648 + }, + { + "epoch": 1.8283549308148475, + "grad_norm": 2.1855921745300293, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7268326878547668, + "num_tokens": 427161864.0, + "step": 16649 + }, + { + "epoch": 1.828464748517461, + "grad_norm": 2.4047648906707764, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7015494704246521, + "num_tokens": 427184257.0, + "step": 16650 + }, + { + "epoch": 1.8285745662200745, + "grad_norm": 2.133988618850708, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7177419066429138, + "num_tokens": 427212337.0, + "step": 16651 + }, + { + "epoch": 1.8286843839226883, + "grad_norm": 2.040799617767334, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7194133996963501, + "num_tokens": 427240909.0, + "step": 16652 + }, + { + "epoch": 1.828794201625302, + "grad_norm": 2.1933958530426025, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7335211634635925, + "num_tokens": 427266182.0, + "step": 16653 + }, + { + "epoch": 1.8289040193279158, + "grad_norm": 2.3639678955078125, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7141283750534058, + "num_tokens": 427290480.0, + "step": 16654 + }, + { + "epoch": 1.8290138370305293, + "grad_norm": 2.1285977363586426, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.721605122089386, + "num_tokens": 427319031.0, + "step": 16655 + }, + { + "epoch": 1.8291236547331429, + "grad_norm": 1.9867156744003296, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7125000953674316, + "num_tokens": 427349673.0, + "step": 16656 + }, + { + "epoch": 1.8292334724357566, + "grad_norm": 2.46002459526062, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7242082357406616, + "num_tokens": 427370493.0, + "step": 16657 + }, + { + "epoch": 1.8293432901383704, + "grad_norm": 2.445692777633667, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7050043344497681, + "num_tokens": 427396252.0, + "step": 16658 + }, + { + "epoch": 1.829453107840984, + "grad_norm": 2.8403735160827637, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.721136748790741, + "num_tokens": 427414194.0, + "step": 16659 + }, + { + "epoch": 1.8295629255435977, + "grad_norm": 2.2893893718719482, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7169306874275208, + "num_tokens": 427442770.0, + "step": 16660 + }, + { + "epoch": 1.8296727432462112, + "grad_norm": 2.4229977130889893, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7161588668823242, + "num_tokens": 427467497.0, + "step": 16661 + }, + { + "epoch": 1.829782560948825, + "grad_norm": 2.046049118041992, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.716771125793457, + "num_tokens": 427495814.0, + "step": 16662 + }, + { + "epoch": 1.8298923786514387, + "grad_norm": 2.131409168243408, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7327715158462524, + "num_tokens": 427521036.0, + "step": 16663 + }, + { + "epoch": 1.8300021963540523, + "grad_norm": 2.3138015270233154, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7079017758369446, + "num_tokens": 427544760.0, + "step": 16664 + }, + { + "epoch": 1.8301120140566658, + "grad_norm": 2.3217785358428955, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7127827405929565, + "num_tokens": 427569685.0, + "step": 16665 + }, + { + "epoch": 1.8302218317592795, + "grad_norm": 2.0862019062042236, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7151848077774048, + "num_tokens": 427598417.0, + "step": 16666 + }, + { + "epoch": 1.8303316494618933, + "grad_norm": 2.0371439456939697, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7417446970939636, + "num_tokens": 427628930.0, + "step": 16667 + }, + { + "epoch": 1.830441467164507, + "grad_norm": 2.2804009914398193, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7558363676071167, + "num_tokens": 427652897.0, + "step": 16668 + }, + { + "epoch": 1.8305512848671206, + "grad_norm": 2.212531328201294, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7298946380615234, + "num_tokens": 427676996.0, + "step": 16669 + }, + { + "epoch": 1.8306611025697341, + "grad_norm": 2.5178723335266113, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7267470359802246, + "num_tokens": 427697197.0, + "step": 16670 + }, + { + "epoch": 1.830770920272348, + "grad_norm": 1.996841549873352, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7220414876937866, + "num_tokens": 427725466.0, + "step": 16671 + }, + { + "epoch": 1.8308807379749616, + "grad_norm": 2.160426378250122, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7197169065475464, + "num_tokens": 427756031.0, + "step": 16672 + }, + { + "epoch": 1.8309905556775752, + "grad_norm": 1.982665777206421, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.6949334144592285, + "num_tokens": 427787502.0, + "step": 16673 + }, + { + "epoch": 1.831100373380189, + "grad_norm": 2.594700574874878, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7306711673736572, + "num_tokens": 427808951.0, + "step": 16674 + }, + { + "epoch": 1.8312101910828025, + "grad_norm": 2.384141683578491, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7486090660095215, + "num_tokens": 427831579.0, + "step": 16675 + }, + { + "epoch": 1.8313200087854162, + "grad_norm": 2.494847297668457, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7102336883544922, + "num_tokens": 427855695.0, + "step": 16676 + }, + { + "epoch": 1.83142982648803, + "grad_norm": 2.276562452316284, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7243376970291138, + "num_tokens": 427881056.0, + "step": 16677 + }, + { + "epoch": 1.8315396441906435, + "grad_norm": 2.5598251819610596, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7130663394927979, + "num_tokens": 427902898.0, + "step": 16678 + }, + { + "epoch": 1.831649461893257, + "grad_norm": 2.3523454666137695, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7147262096405029, + "num_tokens": 427926571.0, + "step": 16679 + }, + { + "epoch": 1.8317592795958708, + "grad_norm": 2.0223007202148438, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7343201041221619, + "num_tokens": 427955940.0, + "step": 16680 + }, + { + "epoch": 1.8318690972984846, + "grad_norm": 2.2227866649627686, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.713160514831543, + "num_tokens": 427981259.0, + "step": 16681 + }, + { + "epoch": 1.8319789150010983, + "grad_norm": 2.3748438358306885, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7403464913368225, + "num_tokens": 428004985.0, + "step": 16682 + }, + { + "epoch": 1.8320887327037119, + "grad_norm": 2.3222615718841553, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7274695038795471, + "num_tokens": 428026779.0, + "step": 16683 + }, + { + "epoch": 1.8321985504063254, + "grad_norm": 2.1598293781280518, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6955968737602234, + "num_tokens": 428054982.0, + "step": 16684 + }, + { + "epoch": 1.8323083681089392, + "grad_norm": 2.4734134674072266, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7152479887008667, + "num_tokens": 428076553.0, + "step": 16685 + }, + { + "epoch": 1.832418185811553, + "grad_norm": 2.3163833618164062, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7236249446868896, + "num_tokens": 428099449.0, + "step": 16686 + }, + { + "epoch": 1.8325280035141664, + "grad_norm": 2.099152088165283, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.709409773349762, + "num_tokens": 428127889.0, + "step": 16687 + }, + { + "epoch": 1.83263782121678, + "grad_norm": 2.399803400039673, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7390820384025574, + "num_tokens": 428149921.0, + "step": 16688 + }, + { + "epoch": 1.8327476389193937, + "grad_norm": 2.3610403537750244, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.727557897567749, + "num_tokens": 428171739.0, + "step": 16689 + }, + { + "epoch": 1.8328574566220075, + "grad_norm": 2.338390588760376, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7183427214622498, + "num_tokens": 428198176.0, + "step": 16690 + }, + { + "epoch": 1.8329672743246213, + "grad_norm": 2.276064872741699, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7051695585250854, + "num_tokens": 428223322.0, + "step": 16691 + }, + { + "epoch": 1.8330770920272348, + "grad_norm": 2.052912473678589, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7124987840652466, + "num_tokens": 428251830.0, + "step": 16692 + }, + { + "epoch": 1.8331869097298483, + "grad_norm": 2.1301991939544678, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7308098077774048, + "num_tokens": 428280061.0, + "step": 16693 + }, + { + "epoch": 1.833296727432462, + "grad_norm": 1.9336966276168823, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7098134160041809, + "num_tokens": 428313030.0, + "step": 16694 + }, + { + "epoch": 1.8334065451350758, + "grad_norm": 2.0992414951324463, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7325263023376465, + "num_tokens": 428338904.0, + "step": 16695 + }, + { + "epoch": 1.8335163628376896, + "grad_norm": 2.610380172729492, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7276828289031982, + "num_tokens": 428358482.0, + "step": 16696 + }, + { + "epoch": 1.8336261805403031, + "grad_norm": 2.0481514930725098, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7003408670425415, + "num_tokens": 428388886.0, + "step": 16697 + }, + { + "epoch": 1.8337359982429167, + "grad_norm": 2.3994076251983643, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7418837547302246, + "num_tokens": 428411669.0, + "step": 16698 + }, + { + "epoch": 1.8338458159455304, + "grad_norm": 2.1606743335723877, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7213367819786072, + "num_tokens": 428440836.0, + "step": 16699 + }, + { + "epoch": 1.8339556336481442, + "grad_norm": 2.2823598384857178, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7269813418388367, + "num_tokens": 428466983.0, + "step": 16700 + }, + { + "epoch": 1.8340654513507577, + "grad_norm": 2.1708202362060547, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7178263664245605, + "num_tokens": 428492310.0, + "step": 16701 + }, + { + "epoch": 1.8341752690533712, + "grad_norm": 2.640928268432617, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7231635451316833, + "num_tokens": 428512947.0, + "step": 16702 + }, + { + "epoch": 1.834285086755985, + "grad_norm": 2.246168375015259, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7142820358276367, + "num_tokens": 428538726.0, + "step": 16703 + }, + { + "epoch": 1.8343949044585988, + "grad_norm": 2.0981409549713135, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7278019189834595, + "num_tokens": 428566549.0, + "step": 16704 + }, + { + "epoch": 1.8345047221612125, + "grad_norm": 2.0149495601654053, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7117235660552979, + "num_tokens": 428598354.0, + "step": 16705 + }, + { + "epoch": 1.834614539863826, + "grad_norm": 2.2152743339538574, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.720096230506897, + "num_tokens": 428621696.0, + "step": 16706 + }, + { + "epoch": 1.8347243575664396, + "grad_norm": 2.393954277038574, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7334975004196167, + "num_tokens": 428644492.0, + "step": 16707 + }, + { + "epoch": 1.8348341752690533, + "grad_norm": 2.2040140628814697, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7123820781707764, + "num_tokens": 428672518.0, + "step": 16708 + }, + { + "epoch": 1.834943992971667, + "grad_norm": 2.2185730934143066, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7052404880523682, + "num_tokens": 428699115.0, + "step": 16709 + }, + { + "epoch": 1.8350538106742806, + "grad_norm": 2.2615725994110107, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.713570773601532, + "num_tokens": 428724008.0, + "step": 16710 + }, + { + "epoch": 1.8351636283768944, + "grad_norm": 2.4735920429229736, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7370079159736633, + "num_tokens": 428744890.0, + "step": 16711 + }, + { + "epoch": 1.835273446079508, + "grad_norm": 2.3887767791748047, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7235480546951294, + "num_tokens": 428768148.0, + "step": 16712 + }, + { + "epoch": 1.8353832637821217, + "grad_norm": 2.43082857131958, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.723431408405304, + "num_tokens": 428791375.0, + "step": 16713 + }, + { + "epoch": 1.8354930814847354, + "grad_norm": 2.4040682315826416, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7110576629638672, + "num_tokens": 428814824.0, + "step": 16714 + }, + { + "epoch": 1.835602899187349, + "grad_norm": 2.0470774173736572, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7225455045700073, + "num_tokens": 428843183.0, + "step": 16715 + }, + { + "epoch": 1.8357127168899625, + "grad_norm": 2.1858835220336914, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7159454822540283, + "num_tokens": 428869362.0, + "step": 16716 + }, + { + "epoch": 1.8358225345925763, + "grad_norm": 2.095614194869995, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.722479522228241, + "num_tokens": 428900161.0, + "step": 16717 + }, + { + "epoch": 1.83593235229519, + "grad_norm": 2.169908046722412, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7328425645828247, + "num_tokens": 428928040.0, + "step": 16718 + }, + { + "epoch": 1.8360421699978038, + "grad_norm": 1.978851079940796, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7023220062255859, + "num_tokens": 428963218.0, + "step": 16719 + }, + { + "epoch": 1.8361519877004173, + "grad_norm": 2.212564468383789, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7387617826461792, + "num_tokens": 428987352.0, + "step": 16720 + }, + { + "epoch": 1.8362618054030309, + "grad_norm": 2.2010498046875, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7367266416549683, + "num_tokens": 429012843.0, + "step": 16721 + }, + { + "epoch": 1.8363716231056446, + "grad_norm": 2.107724666595459, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7060650587081909, + "num_tokens": 429041062.0, + "step": 16722 + }, + { + "epoch": 1.8364814408082584, + "grad_norm": 2.179133415222168, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6990185976028442, + "num_tokens": 429068015.0, + "step": 16723 + }, + { + "epoch": 1.836591258510872, + "grad_norm": 2.181103229522705, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7144163846969604, + "num_tokens": 429097453.0, + "step": 16724 + }, + { + "epoch": 1.8367010762134857, + "grad_norm": 2.439072847366333, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7444741129875183, + "num_tokens": 429119335.0, + "step": 16725 + }, + { + "epoch": 1.8368108939160992, + "grad_norm": 2.6596522331237793, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7354797720909119, + "num_tokens": 429138822.0, + "step": 16726 + }, + { + "epoch": 1.836920711618713, + "grad_norm": 2.2922701835632324, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7269006967544556, + "num_tokens": 429162563.0, + "step": 16727 + }, + { + "epoch": 1.8370305293213267, + "grad_norm": 2.5906996726989746, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7553014755249023, + "num_tokens": 429182035.0, + "step": 16728 + }, + { + "epoch": 1.8371403470239402, + "grad_norm": 2.3604490756988525, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7347712516784668, + "num_tokens": 429206425.0, + "step": 16729 + }, + { + "epoch": 1.8372501647265538, + "grad_norm": 2.0710182189941406, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7241917252540588, + "num_tokens": 429238347.0, + "step": 16730 + }, + { + "epoch": 1.8373599824291675, + "grad_norm": 2.581094741821289, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7271819710731506, + "num_tokens": 429259329.0, + "step": 16731 + }, + { + "epoch": 1.8374698001317813, + "grad_norm": 2.1383097171783447, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7111750841140747, + "num_tokens": 429286360.0, + "step": 16732 + }, + { + "epoch": 1.837579617834395, + "grad_norm": 2.185826063156128, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7200560569763184, + "num_tokens": 429311171.0, + "step": 16733 + }, + { + "epoch": 1.8376894355370086, + "grad_norm": 2.5327682495117188, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7093926668167114, + "num_tokens": 429333080.0, + "step": 16734 + }, + { + "epoch": 1.8377992532396221, + "grad_norm": 2.2407360076904297, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7248035669326782, + "num_tokens": 429356983.0, + "step": 16735 + }, + { + "epoch": 1.8379090709422359, + "grad_norm": 2.3009026050567627, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7265681028366089, + "num_tokens": 429381874.0, + "step": 16736 + }, + { + "epoch": 1.8380188886448496, + "grad_norm": 2.069373607635498, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.716335654258728, + "num_tokens": 429411805.0, + "step": 16737 + }, + { + "epoch": 1.8381287063474632, + "grad_norm": 2.6755597591400146, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7315776348114014, + "num_tokens": 429431913.0, + "step": 16738 + }, + { + "epoch": 1.838238524050077, + "grad_norm": 2.0242671966552734, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7189515829086304, + "num_tokens": 429461469.0, + "step": 16739 + }, + { + "epoch": 1.8383483417526905, + "grad_norm": 2.1959269046783447, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7128573656082153, + "num_tokens": 429488768.0, + "step": 16740 + }, + { + "epoch": 1.8384581594553042, + "grad_norm": 2.075770854949951, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7261574268341064, + "num_tokens": 429517823.0, + "step": 16741 + }, + { + "epoch": 1.838567977157918, + "grad_norm": 2.365530490875244, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7402654886245728, + "num_tokens": 429539088.0, + "step": 16742 + }, + { + "epoch": 1.8386777948605315, + "grad_norm": 2.2220449447631836, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.704512357711792, + "num_tokens": 429565855.0, + "step": 16743 + }, + { + "epoch": 1.838787612563145, + "grad_norm": 2.411451578140259, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7242421507835388, + "num_tokens": 429588330.0, + "step": 16744 + }, + { + "epoch": 1.8388974302657588, + "grad_norm": 3.0671703815460205, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7547601461410522, + "num_tokens": 429603591.0, + "step": 16745 + }, + { + "epoch": 1.8390072479683726, + "grad_norm": 2.2969584465026855, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6965454816818237, + "num_tokens": 429632260.0, + "step": 16746 + }, + { + "epoch": 1.8391170656709863, + "grad_norm": 2.1511881351470947, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7258408665657043, + "num_tokens": 429658172.0, + "step": 16747 + }, + { + "epoch": 1.8392268833735999, + "grad_norm": 2.1013195514678955, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7282155752182007, + "num_tokens": 429688606.0, + "step": 16748 + }, + { + "epoch": 1.8393367010762134, + "grad_norm": 2.1351587772369385, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7266145944595337, + "num_tokens": 429715713.0, + "step": 16749 + }, + { + "epoch": 1.8394465187788271, + "grad_norm": 2.6340203285217285, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7395250201225281, + "num_tokens": 429735382.0, + "step": 16750 + }, + { + "epoch": 1.839556336481441, + "grad_norm": 2.0878987312316895, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.698068380355835, + "num_tokens": 429765963.0, + "step": 16751 + }, + { + "epoch": 1.8396661541840544, + "grad_norm": 2.1238601207733154, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7059609293937683, + "num_tokens": 429793395.0, + "step": 16752 + }, + { + "epoch": 1.839775971886668, + "grad_norm": 2.27478289604187, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7137136459350586, + "num_tokens": 429819274.0, + "step": 16753 + }, + { + "epoch": 1.8398857895892817, + "grad_norm": 2.267735004425049, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7278488278388977, + "num_tokens": 429845333.0, + "step": 16754 + }, + { + "epoch": 1.8399956072918955, + "grad_norm": 2.5859103202819824, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7405322194099426, + "num_tokens": 429864351.0, + "step": 16755 + }, + { + "epoch": 1.8401054249945092, + "grad_norm": 2.3084774017333984, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7319678068161011, + "num_tokens": 429886973.0, + "step": 16756 + }, + { + "epoch": 1.8402152426971228, + "grad_norm": 2.075946092605591, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7240407466888428, + "num_tokens": 429913069.0, + "step": 16757 + }, + { + "epoch": 1.8403250603997363, + "grad_norm": 2.275779962539673, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7131457328796387, + "num_tokens": 429940243.0, + "step": 16758 + }, + { + "epoch": 1.84043487810235, + "grad_norm": 2.329367160797119, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7200474739074707, + "num_tokens": 429964422.0, + "step": 16759 + }, + { + "epoch": 1.8405446958049638, + "grad_norm": 2.3580310344696045, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7088528275489807, + "num_tokens": 429987581.0, + "step": 16760 + }, + { + "epoch": 1.8406545135075776, + "grad_norm": 1.9201397895812988, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7134987115859985, + "num_tokens": 430020446.0, + "step": 16761 + }, + { + "epoch": 1.8407643312101911, + "grad_norm": 2.2830522060394287, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7132782936096191, + "num_tokens": 430046217.0, + "step": 16762 + }, + { + "epoch": 1.8408741489128047, + "grad_norm": 2.197640895843506, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7218232154846191, + "num_tokens": 430073308.0, + "step": 16763 + }, + { + "epoch": 1.8409839666154184, + "grad_norm": 2.2422409057617188, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7134106159210205, + "num_tokens": 430101614.0, + "step": 16764 + }, + { + "epoch": 1.8410937843180322, + "grad_norm": 2.480968713760376, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7305562496185303, + "num_tokens": 430123742.0, + "step": 16765 + }, + { + "epoch": 1.8412036020206457, + "grad_norm": 2.143710136413574, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7217775583267212, + "num_tokens": 430150135.0, + "step": 16766 + }, + { + "epoch": 1.8413134197232592, + "grad_norm": 1.978850245475769, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7331514358520508, + "num_tokens": 430182271.0, + "step": 16767 + }, + { + "epoch": 1.841423237425873, + "grad_norm": 2.4011306762695312, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7266754508018494, + "num_tokens": 430202758.0, + "step": 16768 + }, + { + "epoch": 1.8415330551284868, + "grad_norm": 2.135080575942993, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7135171890258789, + "num_tokens": 430232212.0, + "step": 16769 + }, + { + "epoch": 1.8416428728311005, + "grad_norm": 2.051805257797241, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7093409299850464, + "num_tokens": 430263628.0, + "step": 16770 + }, + { + "epoch": 1.841752690533714, + "grad_norm": 2.212482452392578, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7267460823059082, + "num_tokens": 430289978.0, + "step": 16771 + }, + { + "epoch": 1.8418625082363276, + "grad_norm": 2.4661409854888916, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7197955846786499, + "num_tokens": 430311942.0, + "step": 16772 + }, + { + "epoch": 1.8419723259389413, + "grad_norm": 2.3605055809020996, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7068332433700562, + "num_tokens": 430338356.0, + "step": 16773 + }, + { + "epoch": 1.842082143641555, + "grad_norm": 2.075544834136963, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7494993805885315, + "num_tokens": 430363392.0, + "step": 16774 + }, + { + "epoch": 1.8421919613441686, + "grad_norm": 1.9924854040145874, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7182987332344055, + "num_tokens": 430393568.0, + "step": 16775 + }, + { + "epoch": 1.8423017790467824, + "grad_norm": 2.3695104122161865, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7557247281074524, + "num_tokens": 430417158.0, + "step": 16776 + }, + { + "epoch": 1.842411596749396, + "grad_norm": 1.9725658893585205, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7278224229812622, + "num_tokens": 430447800.0, + "step": 16777 + }, + { + "epoch": 1.8425214144520097, + "grad_norm": 2.2163021564483643, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7316633462905884, + "num_tokens": 430472870.0, + "step": 16778 + }, + { + "epoch": 1.8426312321546234, + "grad_norm": 2.376875400543213, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7121363878250122, + "num_tokens": 430498700.0, + "step": 16779 + }, + { + "epoch": 1.842741049857237, + "grad_norm": 2.2286953926086426, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7283117771148682, + "num_tokens": 430524155.0, + "step": 16780 + }, + { + "epoch": 1.8428508675598505, + "grad_norm": 2.042902946472168, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7286188006401062, + "num_tokens": 430551944.0, + "step": 16781 + }, + { + "epoch": 1.8429606852624643, + "grad_norm": 2.0863747596740723, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7042840719223022, + "num_tokens": 430579526.0, + "step": 16782 + }, + { + "epoch": 1.843070502965078, + "grad_norm": 2.233059883117676, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7197071313858032, + "num_tokens": 430605984.0, + "step": 16783 + }, + { + "epoch": 1.8431803206676918, + "grad_norm": 2.438032388687134, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7356339693069458, + "num_tokens": 430627664.0, + "step": 16784 + }, + { + "epoch": 1.8432901383703053, + "grad_norm": 2.42629337310791, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7300370335578918, + "num_tokens": 430651260.0, + "step": 16785 + }, + { + "epoch": 1.8433999560729188, + "grad_norm": 2.1289000511169434, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7223631739616394, + "num_tokens": 430678258.0, + "step": 16786 + }, + { + "epoch": 1.8435097737755326, + "grad_norm": 2.126051425933838, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.6981582641601562, + "num_tokens": 430706090.0, + "step": 16787 + }, + { + "epoch": 1.8436195914781464, + "grad_norm": 2.2699384689331055, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7282149791717529, + "num_tokens": 430730716.0, + "step": 16788 + }, + { + "epoch": 1.84372940918076, + "grad_norm": 2.2118754386901855, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7339807152748108, + "num_tokens": 430755421.0, + "step": 16789 + }, + { + "epoch": 1.8438392268833736, + "grad_norm": 2.2898526191711426, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7347133159637451, + "num_tokens": 430779523.0, + "step": 16790 + }, + { + "epoch": 1.8439490445859872, + "grad_norm": 2.185429096221924, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7366564273834229, + "num_tokens": 430806462.0, + "step": 16791 + }, + { + "epoch": 1.844058862288601, + "grad_norm": 2.2710413932800293, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7272056341171265, + "num_tokens": 430829693.0, + "step": 16792 + }, + { + "epoch": 1.8441686799912147, + "grad_norm": 2.4148335456848145, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7091183066368103, + "num_tokens": 430853620.0, + "step": 16793 + }, + { + "epoch": 1.8442784976938282, + "grad_norm": 2.3407163619995117, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7311354875564575, + "num_tokens": 430877755.0, + "step": 16794 + }, + { + "epoch": 1.8443883153964418, + "grad_norm": 2.0505082607269287, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7089937925338745, + "num_tokens": 430907793.0, + "step": 16795 + }, + { + "epoch": 1.8444981330990555, + "grad_norm": 2.159999132156372, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7002785205841064, + "num_tokens": 430934411.0, + "step": 16796 + }, + { + "epoch": 1.8446079508016693, + "grad_norm": 1.8877733945846558, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7203066349029541, + "num_tokens": 430966569.0, + "step": 16797 + }, + { + "epoch": 1.844717768504283, + "grad_norm": 2.1529502868652344, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7005844116210938, + "num_tokens": 430996867.0, + "step": 16798 + }, + { + "epoch": 1.8448275862068966, + "grad_norm": 2.409921646118164, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7208271026611328, + "num_tokens": 431018546.0, + "step": 16799 + }, + { + "epoch": 1.84493740390951, + "grad_norm": 2.163947343826294, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6945406198501587, + "num_tokens": 431046989.0, + "step": 16800 + }, + { + "epoch": 1.8450472216121239, + "grad_norm": 2.458820343017578, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6937620639801025, + "num_tokens": 431069904.0, + "step": 16801 + }, + { + "epoch": 1.8451570393147376, + "grad_norm": 2.4439518451690674, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7201370596885681, + "num_tokens": 431090851.0, + "step": 16802 + }, + { + "epoch": 1.8452668570173512, + "grad_norm": 2.4440231323242188, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.715752899646759, + "num_tokens": 431113149.0, + "step": 16803 + }, + { + "epoch": 1.8453766747199647, + "grad_norm": 2.18457293510437, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7237226963043213, + "num_tokens": 431139928.0, + "step": 16804 + }, + { + "epoch": 1.8454864924225785, + "grad_norm": 2.1717278957366943, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7125993967056274, + "num_tokens": 431166142.0, + "step": 16805 + }, + { + "epoch": 1.8455963101251922, + "grad_norm": 2.0147321224212646, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7150512933731079, + "num_tokens": 431196643.0, + "step": 16806 + }, + { + "epoch": 1.845706127827806, + "grad_norm": 2.272435426712036, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7193615436553955, + "num_tokens": 431221636.0, + "step": 16807 + }, + { + "epoch": 1.8458159455304195, + "grad_norm": 2.2423007488250732, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7317730188369751, + "num_tokens": 431247985.0, + "step": 16808 + }, + { + "epoch": 1.845925763233033, + "grad_norm": 2.070772409439087, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7240096926689148, + "num_tokens": 431276061.0, + "step": 16809 + }, + { + "epoch": 1.8460355809356468, + "grad_norm": 2.070816993713379, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7081250548362732, + "num_tokens": 431305138.0, + "step": 16810 + }, + { + "epoch": 1.8461453986382605, + "grad_norm": 2.6388230323791504, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7470161318778992, + "num_tokens": 431324590.0, + "step": 16811 + }, + { + "epoch": 1.8462552163408743, + "grad_norm": 2.1210834980010986, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7105023860931396, + "num_tokens": 431352018.0, + "step": 16812 + }, + { + "epoch": 1.8463650340434878, + "grad_norm": 2.328139543533325, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7274137139320374, + "num_tokens": 431372898.0, + "step": 16813 + }, + { + "epoch": 1.8464748517461014, + "grad_norm": 2.2834181785583496, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7137222290039062, + "num_tokens": 431398032.0, + "step": 16814 + }, + { + "epoch": 1.8465846694487151, + "grad_norm": 2.188356399536133, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7095911502838135, + "num_tokens": 431426154.0, + "step": 16815 + }, + { + "epoch": 1.846694487151329, + "grad_norm": 2.1226208209991455, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7277086973190308, + "num_tokens": 431452262.0, + "step": 16816 + }, + { + "epoch": 1.8468043048539424, + "grad_norm": 2.1135363578796387, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7309616804122925, + "num_tokens": 431481948.0, + "step": 16817 + }, + { + "epoch": 1.846914122556556, + "grad_norm": 2.7947957515716553, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7288796901702881, + "num_tokens": 431499431.0, + "step": 16818 + }, + { + "epoch": 1.8470239402591697, + "grad_norm": 2.144662618637085, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7235836386680603, + "num_tokens": 431526218.0, + "step": 16819 + }, + { + "epoch": 1.8471337579617835, + "grad_norm": 2.5475833415985107, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7311704158782959, + "num_tokens": 431547081.0, + "step": 16820 + }, + { + "epoch": 1.8472435756643972, + "grad_norm": 2.3216605186462402, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7073878049850464, + "num_tokens": 431572761.0, + "step": 16821 + }, + { + "epoch": 1.8473533933670108, + "grad_norm": 2.189157247543335, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7393380403518677, + "num_tokens": 431597169.0, + "step": 16822 + }, + { + "epoch": 1.8474632110696243, + "grad_norm": 1.9972796440124512, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.721793532371521, + "num_tokens": 431626554.0, + "step": 16823 + }, + { + "epoch": 1.847573028772238, + "grad_norm": 2.057333469390869, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7371499538421631, + "num_tokens": 431654414.0, + "step": 16824 + }, + { + "epoch": 1.8476828464748518, + "grad_norm": 1.9644616842269897, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7045168876647949, + "num_tokens": 431686363.0, + "step": 16825 + }, + { + "epoch": 1.8477926641774656, + "grad_norm": 2.0381619930267334, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7386239767074585, + "num_tokens": 431715000.0, + "step": 16826 + }, + { + "epoch": 1.847902481880079, + "grad_norm": 2.1908414363861084, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7187529802322388, + "num_tokens": 431740229.0, + "step": 16827 + }, + { + "epoch": 1.8480122995826926, + "grad_norm": 2.5871646404266357, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7329392433166504, + "num_tokens": 431759869.0, + "step": 16828 + }, + { + "epoch": 1.8481221172853064, + "grad_norm": 2.083195686340332, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7342507839202881, + "num_tokens": 431786169.0, + "step": 16829 + }, + { + "epoch": 1.8482319349879202, + "grad_norm": 2.3217968940734863, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7386271357536316, + "num_tokens": 431809034.0, + "step": 16830 + }, + { + "epoch": 1.8483417526905337, + "grad_norm": 1.9793418645858765, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7099496126174927, + "num_tokens": 431841073.0, + "step": 16831 + }, + { + "epoch": 1.8484515703931472, + "grad_norm": 2.2354400157928467, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7184978127479553, + "num_tokens": 431866120.0, + "step": 16832 + }, + { + "epoch": 1.848561388095761, + "grad_norm": 2.114506721496582, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6781002283096313, + "num_tokens": 431896245.0, + "step": 16833 + }, + { + "epoch": 1.8486712057983747, + "grad_norm": 2.0597245693206787, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7161827683448792, + "num_tokens": 431925247.0, + "step": 16834 + }, + { + "epoch": 1.8487810235009885, + "grad_norm": 2.2362632751464844, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7192049026489258, + "num_tokens": 431950432.0, + "step": 16835 + }, + { + "epoch": 1.848890841203602, + "grad_norm": 2.052197217941284, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6973109841346741, + "num_tokens": 431981639.0, + "step": 16836 + }, + { + "epoch": 1.8490006589062156, + "grad_norm": 2.1657514572143555, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7020468711853027, + "num_tokens": 432009934.0, + "step": 16837 + }, + { + "epoch": 1.8491104766088293, + "grad_norm": 2.548708200454712, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.730792760848999, + "num_tokens": 432030478.0, + "step": 16838 + }, + { + "epoch": 1.849220294311443, + "grad_norm": 2.3281970024108887, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7247130870819092, + "num_tokens": 432055962.0, + "step": 16839 + }, + { + "epoch": 1.8493301120140566, + "grad_norm": 1.8991646766662598, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7025207877159119, + "num_tokens": 432090625.0, + "step": 16840 + }, + { + "epoch": 1.8494399297166704, + "grad_norm": 2.558241605758667, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7494404315948486, + "num_tokens": 432109964.0, + "step": 16841 + }, + { + "epoch": 1.849549747419284, + "grad_norm": 1.9241667985916138, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6995230913162231, + "num_tokens": 432146586.0, + "step": 16842 + }, + { + "epoch": 1.8496595651218977, + "grad_norm": 2.1868464946746826, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7234881520271301, + "num_tokens": 432171416.0, + "step": 16843 + }, + { + "epoch": 1.8497693828245114, + "grad_norm": 2.548140287399292, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7358123660087585, + "num_tokens": 432191296.0, + "step": 16844 + }, + { + "epoch": 1.849879200527125, + "grad_norm": 1.9640716314315796, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7273036241531372, + "num_tokens": 432221713.0, + "step": 16845 + }, + { + "epoch": 1.8499890182297385, + "grad_norm": 2.11806321144104, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6977421045303345, + "num_tokens": 432249338.0, + "step": 16846 + }, + { + "epoch": 1.8500988359323522, + "grad_norm": 2.1768484115600586, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7004925012588501, + "num_tokens": 432279028.0, + "step": 16847 + }, + { + "epoch": 1.850208653634966, + "grad_norm": 2.2622973918914795, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7175419330596924, + "num_tokens": 432305331.0, + "step": 16848 + }, + { + "epoch": 1.8503184713375798, + "grad_norm": 2.4988932609558105, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7263602018356323, + "num_tokens": 432326642.0, + "step": 16849 + }, + { + "epoch": 1.8504282890401933, + "grad_norm": 2.1260411739349365, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7283483743667603, + "num_tokens": 432355773.0, + "step": 16850 + }, + { + "epoch": 1.8505381067428068, + "grad_norm": 2.04618763923645, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7208758592605591, + "num_tokens": 432383866.0, + "step": 16851 + }, + { + "epoch": 1.8506479244454206, + "grad_norm": 2.412612199783325, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.740345299243927, + "num_tokens": 432405314.0, + "step": 16852 + }, + { + "epoch": 1.8507577421480343, + "grad_norm": 2.318195104598999, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7241540551185608, + "num_tokens": 432428136.0, + "step": 16853 + }, + { + "epoch": 1.8508675598506479, + "grad_norm": 2.3034095764160156, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7234119772911072, + "num_tokens": 432453392.0, + "step": 16854 + }, + { + "epoch": 1.8509773775532616, + "grad_norm": 2.3730623722076416, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7087376117706299, + "num_tokens": 432477937.0, + "step": 16855 + }, + { + "epoch": 1.8510871952558752, + "grad_norm": 2.1933400630950928, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7173954248428345, + "num_tokens": 432505873.0, + "step": 16856 + }, + { + "epoch": 1.851197012958489, + "grad_norm": 2.1292924880981445, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7268916368484497, + "num_tokens": 432531837.0, + "step": 16857 + }, + { + "epoch": 1.8513068306611027, + "grad_norm": 2.229863405227661, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6966536641120911, + "num_tokens": 432558450.0, + "step": 16858 + }, + { + "epoch": 1.8514166483637162, + "grad_norm": 2.514228105545044, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7184178233146667, + "num_tokens": 432580635.0, + "step": 16859 + }, + { + "epoch": 1.8515264660663298, + "grad_norm": 2.4090158939361572, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7362532019615173, + "num_tokens": 432603707.0, + "step": 16860 + }, + { + "epoch": 1.8516362837689435, + "grad_norm": 2.2508764266967773, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7393726110458374, + "num_tokens": 432628419.0, + "step": 16861 + }, + { + "epoch": 1.8517461014715573, + "grad_norm": 2.3352487087249756, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7279008626937866, + "num_tokens": 432652104.0, + "step": 16862 + }, + { + "epoch": 1.851855919174171, + "grad_norm": 2.110461711883545, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7175750136375427, + "num_tokens": 432678727.0, + "step": 16863 + }, + { + "epoch": 1.8519657368767846, + "grad_norm": 2.361473798751831, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7427409887313843, + "num_tokens": 432700113.0, + "step": 16864 + }, + { + "epoch": 1.852075554579398, + "grad_norm": 1.92267644405365, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7101094722747803, + "num_tokens": 432730841.0, + "step": 16865 + }, + { + "epoch": 1.8521853722820119, + "grad_norm": 2.250236988067627, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7084826231002808, + "num_tokens": 432756431.0, + "step": 16866 + }, + { + "epoch": 1.8522951899846256, + "grad_norm": 1.9963116645812988, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7155083417892456, + "num_tokens": 432788817.0, + "step": 16867 + }, + { + "epoch": 1.8524050076872391, + "grad_norm": 2.228377342224121, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7242639064788818, + "num_tokens": 432814584.0, + "step": 16868 + }, + { + "epoch": 1.8525148253898527, + "grad_norm": 2.1700940132141113, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7185781002044678, + "num_tokens": 432842094.0, + "step": 16869 + }, + { + "epoch": 1.8526246430924664, + "grad_norm": 1.930933952331543, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7397936582565308, + "num_tokens": 432872494.0, + "step": 16870 + }, + { + "epoch": 1.8527344607950802, + "grad_norm": 2.109816074371338, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.71769118309021, + "num_tokens": 432900035.0, + "step": 16871 + }, + { + "epoch": 1.852844278497694, + "grad_norm": 2.1296703815460205, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.721685528755188, + "num_tokens": 432929431.0, + "step": 16872 + }, + { + "epoch": 1.8529540962003075, + "grad_norm": 2.259793758392334, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7167238593101501, + "num_tokens": 432954977.0, + "step": 16873 + }, + { + "epoch": 1.853063913902921, + "grad_norm": 2.3549911975860596, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7195267081260681, + "num_tokens": 432979864.0, + "step": 16874 + }, + { + "epoch": 1.8531737316055348, + "grad_norm": 2.2745823860168457, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7174334526062012, + "num_tokens": 433006331.0, + "step": 16875 + }, + { + "epoch": 1.8532835493081485, + "grad_norm": 2.264968156814575, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7306825518608093, + "num_tokens": 433030589.0, + "step": 16876 + }, + { + "epoch": 1.8533933670107623, + "grad_norm": 2.380213499069214, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7053136825561523, + "num_tokens": 433055087.0, + "step": 16877 + }, + { + "epoch": 1.8535031847133758, + "grad_norm": 2.217602014541626, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7568076848983765, + "num_tokens": 433079916.0, + "step": 16878 + }, + { + "epoch": 1.8536130024159894, + "grad_norm": 2.00343656539917, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7284018993377686, + "num_tokens": 433111738.0, + "step": 16879 + }, + { + "epoch": 1.8537228201186031, + "grad_norm": 2.3572123050689697, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7358062267303467, + "num_tokens": 433135726.0, + "step": 16880 + }, + { + "epoch": 1.8538326378212169, + "grad_norm": 2.2544379234313965, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6981890201568604, + "num_tokens": 433160970.0, + "step": 16881 + }, + { + "epoch": 1.8539424555238304, + "grad_norm": 2.130384922027588, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7164140939712524, + "num_tokens": 433188416.0, + "step": 16882 + }, + { + "epoch": 1.854052273226444, + "grad_norm": 2.040658473968506, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7164937257766724, + "num_tokens": 433217491.0, + "step": 16883 + }, + { + "epoch": 1.8541620909290577, + "grad_norm": 2.440258502960205, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7391178607940674, + "num_tokens": 433238188.0, + "step": 16884 + }, + { + "epoch": 1.8542719086316715, + "grad_norm": 2.3761825561523438, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7406240105628967, + "num_tokens": 433258978.0, + "step": 16885 + }, + { + "epoch": 1.8543817263342852, + "grad_norm": 2.432715654373169, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7410804629325867, + "num_tokens": 433279956.0, + "step": 16886 + }, + { + "epoch": 1.8544915440368988, + "grad_norm": 2.300757646560669, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7157536149024963, + "num_tokens": 433303278.0, + "step": 16887 + }, + { + "epoch": 1.8546013617395123, + "grad_norm": 2.2103898525238037, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7095799446105957, + "num_tokens": 433331647.0, + "step": 16888 + }, + { + "epoch": 1.854711179442126, + "grad_norm": 2.1770622730255127, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7188155055046082, + "num_tokens": 433357398.0, + "step": 16889 + }, + { + "epoch": 1.8548209971447398, + "grad_norm": 2.5501794815063477, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7407537698745728, + "num_tokens": 433377642.0, + "step": 16890 + }, + { + "epoch": 1.8549308148473536, + "grad_norm": 2.17867374420166, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6960698366165161, + "num_tokens": 433407100.0, + "step": 16891 + }, + { + "epoch": 1.855040632549967, + "grad_norm": 2.474792957305908, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.735394299030304, + "num_tokens": 433426342.0, + "step": 16892 + }, + { + "epoch": 1.8551504502525806, + "grad_norm": 2.5166876316070557, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7117187976837158, + "num_tokens": 433448015.0, + "step": 16893 + }, + { + "epoch": 1.8552602679551944, + "grad_norm": 2.2124457359313965, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6915289163589478, + "num_tokens": 433476311.0, + "step": 16894 + }, + { + "epoch": 1.8553700856578081, + "grad_norm": 2.477362871170044, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.727512001991272, + "num_tokens": 433498215.0, + "step": 16895 + }, + { + "epoch": 1.8554799033604217, + "grad_norm": 2.284602403640747, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7272204160690308, + "num_tokens": 433522578.0, + "step": 16896 + }, + { + "epoch": 1.8555897210630352, + "grad_norm": 2.28493332862854, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7092584371566772, + "num_tokens": 433548752.0, + "step": 16897 + }, + { + "epoch": 1.855699538765649, + "grad_norm": 2.403977632522583, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7328680753707886, + "num_tokens": 433570937.0, + "step": 16898 + }, + { + "epoch": 1.8558093564682627, + "grad_norm": 2.171370506286621, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.692716121673584, + "num_tokens": 433601423.0, + "step": 16899 + }, + { + "epoch": 1.8559191741708765, + "grad_norm": 2.3647656440734863, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7055727243423462, + "num_tokens": 433624239.0, + "step": 16900 + }, + { + "epoch": 1.85602899187349, + "grad_norm": 2.1453137397766113, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7323243618011475, + "num_tokens": 433653010.0, + "step": 16901 + }, + { + "epoch": 1.8561388095761036, + "grad_norm": 2.2934632301330566, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7208455801010132, + "num_tokens": 433675666.0, + "step": 16902 + }, + { + "epoch": 1.8562486272787173, + "grad_norm": 2.453181505203247, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7019959688186646, + "num_tokens": 433699597.0, + "step": 16903 + }, + { + "epoch": 1.856358444981331, + "grad_norm": 2.0857954025268555, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7122709155082703, + "num_tokens": 433728118.0, + "step": 16904 + }, + { + "epoch": 1.8564682626839446, + "grad_norm": 2.0539629459381104, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7252272963523865, + "num_tokens": 433758667.0, + "step": 16905 + }, + { + "epoch": 1.8565780803865584, + "grad_norm": 2.213200330734253, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7101234197616577, + "num_tokens": 433784976.0, + "step": 16906 + }, + { + "epoch": 1.856687898089172, + "grad_norm": 2.3269848823547363, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7319974899291992, + "num_tokens": 433809306.0, + "step": 16907 + }, + { + "epoch": 1.8567977157917857, + "grad_norm": 2.206313133239746, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7284836173057556, + "num_tokens": 433834922.0, + "step": 16908 + }, + { + "epoch": 1.8569075334943994, + "grad_norm": 2.643700361251831, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7020823359489441, + "num_tokens": 433857306.0, + "step": 16909 + }, + { + "epoch": 1.857017351197013, + "grad_norm": 1.9942989349365234, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7161044478416443, + "num_tokens": 433889297.0, + "step": 16910 + }, + { + "epoch": 1.8571271688996265, + "grad_norm": 2.161735773086548, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7014384269714355, + "num_tokens": 433917480.0, + "step": 16911 + }, + { + "epoch": 1.8572369866022402, + "grad_norm": 2.349365234375, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7366567850112915, + "num_tokens": 433939205.0, + "step": 16912 + }, + { + "epoch": 1.857346804304854, + "grad_norm": 2.209045648574829, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7075725793838501, + "num_tokens": 433966239.0, + "step": 16913 + }, + { + "epoch": 1.8574566220074678, + "grad_norm": 2.3778810501098633, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7186644673347473, + "num_tokens": 433988790.0, + "step": 16914 + }, + { + "epoch": 1.8575664397100813, + "grad_norm": 2.210728645324707, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7438650131225586, + "num_tokens": 434013669.0, + "step": 16915 + }, + { + "epoch": 1.8576762574126948, + "grad_norm": 2.4501242637634277, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7191318273544312, + "num_tokens": 434036808.0, + "step": 16916 + }, + { + "epoch": 1.8577860751153086, + "grad_norm": 2.1152617931365967, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6914511322975159, + "num_tokens": 434065783.0, + "step": 16917 + }, + { + "epoch": 1.8578958928179223, + "grad_norm": 2.059126615524292, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7056591510772705, + "num_tokens": 434097182.0, + "step": 16918 + }, + { + "epoch": 1.8580057105205359, + "grad_norm": 2.3154397010803223, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7032089829444885, + "num_tokens": 434120364.0, + "step": 16919 + }, + { + "epoch": 1.8581155282231496, + "grad_norm": 2.604987859725952, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7356355786323547, + "num_tokens": 434140433.0, + "step": 16920 + }, + { + "epoch": 1.8582253459257632, + "grad_norm": 2.084963321685791, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6911844611167908, + "num_tokens": 434170360.0, + "step": 16921 + }, + { + "epoch": 1.858335163628377, + "grad_norm": 2.168654680252075, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7037592530250549, + "num_tokens": 434199113.0, + "step": 16922 + }, + { + "epoch": 1.8584449813309907, + "grad_norm": 2.528834819793701, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7352327704429626, + "num_tokens": 434220019.0, + "step": 16923 + }, + { + "epoch": 1.8585547990336042, + "grad_norm": 2.2804319858551025, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.724787712097168, + "num_tokens": 434244378.0, + "step": 16924 + }, + { + "epoch": 1.8586646167362177, + "grad_norm": 2.0284016132354736, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7319801449775696, + "num_tokens": 434272606.0, + "step": 16925 + }, + { + "epoch": 1.8587744344388315, + "grad_norm": 1.9930609464645386, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7248809337615967, + "num_tokens": 434302495.0, + "step": 16926 + }, + { + "epoch": 1.8588842521414453, + "grad_norm": 2.2049427032470703, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7115899324417114, + "num_tokens": 434328319.0, + "step": 16927 + }, + { + "epoch": 1.858994069844059, + "grad_norm": 2.458772659301758, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7318251729011536, + "num_tokens": 434351213.0, + "step": 16928 + }, + { + "epoch": 1.8591038875466726, + "grad_norm": 2.1133735179901123, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.686760425567627, + "num_tokens": 434381470.0, + "step": 16929 + }, + { + "epoch": 1.859213705249286, + "grad_norm": 2.324737548828125, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.704484224319458, + "num_tokens": 434406181.0, + "step": 16930 + }, + { + "epoch": 1.8593235229518998, + "grad_norm": 2.7404019832611084, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7469928860664368, + "num_tokens": 434425663.0, + "step": 16931 + }, + { + "epoch": 1.8594333406545136, + "grad_norm": 2.5743417739868164, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7253476977348328, + "num_tokens": 434447139.0, + "step": 16932 + }, + { + "epoch": 1.8595431583571271, + "grad_norm": 2.2772862911224365, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7189500331878662, + "num_tokens": 434473195.0, + "step": 16933 + }, + { + "epoch": 1.8596529760597407, + "grad_norm": 2.1665048599243164, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7125070095062256, + "num_tokens": 434500702.0, + "step": 16934 + }, + { + "epoch": 1.8597627937623544, + "grad_norm": 2.3289382457733154, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.739016592502594, + "num_tokens": 434524200.0, + "step": 16935 + }, + { + "epoch": 1.8598726114649682, + "grad_norm": 2.198723554611206, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.6975280046463013, + "num_tokens": 434550646.0, + "step": 16936 + }, + { + "epoch": 1.859982429167582, + "grad_norm": 2.4192380905151367, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7203896045684814, + "num_tokens": 434573399.0, + "step": 16937 + }, + { + "epoch": 1.8600922468701955, + "grad_norm": 2.489049196243286, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7216649055480957, + "num_tokens": 434595284.0, + "step": 16938 + }, + { + "epoch": 1.860202064572809, + "grad_norm": 2.5496037006378174, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7509981393814087, + "num_tokens": 434614805.0, + "step": 16939 + }, + { + "epoch": 1.8603118822754228, + "grad_norm": 2.0670688152313232, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7178385257720947, + "num_tokens": 434643230.0, + "step": 16940 + }, + { + "epoch": 1.8604216999780365, + "grad_norm": 2.181352138519287, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.717072606086731, + "num_tokens": 434668648.0, + "step": 16941 + }, + { + "epoch": 1.8605315176806503, + "grad_norm": 2.23140287399292, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7325804233551025, + "num_tokens": 434693087.0, + "step": 16942 + }, + { + "epoch": 1.8606413353832638, + "grad_norm": 1.9578548669815063, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7240961790084839, + "num_tokens": 434724730.0, + "step": 16943 + }, + { + "epoch": 1.8607511530858774, + "grad_norm": 2.3373143672943115, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7315738201141357, + "num_tokens": 434748160.0, + "step": 16944 + }, + { + "epoch": 1.860860970788491, + "grad_norm": 2.2194507122039795, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7190563082695007, + "num_tokens": 434775107.0, + "step": 16945 + }, + { + "epoch": 1.8609707884911049, + "grad_norm": 2.259127378463745, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7177869081497192, + "num_tokens": 434800685.0, + "step": 16946 + }, + { + "epoch": 1.8610806061937184, + "grad_norm": 2.216900110244751, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7019019722938538, + "num_tokens": 434828457.0, + "step": 16947 + }, + { + "epoch": 1.861190423896332, + "grad_norm": 2.2614781856536865, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7248349189758301, + "num_tokens": 434853473.0, + "step": 16948 + }, + { + "epoch": 1.8613002415989457, + "grad_norm": 1.8576048612594604, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7199400067329407, + "num_tokens": 434888353.0, + "step": 16949 + }, + { + "epoch": 1.8614100593015594, + "grad_norm": 1.924302101135254, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6921908855438232, + "num_tokens": 434922473.0, + "step": 16950 + }, + { + "epoch": 1.8615198770041732, + "grad_norm": 2.244593620300293, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7210097312927246, + "num_tokens": 434947318.0, + "step": 16951 + }, + { + "epoch": 1.8616296947067867, + "grad_norm": 2.4181859493255615, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7250826358795166, + "num_tokens": 434968983.0, + "step": 16952 + }, + { + "epoch": 1.8617395124094003, + "grad_norm": 2.323284864425659, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7021607756614685, + "num_tokens": 434992902.0, + "step": 16953 + }, + { + "epoch": 1.861849330112014, + "grad_norm": 2.19490385055542, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7130410671234131, + "num_tokens": 435018899.0, + "step": 16954 + }, + { + "epoch": 1.8619591478146278, + "grad_norm": 2.3221466541290283, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7333620190620422, + "num_tokens": 435041860.0, + "step": 16955 + }, + { + "epoch": 1.8620689655172413, + "grad_norm": 2.0936713218688965, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.6955937743186951, + "num_tokens": 435070342.0, + "step": 16956 + }, + { + "epoch": 1.862178783219855, + "grad_norm": 2.048549175262451, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7020343542098999, + "num_tokens": 435098216.0, + "step": 16957 + }, + { + "epoch": 1.8622886009224686, + "grad_norm": 2.220224618911743, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6966904997825623, + "num_tokens": 435127630.0, + "step": 16958 + }, + { + "epoch": 1.8623984186250824, + "grad_norm": 2.120936155319214, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7099462747573853, + "num_tokens": 435157201.0, + "step": 16959 + }, + { + "epoch": 1.8625082363276961, + "grad_norm": 1.8974181413650513, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7123950719833374, + "num_tokens": 435189938.0, + "step": 16960 + }, + { + "epoch": 1.8626180540303097, + "grad_norm": 2.1748931407928467, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7216448187828064, + "num_tokens": 435214696.0, + "step": 16961 + }, + { + "epoch": 1.8627278717329232, + "grad_norm": 2.46818208694458, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7161315679550171, + "num_tokens": 435236698.0, + "step": 16962 + }, + { + "epoch": 1.862837689435537, + "grad_norm": 2.2009007930755615, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.692358136177063, + "num_tokens": 435263775.0, + "step": 16963 + }, + { + "epoch": 1.8629475071381507, + "grad_norm": 2.4884462356567383, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7065236568450928, + "num_tokens": 435286820.0, + "step": 16964 + }, + { + "epoch": 1.8630573248407645, + "grad_norm": 2.062115430831909, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7080928087234497, + "num_tokens": 435317622.0, + "step": 16965 + }, + { + "epoch": 1.863167142543378, + "grad_norm": 2.1250813007354736, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7341002821922302, + "num_tokens": 435344862.0, + "step": 16966 + }, + { + "epoch": 1.8632769602459915, + "grad_norm": 2.706772565841675, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.727290153503418, + "num_tokens": 435363994.0, + "step": 16967 + }, + { + "epoch": 1.8633867779486053, + "grad_norm": 2.2402076721191406, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7082125544548035, + "num_tokens": 435391717.0, + "step": 16968 + }, + { + "epoch": 1.863496595651219, + "grad_norm": 1.9895070791244507, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.695159912109375, + "num_tokens": 435422497.0, + "step": 16969 + }, + { + "epoch": 1.8636064133538326, + "grad_norm": 2.078756809234619, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6953640580177307, + "num_tokens": 435452817.0, + "step": 16970 + }, + { + "epoch": 1.8637162310564463, + "grad_norm": 2.325714349746704, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7156313061714172, + "num_tokens": 435476704.0, + "step": 16971 + }, + { + "epoch": 1.8638260487590599, + "grad_norm": 2.247964859008789, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7284072041511536, + "num_tokens": 435501371.0, + "step": 16972 + }, + { + "epoch": 1.8639358664616736, + "grad_norm": 2.2986440658569336, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7304264307022095, + "num_tokens": 435524342.0, + "step": 16973 + }, + { + "epoch": 1.8640456841642874, + "grad_norm": 2.0898244380950928, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7087570428848267, + "num_tokens": 435553879.0, + "step": 16974 + }, + { + "epoch": 1.864155501866901, + "grad_norm": 2.072545051574707, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7188761234283447, + "num_tokens": 435582793.0, + "step": 16975 + }, + { + "epoch": 1.8642653195695145, + "grad_norm": 2.042274236679077, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7202707529067993, + "num_tokens": 435610510.0, + "step": 16976 + }, + { + "epoch": 1.8643751372721282, + "grad_norm": 2.2815513610839844, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7224744558334351, + "num_tokens": 435633727.0, + "step": 16977 + }, + { + "epoch": 1.864484954974742, + "grad_norm": 2.124739646911621, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7041728496551514, + "num_tokens": 435663203.0, + "step": 16978 + }, + { + "epoch": 1.8645947726773557, + "grad_norm": 2.721053123474121, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7434374094009399, + "num_tokens": 435681109.0, + "step": 16979 + }, + { + "epoch": 1.8647045903799693, + "grad_norm": 2.191716194152832, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.704750657081604, + "num_tokens": 435707769.0, + "step": 16980 + }, + { + "epoch": 1.8648144080825828, + "grad_norm": 2.0779457092285156, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7266649007797241, + "num_tokens": 435734409.0, + "step": 16981 + }, + { + "epoch": 1.8649242257851966, + "grad_norm": 2.3375191688537598, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7075544595718384, + "num_tokens": 435759364.0, + "step": 16982 + }, + { + "epoch": 1.8650340434878103, + "grad_norm": 2.4643805027008057, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7086405754089355, + "num_tokens": 435781768.0, + "step": 16983 + }, + { + "epoch": 1.8651438611904239, + "grad_norm": 2.25441837310791, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7276421785354614, + "num_tokens": 435806501.0, + "step": 16984 + }, + { + "epoch": 1.8652536788930374, + "grad_norm": 2.2546873092651367, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7239023447036743, + "num_tokens": 435830515.0, + "step": 16985 + }, + { + "epoch": 1.8653634965956511, + "grad_norm": 2.1886255741119385, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6992896795272827, + "num_tokens": 435859911.0, + "step": 16986 + }, + { + "epoch": 1.865473314298265, + "grad_norm": 2.14520001411438, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7184156775474548, + "num_tokens": 435888366.0, + "step": 16987 + }, + { + "epoch": 1.8655831320008787, + "grad_norm": 2.332825183868408, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7088881134986877, + "num_tokens": 435914200.0, + "step": 16988 + }, + { + "epoch": 1.8656929497034922, + "grad_norm": 2.3333840370178223, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7198780179023743, + "num_tokens": 435938543.0, + "step": 16989 + }, + { + "epoch": 1.8658027674061057, + "grad_norm": 2.499408006668091, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.721268892288208, + "num_tokens": 435960212.0, + "step": 16990 + }, + { + "epoch": 1.8659125851087195, + "grad_norm": 2.2989280223846436, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7170737981796265, + "num_tokens": 435986933.0, + "step": 16991 + }, + { + "epoch": 1.8660224028113332, + "grad_norm": 2.214064836502075, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7012475728988647, + "num_tokens": 436015370.0, + "step": 16992 + }, + { + "epoch": 1.866132220513947, + "grad_norm": 2.5463716983795166, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.735803484916687, + "num_tokens": 436036722.0, + "step": 16993 + }, + { + "epoch": 1.8662420382165605, + "grad_norm": 2.1610803604125977, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7284802794456482, + "num_tokens": 436063499.0, + "step": 16994 + }, + { + "epoch": 1.866351855919174, + "grad_norm": 2.5488297939300537, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7439746856689453, + "num_tokens": 436083287.0, + "step": 16995 + }, + { + "epoch": 1.8664616736217878, + "grad_norm": 2.376033306121826, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.730924665927887, + "num_tokens": 436106995.0, + "step": 16996 + }, + { + "epoch": 1.8665714913244016, + "grad_norm": 2.4124011993408203, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7304821014404297, + "num_tokens": 436129329.0, + "step": 16997 + }, + { + "epoch": 1.8666813090270151, + "grad_norm": 2.1935911178588867, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7186737060546875, + "num_tokens": 436156548.0, + "step": 16998 + }, + { + "epoch": 1.8667911267296287, + "grad_norm": 2.617997407913208, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7384943962097168, + "num_tokens": 436176534.0, + "step": 16999 + }, + { + "epoch": 1.8669009444322424, + "grad_norm": 2.0890917778015137, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7518178820610046, + "num_tokens": 436204647.0, + "step": 17000 + }, + { + "epoch": 1.8670107621348562, + "grad_norm": 2.0717031955718994, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7139682769775391, + "num_tokens": 436232422.0, + "step": 17001 + }, + { + "epoch": 1.86712057983747, + "grad_norm": 2.3634283542633057, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7498492002487183, + "num_tokens": 436253217.0, + "step": 17002 + }, + { + "epoch": 1.8672303975400835, + "grad_norm": 2.208828926086426, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7393806576728821, + "num_tokens": 436277942.0, + "step": 17003 + }, + { + "epoch": 1.867340215242697, + "grad_norm": 2.1920511722564697, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7160065174102783, + "num_tokens": 436303078.0, + "step": 17004 + }, + { + "epoch": 1.8674500329453108, + "grad_norm": 2.1873669624328613, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7381032705307007, + "num_tokens": 436327513.0, + "step": 17005 + }, + { + "epoch": 1.8675598506479245, + "grad_norm": 2.223125457763672, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7156972885131836, + "num_tokens": 436351370.0, + "step": 17006 + }, + { + "epoch": 1.8676696683505383, + "grad_norm": 2.3795366287231445, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7203255891799927, + "num_tokens": 436375731.0, + "step": 17007 + }, + { + "epoch": 1.8677794860531518, + "grad_norm": 2.3851850032806396, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7227497100830078, + "num_tokens": 436398535.0, + "step": 17008 + }, + { + "epoch": 1.8678893037557653, + "grad_norm": 2.583068609237671, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7174874544143677, + "num_tokens": 436418795.0, + "step": 17009 + }, + { + "epoch": 1.867999121458379, + "grad_norm": 2.3661673069000244, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7139264941215515, + "num_tokens": 436442854.0, + "step": 17010 + }, + { + "epoch": 1.8681089391609929, + "grad_norm": 2.315354585647583, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7306256294250488, + "num_tokens": 436466756.0, + "step": 17011 + }, + { + "epoch": 1.8682187568636064, + "grad_norm": 2.210623025894165, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7274765968322754, + "num_tokens": 436493024.0, + "step": 17012 + }, + { + "epoch": 1.86832857456622, + "grad_norm": 2.1397032737731934, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7198206186294556, + "num_tokens": 436520728.0, + "step": 17013 + }, + { + "epoch": 1.8684383922688337, + "grad_norm": 2.3500444889068604, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7171300649642944, + "num_tokens": 436545475.0, + "step": 17014 + }, + { + "epoch": 1.8685482099714474, + "grad_norm": 2.393051862716675, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7100070714950562, + "num_tokens": 436568258.0, + "step": 17015 + }, + { + "epoch": 1.8686580276740612, + "grad_norm": 2.2392454147338867, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.706878662109375, + "num_tokens": 436594483.0, + "step": 17016 + }, + { + "epoch": 1.8687678453766747, + "grad_norm": 2.0087387561798096, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7132622003555298, + "num_tokens": 436624867.0, + "step": 17017 + }, + { + "epoch": 1.8688776630792883, + "grad_norm": 2.1055407524108887, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7214095592498779, + "num_tokens": 436653146.0, + "step": 17018 + }, + { + "epoch": 1.868987480781902, + "grad_norm": 1.9488409757614136, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7043120265007019, + "num_tokens": 436686595.0, + "step": 17019 + }, + { + "epoch": 1.8690972984845158, + "grad_norm": 2.3689849376678467, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7523950338363647, + "num_tokens": 436709022.0, + "step": 17020 + }, + { + "epoch": 1.8692071161871293, + "grad_norm": 2.1632003784179688, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7228976488113403, + "num_tokens": 436734500.0, + "step": 17021 + }, + { + "epoch": 1.869316933889743, + "grad_norm": 2.09248423576355, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7263001203536987, + "num_tokens": 436760178.0, + "step": 17022 + }, + { + "epoch": 1.8694267515923566, + "grad_norm": 2.150690793991089, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7295699715614319, + "num_tokens": 436786380.0, + "step": 17023 + }, + { + "epoch": 1.8695365692949704, + "grad_norm": 2.2089266777038574, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7168529629707336, + "num_tokens": 436812824.0, + "step": 17024 + }, + { + "epoch": 1.8696463869975841, + "grad_norm": 2.4203948974609375, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7399870157241821, + "num_tokens": 436835903.0, + "step": 17025 + }, + { + "epoch": 1.8697562047001977, + "grad_norm": 2.29060959815979, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7366877198219299, + "num_tokens": 436858864.0, + "step": 17026 + }, + { + "epoch": 1.8698660224028112, + "grad_norm": 2.3020763397216797, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6996683478355408, + "num_tokens": 436884114.0, + "step": 17027 + }, + { + "epoch": 1.869975840105425, + "grad_norm": 2.3879635334014893, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7270793914794922, + "num_tokens": 436906309.0, + "step": 17028 + }, + { + "epoch": 1.8700856578080387, + "grad_norm": 2.2761600017547607, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6993027925491333, + "num_tokens": 436932017.0, + "step": 17029 + }, + { + "epoch": 1.8701954755106525, + "grad_norm": 2.2708518505096436, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7112441658973694, + "num_tokens": 436958524.0, + "step": 17030 + }, + { + "epoch": 1.870305293213266, + "grad_norm": 2.9738306999206543, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7500410676002502, + "num_tokens": 436973632.0, + "step": 17031 + }, + { + "epoch": 1.8704151109158795, + "grad_norm": 2.1935019493103027, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7268486618995667, + "num_tokens": 436998300.0, + "step": 17032 + }, + { + "epoch": 1.8705249286184933, + "grad_norm": 2.4522922039031982, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7307149171829224, + "num_tokens": 437020588.0, + "step": 17033 + }, + { + "epoch": 1.870634746321107, + "grad_norm": 2.0826876163482666, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7217150926589966, + "num_tokens": 437049725.0, + "step": 17034 + }, + { + "epoch": 1.8707445640237206, + "grad_norm": 2.244317054748535, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7186229825019836, + "num_tokens": 437075298.0, + "step": 17035 + }, + { + "epoch": 1.8708543817263343, + "grad_norm": 2.1691315174102783, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7062633633613586, + "num_tokens": 437103104.0, + "step": 17036 + }, + { + "epoch": 1.8709641994289479, + "grad_norm": 2.1609914302825928, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6951442360877991, + "num_tokens": 437130160.0, + "step": 17037 + }, + { + "epoch": 1.8710740171315616, + "grad_norm": 2.437894344329834, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7408641576766968, + "num_tokens": 437152741.0, + "step": 17038 + }, + { + "epoch": 1.8711838348341754, + "grad_norm": 2.146044969558716, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7252829074859619, + "num_tokens": 437178184.0, + "step": 17039 + }, + { + "epoch": 1.871293652536789, + "grad_norm": 2.232748031616211, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7011116743087769, + "num_tokens": 437204323.0, + "step": 17040 + }, + { + "epoch": 1.8714034702394025, + "grad_norm": 2.4389374256134033, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7330209016799927, + "num_tokens": 437226598.0, + "step": 17041 + }, + { + "epoch": 1.8715132879420162, + "grad_norm": 2.2487642765045166, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7353758811950684, + "num_tokens": 437251512.0, + "step": 17042 + }, + { + "epoch": 1.87162310564463, + "grad_norm": 2.1407084465026855, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7408643960952759, + "num_tokens": 437277103.0, + "step": 17043 + }, + { + "epoch": 1.8717329233472437, + "grad_norm": 2.136469602584839, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7179322838783264, + "num_tokens": 437303029.0, + "step": 17044 + }, + { + "epoch": 1.8718427410498573, + "grad_norm": 2.3446455001831055, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.755689799785614, + "num_tokens": 437326229.0, + "step": 17045 + }, + { + "epoch": 1.8719525587524708, + "grad_norm": 2.053936004638672, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7038478851318359, + "num_tokens": 437360518.0, + "step": 17046 + }, + { + "epoch": 1.8720623764550846, + "grad_norm": 2.3607337474823, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7181804180145264, + "num_tokens": 437381900.0, + "step": 17047 + }, + { + "epoch": 1.8721721941576983, + "grad_norm": 2.173154592514038, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7263182401657104, + "num_tokens": 437407350.0, + "step": 17048 + }, + { + "epoch": 1.8722820118603118, + "grad_norm": 2.1805994510650635, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7041054964065552, + "num_tokens": 437435024.0, + "step": 17049 + }, + { + "epoch": 1.8723918295629254, + "grad_norm": 2.2887322902679443, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7232499718666077, + "num_tokens": 437461132.0, + "step": 17050 + }, + { + "epoch": 1.8725016472655391, + "grad_norm": 2.3016340732574463, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7197848558425903, + "num_tokens": 437486203.0, + "step": 17051 + }, + { + "epoch": 1.872611464968153, + "grad_norm": 2.1572885513305664, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7021955251693726, + "num_tokens": 437514241.0, + "step": 17052 + }, + { + "epoch": 1.8727212826707667, + "grad_norm": 2.10667085647583, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7181925773620605, + "num_tokens": 437542243.0, + "step": 17053 + }, + { + "epoch": 1.8728311003733802, + "grad_norm": 2.160233736038208, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7043383121490479, + "num_tokens": 437569913.0, + "step": 17054 + }, + { + "epoch": 1.8729409180759937, + "grad_norm": 2.1442840099334717, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7099942564964294, + "num_tokens": 437598229.0, + "step": 17055 + }, + { + "epoch": 1.8730507357786075, + "grad_norm": 2.2795228958129883, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7183359861373901, + "num_tokens": 437622119.0, + "step": 17056 + }, + { + "epoch": 1.8731605534812212, + "grad_norm": 2.144937515258789, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7225145101547241, + "num_tokens": 437650462.0, + "step": 17057 + }, + { + "epoch": 1.873270371183835, + "grad_norm": 2.7711024284362793, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7480319142341614, + "num_tokens": 437667750.0, + "step": 17058 + }, + { + "epoch": 1.8733801888864485, + "grad_norm": 2.666349172592163, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7328715324401855, + "num_tokens": 437687001.0, + "step": 17059 + }, + { + "epoch": 1.873490006589062, + "grad_norm": 2.0123579502105713, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7100098133087158, + "num_tokens": 437719049.0, + "step": 17060 + }, + { + "epoch": 1.8735998242916758, + "grad_norm": 1.990000605583191, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6878342628479004, + "num_tokens": 437753969.0, + "step": 17061 + }, + { + "epoch": 1.8737096419942896, + "grad_norm": 2.3707284927368164, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.735387921333313, + "num_tokens": 437775884.0, + "step": 17062 + }, + { + "epoch": 1.8738194596969031, + "grad_norm": 2.3304498195648193, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7147315740585327, + "num_tokens": 437800663.0, + "step": 17063 + }, + { + "epoch": 1.8739292773995166, + "grad_norm": 2.1062185764312744, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7066053152084351, + "num_tokens": 437830286.0, + "step": 17064 + }, + { + "epoch": 1.8740390951021304, + "grad_norm": 2.2074036598205566, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.745132327079773, + "num_tokens": 437854984.0, + "step": 17065 + }, + { + "epoch": 1.8741489128047442, + "grad_norm": 2.5256216526031494, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7351280450820923, + "num_tokens": 437875305.0, + "step": 17066 + }, + { + "epoch": 1.874258730507358, + "grad_norm": 2.1691908836364746, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.6963234543800354, + "num_tokens": 437901905.0, + "step": 17067 + }, + { + "epoch": 1.8743685482099715, + "grad_norm": 2.1023855209350586, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7276365160942078, + "num_tokens": 437931204.0, + "step": 17068 + }, + { + "epoch": 1.874478365912585, + "grad_norm": 2.151169538497925, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.729827344417572, + "num_tokens": 437957961.0, + "step": 17069 + }, + { + "epoch": 1.8745881836151987, + "grad_norm": 2.1434054374694824, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7301609516143799, + "num_tokens": 437984674.0, + "step": 17070 + }, + { + "epoch": 1.8746980013178125, + "grad_norm": 2.3023993968963623, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7212029099464417, + "num_tokens": 438010888.0, + "step": 17071 + }, + { + "epoch": 1.8748078190204263, + "grad_norm": 2.0540735721588135, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7078123688697815, + "num_tokens": 438042000.0, + "step": 17072 + }, + { + "epoch": 1.8749176367230398, + "grad_norm": 2.235227346420288, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7211883068084717, + "num_tokens": 438067000.0, + "step": 17073 + }, + { + "epoch": 1.8750274544256533, + "grad_norm": 2.672506093978882, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7116197347640991, + "num_tokens": 438086813.0, + "step": 17074 + }, + { + "epoch": 1.875137272128267, + "grad_norm": 2.3531274795532227, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7268831729888916, + "num_tokens": 438109868.0, + "step": 17075 + }, + { + "epoch": 1.8752470898308808, + "grad_norm": 2.256836175918579, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7152635455131531, + "num_tokens": 438134919.0, + "step": 17076 + }, + { + "epoch": 1.8753569075334944, + "grad_norm": 2.320945978164673, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7195173501968384, + "num_tokens": 438159874.0, + "step": 17077 + }, + { + "epoch": 1.875466725236108, + "grad_norm": 2.2869338989257812, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7148882150650024, + "num_tokens": 438184607.0, + "step": 17078 + }, + { + "epoch": 1.8755765429387217, + "grad_norm": 2.0193145275115967, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7287877798080444, + "num_tokens": 438216052.0, + "step": 17079 + }, + { + "epoch": 1.8756863606413354, + "grad_norm": 2.402575969696045, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7414738535881042, + "num_tokens": 438240686.0, + "step": 17080 + }, + { + "epoch": 1.8757961783439492, + "grad_norm": 2.4067180156707764, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7212917804718018, + "num_tokens": 438266759.0, + "step": 17081 + }, + { + "epoch": 1.8759059960465627, + "grad_norm": 2.195803165435791, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.72893226146698, + "num_tokens": 438294371.0, + "step": 17082 + }, + { + "epoch": 1.8760158137491763, + "grad_norm": 2.5865368843078613, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.724515974521637, + "num_tokens": 438315083.0, + "step": 17083 + }, + { + "epoch": 1.87612563145179, + "grad_norm": 2.182842493057251, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7258376479148865, + "num_tokens": 438340311.0, + "step": 17084 + }, + { + "epoch": 1.8762354491544038, + "grad_norm": 2.1564741134643555, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7298054695129395, + "num_tokens": 438367629.0, + "step": 17085 + }, + { + "epoch": 1.8763452668570173, + "grad_norm": 2.0274453163146973, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7111552953720093, + "num_tokens": 438399514.0, + "step": 17086 + }, + { + "epoch": 1.876455084559631, + "grad_norm": 2.3138465881347656, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7288237810134888, + "num_tokens": 438423438.0, + "step": 17087 + }, + { + "epoch": 1.8765649022622446, + "grad_norm": 2.4590275287628174, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7260081768035889, + "num_tokens": 438444140.0, + "step": 17088 + }, + { + "epoch": 1.8766747199648584, + "grad_norm": 2.158600330352783, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.716005802154541, + "num_tokens": 438471530.0, + "step": 17089 + }, + { + "epoch": 1.876784537667472, + "grad_norm": 2.1977312564849854, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7170011401176453, + "num_tokens": 438496177.0, + "step": 17090 + }, + { + "epoch": 1.8768943553700856, + "grad_norm": 2.27288818359375, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7421388626098633, + "num_tokens": 438519078.0, + "step": 17091 + }, + { + "epoch": 1.8770041730726992, + "grad_norm": 2.1391613483428955, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7178347706794739, + "num_tokens": 438546232.0, + "step": 17092 + }, + { + "epoch": 1.877113990775313, + "grad_norm": 2.1309621334075928, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.690361738204956, + "num_tokens": 438573311.0, + "step": 17093 + }, + { + "epoch": 1.8772238084779267, + "grad_norm": 2.6775121688842773, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7186744213104248, + "num_tokens": 438594501.0, + "step": 17094 + }, + { + "epoch": 1.8773336261805404, + "grad_norm": 2.0976085662841797, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7419884204864502, + "num_tokens": 438620964.0, + "step": 17095 + }, + { + "epoch": 1.877443443883154, + "grad_norm": 2.2050957679748535, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7221860885620117, + "num_tokens": 438646533.0, + "step": 17096 + }, + { + "epoch": 1.8775532615857675, + "grad_norm": 2.094080686569214, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7289893627166748, + "num_tokens": 438673219.0, + "step": 17097 + }, + { + "epoch": 1.8776630792883813, + "grad_norm": 2.3794898986816406, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7354681491851807, + "num_tokens": 438694692.0, + "step": 17098 + }, + { + "epoch": 1.877772896990995, + "grad_norm": 2.2133734226226807, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7308408617973328, + "num_tokens": 438719171.0, + "step": 17099 + }, + { + "epoch": 1.8778827146936086, + "grad_norm": 1.9096245765686035, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7082738876342773, + "num_tokens": 438753694.0, + "step": 17100 + }, + { + "epoch": 1.8779925323962223, + "grad_norm": 2.367062568664551, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.708636999130249, + "num_tokens": 438778527.0, + "step": 17101 + }, + { + "epoch": 1.8781023500988359, + "grad_norm": 2.2478525638580322, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7156388163566589, + "num_tokens": 438803685.0, + "step": 17102 + }, + { + "epoch": 1.8782121678014496, + "grad_norm": 2.125253438949585, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6962478160858154, + "num_tokens": 438831540.0, + "step": 17103 + }, + { + "epoch": 1.8783219855040634, + "grad_norm": 2.5845446586608887, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7450290322303772, + "num_tokens": 438851107.0, + "step": 17104 + }, + { + "epoch": 1.878431803206677, + "grad_norm": 2.1410677433013916, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7062714099884033, + "num_tokens": 438879535.0, + "step": 17105 + }, + { + "epoch": 1.8785416209092904, + "grad_norm": 1.9207730293273926, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7224422693252563, + "num_tokens": 438913836.0, + "step": 17106 + }, + { + "epoch": 1.8786514386119042, + "grad_norm": 2.4262044429779053, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7230173349380493, + "num_tokens": 438938349.0, + "step": 17107 + }, + { + "epoch": 1.878761256314518, + "grad_norm": 2.551164150238037, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7271770238876343, + "num_tokens": 438957749.0, + "step": 17108 + }, + { + "epoch": 1.8788710740171317, + "grad_norm": 2.061586618423462, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7218959927558899, + "num_tokens": 438985428.0, + "step": 17109 + }, + { + "epoch": 1.8789808917197452, + "grad_norm": 1.8947055339813232, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7124865055084229, + "num_tokens": 439019074.0, + "step": 17110 + }, + { + "epoch": 1.8790907094223588, + "grad_norm": 2.0168328285217285, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7058119773864746, + "num_tokens": 439048040.0, + "step": 17111 + }, + { + "epoch": 1.8792005271249725, + "grad_norm": 2.354964017868042, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7267301082611084, + "num_tokens": 439071938.0, + "step": 17112 + }, + { + "epoch": 1.8793103448275863, + "grad_norm": 2.2163472175598145, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7281079888343811, + "num_tokens": 439097062.0, + "step": 17113 + }, + { + "epoch": 1.8794201625301998, + "grad_norm": 2.2903554439544678, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7004013061523438, + "num_tokens": 439123494.0, + "step": 17114 + }, + { + "epoch": 1.8795299802328134, + "grad_norm": 2.49607253074646, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7274554967880249, + "num_tokens": 439143868.0, + "step": 17115 + }, + { + "epoch": 1.8796397979354271, + "grad_norm": 1.984092354774475, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7164584398269653, + "num_tokens": 439174977.0, + "step": 17116 + }, + { + "epoch": 1.8797496156380409, + "grad_norm": 2.457460880279541, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.704151451587677, + "num_tokens": 439196680.0, + "step": 17117 + }, + { + "epoch": 1.8798594333406546, + "grad_norm": 2.377772331237793, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7505598068237305, + "num_tokens": 439217990.0, + "step": 17118 + }, + { + "epoch": 1.8799692510432682, + "grad_norm": 2.242553949356079, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.704755425453186, + "num_tokens": 439244729.0, + "step": 17119 + }, + { + "epoch": 1.8800790687458817, + "grad_norm": 2.388262987136841, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7239021062850952, + "num_tokens": 439267361.0, + "step": 17120 + }, + { + "epoch": 1.8801888864484955, + "grad_norm": 2.605905532836914, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7446308135986328, + "num_tokens": 439285885.0, + "step": 17121 + }, + { + "epoch": 1.8802987041511092, + "grad_norm": 1.9946227073669434, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7275563478469849, + "num_tokens": 439315322.0, + "step": 17122 + }, + { + "epoch": 1.880408521853723, + "grad_norm": 2.1780385971069336, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7113109827041626, + "num_tokens": 439342055.0, + "step": 17123 + }, + { + "epoch": 1.8805183395563365, + "grad_norm": 2.207106351852417, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7143083810806274, + "num_tokens": 439367750.0, + "step": 17124 + }, + { + "epoch": 1.88062815725895, + "grad_norm": 2.0405726432800293, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7073126435279846, + "num_tokens": 439397044.0, + "step": 17125 + }, + { + "epoch": 1.8807379749615638, + "grad_norm": 1.968654990196228, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.716547429561615, + "num_tokens": 439428080.0, + "step": 17126 + }, + { + "epoch": 1.8808477926641776, + "grad_norm": 2.3809642791748047, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7235770225524902, + "num_tokens": 439451123.0, + "step": 17127 + }, + { + "epoch": 1.880957610366791, + "grad_norm": 2.0580668449401855, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7182328104972839, + "num_tokens": 439483082.0, + "step": 17128 + }, + { + "epoch": 1.8810674280694046, + "grad_norm": 2.316333293914795, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7283892631530762, + "num_tokens": 439507490.0, + "step": 17129 + }, + { + "epoch": 1.8811772457720184, + "grad_norm": 2.3462562561035156, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.740585207939148, + "num_tokens": 439528472.0, + "step": 17130 + }, + { + "epoch": 1.8812870634746321, + "grad_norm": 2.2827954292297363, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7218571305274963, + "num_tokens": 439554766.0, + "step": 17131 + }, + { + "epoch": 1.881396881177246, + "grad_norm": 2.2707765102386475, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7157772779464722, + "num_tokens": 439579671.0, + "step": 17132 + }, + { + "epoch": 1.8815066988798594, + "grad_norm": 2.111527442932129, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7453924417495728, + "num_tokens": 439606181.0, + "step": 17133 + }, + { + "epoch": 1.881616516582473, + "grad_norm": 2.2011773586273193, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7134657502174377, + "num_tokens": 439634013.0, + "step": 17134 + }, + { + "epoch": 1.8817263342850867, + "grad_norm": 2.29295015335083, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7366973161697388, + "num_tokens": 439658040.0, + "step": 17135 + }, + { + "epoch": 1.8818361519877005, + "grad_norm": 2.3526499271392822, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7386887073516846, + "num_tokens": 439680214.0, + "step": 17136 + }, + { + "epoch": 1.881945969690314, + "grad_norm": 2.394641637802124, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7240897417068481, + "num_tokens": 439704581.0, + "step": 17137 + }, + { + "epoch": 1.8820557873929278, + "grad_norm": 2.5765349864959717, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.747551441192627, + "num_tokens": 439724614.0, + "step": 17138 + }, + { + "epoch": 1.8821656050955413, + "grad_norm": 2.2854509353637695, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7367087602615356, + "num_tokens": 439748540.0, + "step": 17139 + }, + { + "epoch": 1.882275422798155, + "grad_norm": 2.360139846801758, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.701117753982544, + "num_tokens": 439774369.0, + "step": 17140 + }, + { + "epoch": 1.8823852405007688, + "grad_norm": 2.35730242729187, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7285751104354858, + "num_tokens": 439799102.0, + "step": 17141 + }, + { + "epoch": 1.8824950582033824, + "grad_norm": 1.8876551389694214, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7412369847297668, + "num_tokens": 439830525.0, + "step": 17142 + }, + { + "epoch": 1.882604875905996, + "grad_norm": 2.24646258354187, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7199053168296814, + "num_tokens": 439855963.0, + "step": 17143 + }, + { + "epoch": 1.8827146936086097, + "grad_norm": 2.342106342315674, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.72489333152771, + "num_tokens": 439880794.0, + "step": 17144 + }, + { + "epoch": 1.8828245113112234, + "grad_norm": 2.224302291870117, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7125765681266785, + "num_tokens": 439906049.0, + "step": 17145 + }, + { + "epoch": 1.8829343290138372, + "grad_norm": 2.1119625568389893, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7245367765426636, + "num_tokens": 439933646.0, + "step": 17146 + }, + { + "epoch": 1.8830441467164507, + "grad_norm": 2.314911127090454, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7000669240951538, + "num_tokens": 439959554.0, + "step": 17147 + }, + { + "epoch": 1.8831539644190642, + "grad_norm": 2.032630443572998, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7061263918876648, + "num_tokens": 439990078.0, + "step": 17148 + }, + { + "epoch": 1.883263782121678, + "grad_norm": 2.215423107147217, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.73894202709198, + "num_tokens": 440013022.0, + "step": 17149 + }, + { + "epoch": 1.8833735998242918, + "grad_norm": 2.120736837387085, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7162668704986572, + "num_tokens": 440040703.0, + "step": 17150 + }, + { + "epoch": 1.8834834175269053, + "grad_norm": 2.2990784645080566, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7300056219100952, + "num_tokens": 440064726.0, + "step": 17151 + }, + { + "epoch": 1.883593235229519, + "grad_norm": 2.220945119857788, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7409424781799316, + "num_tokens": 440088792.0, + "step": 17152 + }, + { + "epoch": 1.8837030529321326, + "grad_norm": 2.5419299602508545, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7243037223815918, + "num_tokens": 440109414.0, + "step": 17153 + }, + { + "epoch": 1.8838128706347463, + "grad_norm": 2.297677755355835, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7284080982208252, + "num_tokens": 440132806.0, + "step": 17154 + }, + { + "epoch": 1.88392268833736, + "grad_norm": 2.2356762886047363, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7232502698898315, + "num_tokens": 440157328.0, + "step": 17155 + }, + { + "epoch": 1.8840325060399736, + "grad_norm": 2.279435634613037, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7341068387031555, + "num_tokens": 440179481.0, + "step": 17156 + }, + { + "epoch": 1.8841423237425872, + "grad_norm": 2.306946277618408, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.708996593952179, + "num_tokens": 440204215.0, + "step": 17157 + }, + { + "epoch": 1.884252141445201, + "grad_norm": 2.223114013671875, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7258887887001038, + "num_tokens": 440229541.0, + "step": 17158 + }, + { + "epoch": 1.8843619591478147, + "grad_norm": 2.565802812576294, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7282814979553223, + "num_tokens": 440249796.0, + "step": 17159 + }, + { + "epoch": 1.8844717768504284, + "grad_norm": 2.1716599464416504, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7332372665405273, + "num_tokens": 440276289.0, + "step": 17160 + }, + { + "epoch": 1.884581594553042, + "grad_norm": 2.464937210083008, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7209146618843079, + "num_tokens": 440297338.0, + "step": 17161 + }, + { + "epoch": 1.8846914122556555, + "grad_norm": 2.32565975189209, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7330071926116943, + "num_tokens": 440319661.0, + "step": 17162 + }, + { + "epoch": 1.8848012299582693, + "grad_norm": 2.178950071334839, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7376990914344788, + "num_tokens": 440345944.0, + "step": 17163 + }, + { + "epoch": 1.884911047660883, + "grad_norm": 2.115593910217285, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.723345935344696, + "num_tokens": 440373883.0, + "step": 17164 + }, + { + "epoch": 1.8850208653634966, + "grad_norm": 2.0258841514587402, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7081412672996521, + "num_tokens": 440405694.0, + "step": 17165 + }, + { + "epoch": 1.88513068306611, + "grad_norm": 2.1881964206695557, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7154943346977234, + "num_tokens": 440432395.0, + "step": 17166 + }, + { + "epoch": 1.8852405007687238, + "grad_norm": 2.332244396209717, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7246100902557373, + "num_tokens": 440456790.0, + "step": 17167 + }, + { + "epoch": 1.8853503184713376, + "grad_norm": 2.2464041709899902, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7370998859405518, + "num_tokens": 440481203.0, + "step": 17168 + }, + { + "epoch": 1.8854601361739514, + "grad_norm": 2.2208251953125, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7282619476318359, + "num_tokens": 440507120.0, + "step": 17169 + }, + { + "epoch": 1.885569953876565, + "grad_norm": 2.020042896270752, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7097481489181519, + "num_tokens": 440535563.0, + "step": 17170 + }, + { + "epoch": 1.8856797715791784, + "grad_norm": 2.058638334274292, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.710431694984436, + "num_tokens": 440565907.0, + "step": 17171 + }, + { + "epoch": 1.8857895892817922, + "grad_norm": 2.3577864170074463, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7346106767654419, + "num_tokens": 440588280.0, + "step": 17172 + }, + { + "epoch": 1.885899406984406, + "grad_norm": 2.158055067062378, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7028024196624756, + "num_tokens": 440614270.0, + "step": 17173 + }, + { + "epoch": 1.8860092246870197, + "grad_norm": 2.670262575149536, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7238168716430664, + "num_tokens": 440635032.0, + "step": 17174 + }, + { + "epoch": 1.8861190423896332, + "grad_norm": 2.0409979820251465, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6901262402534485, + "num_tokens": 440665536.0, + "step": 17175 + }, + { + "epoch": 1.8862288600922468, + "grad_norm": 2.221525192260742, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7037980556488037, + "num_tokens": 440691488.0, + "step": 17176 + }, + { + "epoch": 1.8863386777948605, + "grad_norm": 2.127310276031494, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.713469922542572, + "num_tokens": 440718147.0, + "step": 17177 + }, + { + "epoch": 1.8864484954974743, + "grad_norm": 2.0841071605682373, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7159836888313293, + "num_tokens": 440746867.0, + "step": 17178 + }, + { + "epoch": 1.8865583132000878, + "grad_norm": 2.143430709838867, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7236164808273315, + "num_tokens": 440773292.0, + "step": 17179 + }, + { + "epoch": 1.8866681309027014, + "grad_norm": 1.8859065771102905, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7249369025230408, + "num_tokens": 440805148.0, + "step": 17180 + }, + { + "epoch": 1.8867779486053151, + "grad_norm": 2.053267002105713, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6906499862670898, + "num_tokens": 440837580.0, + "step": 17181 + }, + { + "epoch": 1.8868877663079289, + "grad_norm": 1.9976540803909302, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.701906681060791, + "num_tokens": 440870492.0, + "step": 17182 + }, + { + "epoch": 1.8869975840105426, + "grad_norm": 2.5615127086639404, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7311440110206604, + "num_tokens": 440890756.0, + "step": 17183 + }, + { + "epoch": 1.8871074017131562, + "grad_norm": 2.162304162979126, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.741348922252655, + "num_tokens": 440917350.0, + "step": 17184 + }, + { + "epoch": 1.8872172194157697, + "grad_norm": 2.257202625274658, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6958929300308228, + "num_tokens": 440944273.0, + "step": 17185 + }, + { + "epoch": 1.8873270371183835, + "grad_norm": 2.5210018157958984, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7309733629226685, + "num_tokens": 440965175.0, + "step": 17186 + }, + { + "epoch": 1.8874368548209972, + "grad_norm": 2.2309532165527344, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7193698287010193, + "num_tokens": 440990812.0, + "step": 17187 + }, + { + "epoch": 1.887546672523611, + "grad_norm": 2.103102445602417, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7295851707458496, + "num_tokens": 441017819.0, + "step": 17188 + }, + { + "epoch": 1.8876564902262245, + "grad_norm": 2.47310733795166, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7377268075942993, + "num_tokens": 441039388.0, + "step": 17189 + }, + { + "epoch": 1.887766307928838, + "grad_norm": 2.384110689163208, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6866265535354614, + "num_tokens": 441063813.0, + "step": 17190 + }, + { + "epoch": 1.8878761256314518, + "grad_norm": 2.1802549362182617, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7311618328094482, + "num_tokens": 441090487.0, + "step": 17191 + }, + { + "epoch": 1.8879859433340656, + "grad_norm": 1.9514142274856567, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7162567377090454, + "num_tokens": 441122951.0, + "step": 17192 + }, + { + "epoch": 1.888095761036679, + "grad_norm": 2.2905869483947754, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7326160669326782, + "num_tokens": 441147857.0, + "step": 17193 + }, + { + "epoch": 1.8882055787392926, + "grad_norm": 2.144757032394409, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7195985317230225, + "num_tokens": 441175531.0, + "step": 17194 + }, + { + "epoch": 1.8883153964419064, + "grad_norm": 2.25677490234375, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.7112628221511841, + "num_tokens": 441200178.0, + "step": 17195 + }, + { + "epoch": 1.8884252141445201, + "grad_norm": 2.1303367614746094, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7270900011062622, + "num_tokens": 441226528.0, + "step": 17196 + }, + { + "epoch": 1.888535031847134, + "grad_norm": 2.4382970333099365, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7336811423301697, + "num_tokens": 441246611.0, + "step": 17197 + }, + { + "epoch": 1.8886448495497474, + "grad_norm": 2.3039729595184326, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7107424736022949, + "num_tokens": 441271561.0, + "step": 17198 + }, + { + "epoch": 1.888754667252361, + "grad_norm": 2.2937417030334473, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6898776292800903, + "num_tokens": 441301013.0, + "step": 17199 + }, + { + "epoch": 1.8888644849549747, + "grad_norm": 2.0963008403778076, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7186477184295654, + "num_tokens": 441329334.0, + "step": 17200 + }, + { + "epoch": 1.8889743026575885, + "grad_norm": 2.406888008117676, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7118206024169922, + "num_tokens": 441353046.0, + "step": 17201 + }, + { + "epoch": 1.889084120360202, + "grad_norm": 2.429048538208008, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7420209646224976, + "num_tokens": 441373991.0, + "step": 17202 + }, + { + "epoch": 1.8891939380628158, + "grad_norm": 2.1289989948272705, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6908848881721497, + "num_tokens": 441403270.0, + "step": 17203 + }, + { + "epoch": 1.8893037557654293, + "grad_norm": 2.3961775302886963, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7404347658157349, + "num_tokens": 441426300.0, + "step": 17204 + }, + { + "epoch": 1.889413573468043, + "grad_norm": 2.0650346279144287, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7160710096359253, + "num_tokens": 441457348.0, + "step": 17205 + }, + { + "epoch": 1.8895233911706568, + "grad_norm": 2.453425407409668, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.70821213722229, + "num_tokens": 441480038.0, + "step": 17206 + }, + { + "epoch": 1.8896332088732704, + "grad_norm": 2.134596109390259, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7419707775115967, + "num_tokens": 441505525.0, + "step": 17207 + }, + { + "epoch": 1.8897430265758839, + "grad_norm": 2.2056422233581543, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.728702962398529, + "num_tokens": 441531142.0, + "step": 17208 + }, + { + "epoch": 1.8898528442784976, + "grad_norm": 2.399287223815918, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7410570383071899, + "num_tokens": 441556237.0, + "step": 17209 + }, + { + "epoch": 1.8899626619811114, + "grad_norm": 1.938428282737732, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.708604097366333, + "num_tokens": 441588267.0, + "step": 17210 + }, + { + "epoch": 1.8900724796837252, + "grad_norm": 2.369896173477173, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7259352803230286, + "num_tokens": 441613167.0, + "step": 17211 + }, + { + "epoch": 1.8901822973863387, + "grad_norm": 2.1572587490081787, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7343841791152954, + "num_tokens": 441638910.0, + "step": 17212 + }, + { + "epoch": 1.8902921150889522, + "grad_norm": 2.4751672744750977, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7314788103103638, + "num_tokens": 441659036.0, + "step": 17213 + }, + { + "epoch": 1.890401932791566, + "grad_norm": 2.541327953338623, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7319197058677673, + "num_tokens": 441680421.0, + "step": 17214 + }, + { + "epoch": 1.8905117504941797, + "grad_norm": 2.3348805904388428, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7178183197975159, + "num_tokens": 441703241.0, + "step": 17215 + }, + { + "epoch": 1.8906215681967933, + "grad_norm": 2.352581024169922, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7376017570495605, + "num_tokens": 441725642.0, + "step": 17216 + }, + { + "epoch": 1.890731385899407, + "grad_norm": 2.3391976356506348, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7211267352104187, + "num_tokens": 441750998.0, + "step": 17217 + }, + { + "epoch": 1.8908412036020206, + "grad_norm": 2.3633220195770264, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7304804921150208, + "num_tokens": 441772465.0, + "step": 17218 + }, + { + "epoch": 1.8909510213046343, + "grad_norm": 2.255527973175049, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7227978706359863, + "num_tokens": 441796082.0, + "step": 17219 + }, + { + "epoch": 1.891060839007248, + "grad_norm": 2.227837562561035, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7243145704269409, + "num_tokens": 441820983.0, + "step": 17220 + }, + { + "epoch": 1.8911706567098616, + "grad_norm": 2.111351728439331, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7205071449279785, + "num_tokens": 441848316.0, + "step": 17221 + }, + { + "epoch": 1.8912804744124752, + "grad_norm": 2.226020097732544, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7273756265640259, + "num_tokens": 441872504.0, + "step": 17222 + }, + { + "epoch": 1.891390292115089, + "grad_norm": 2.2941465377807617, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7235027551651001, + "num_tokens": 441897508.0, + "step": 17223 + }, + { + "epoch": 1.8915001098177027, + "grad_norm": 2.0484743118286133, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7002055644989014, + "num_tokens": 441926048.0, + "step": 17224 + }, + { + "epoch": 1.8916099275203164, + "grad_norm": 2.2847864627838135, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7082754373550415, + "num_tokens": 441954473.0, + "step": 17225 + }, + { + "epoch": 1.89171974522293, + "grad_norm": 2.3985345363616943, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7156757712364197, + "num_tokens": 441977001.0, + "step": 17226 + }, + { + "epoch": 1.8918295629255435, + "grad_norm": 2.064464807510376, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7085047960281372, + "num_tokens": 442006074.0, + "step": 17227 + }, + { + "epoch": 1.8919393806281573, + "grad_norm": 2.291050672531128, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7146885395050049, + "num_tokens": 442029235.0, + "step": 17228 + }, + { + "epoch": 1.892049198330771, + "grad_norm": 2.180145740509033, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7127104997634888, + "num_tokens": 442054583.0, + "step": 17229 + }, + { + "epoch": 1.8921590160333845, + "grad_norm": 2.345686912536621, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7402389049530029, + "num_tokens": 442077505.0, + "step": 17230 + }, + { + "epoch": 1.892268833735998, + "grad_norm": 2.5656116008758545, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7260881066322327, + "num_tokens": 442098576.0, + "step": 17231 + }, + { + "epoch": 1.8923786514386118, + "grad_norm": 2.0557363033294678, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7235864400863647, + "num_tokens": 442127992.0, + "step": 17232 + }, + { + "epoch": 1.8924884691412256, + "grad_norm": 2.2216835021972656, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7405593395233154, + "num_tokens": 442151814.0, + "step": 17233 + }, + { + "epoch": 1.8925982868438394, + "grad_norm": 2.0513410568237305, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7471620440483093, + "num_tokens": 442178381.0, + "step": 17234 + }, + { + "epoch": 1.8927081045464529, + "grad_norm": 2.2328009605407715, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7226136922836304, + "num_tokens": 442207853.0, + "step": 17235 + }, + { + "epoch": 1.8928179222490664, + "grad_norm": 2.210042953491211, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7407448291778564, + "num_tokens": 442232244.0, + "step": 17236 + }, + { + "epoch": 1.8929277399516802, + "grad_norm": 2.1583356857299805, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7387394905090332, + "num_tokens": 442260548.0, + "step": 17237 + }, + { + "epoch": 1.893037557654294, + "grad_norm": 2.245806932449341, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7139989137649536, + "num_tokens": 442286286.0, + "step": 17238 + }, + { + "epoch": 1.8931473753569077, + "grad_norm": 2.3147408962249756, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7265856862068176, + "num_tokens": 442310836.0, + "step": 17239 + }, + { + "epoch": 1.8932571930595212, + "grad_norm": 1.9971472024917603, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7012214660644531, + "num_tokens": 442342651.0, + "step": 17240 + }, + { + "epoch": 1.8933670107621348, + "grad_norm": 2.258380174636841, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7416427135467529, + "num_tokens": 442365488.0, + "step": 17241 + }, + { + "epoch": 1.8934768284647485, + "grad_norm": 2.247357130050659, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7067540884017944, + "num_tokens": 442394079.0, + "step": 17242 + }, + { + "epoch": 1.8935866461673623, + "grad_norm": 2.3752779960632324, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7328802943229675, + "num_tokens": 442417482.0, + "step": 17243 + }, + { + "epoch": 1.8936964638699758, + "grad_norm": 2.19633150100708, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7261356115341187, + "num_tokens": 442441782.0, + "step": 17244 + }, + { + "epoch": 1.8938062815725893, + "grad_norm": 2.31946063041687, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7159526348114014, + "num_tokens": 442466107.0, + "step": 17245 + }, + { + "epoch": 1.893916099275203, + "grad_norm": 2.2177040576934814, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7264360785484314, + "num_tokens": 442492400.0, + "step": 17246 + }, + { + "epoch": 1.8940259169778169, + "grad_norm": 2.0103840827941895, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7017440795898438, + "num_tokens": 442524517.0, + "step": 17247 + }, + { + "epoch": 1.8941357346804306, + "grad_norm": 2.217693567276001, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7123563885688782, + "num_tokens": 442552207.0, + "step": 17248 + }, + { + "epoch": 1.8942455523830442, + "grad_norm": 2.3050594329833984, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7382320165634155, + "num_tokens": 442576314.0, + "step": 17249 + }, + { + "epoch": 1.8943553700856577, + "grad_norm": 2.161648750305176, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7199407815933228, + "num_tokens": 442601942.0, + "step": 17250 + }, + { + "epoch": 1.8944651877882714, + "grad_norm": 2.041621208190918, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7206329703330994, + "num_tokens": 442631801.0, + "step": 17251 + }, + { + "epoch": 1.8945750054908852, + "grad_norm": 2.0899014472961426, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7017550468444824, + "num_tokens": 442661155.0, + "step": 17252 + }, + { + "epoch": 1.894684823193499, + "grad_norm": 2.0915446281433105, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7204408645629883, + "num_tokens": 442691004.0, + "step": 17253 + }, + { + "epoch": 1.8947946408961125, + "grad_norm": 2.382521867752075, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7644115090370178, + "num_tokens": 442712727.0, + "step": 17254 + }, + { + "epoch": 1.894904458598726, + "grad_norm": 2.340768575668335, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7178802490234375, + "num_tokens": 442736075.0, + "step": 17255 + }, + { + "epoch": 1.8950142763013398, + "grad_norm": 2.055298328399658, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7231854796409607, + "num_tokens": 442765652.0, + "step": 17256 + }, + { + "epoch": 1.8951240940039535, + "grad_norm": 2.326195478439331, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7176659107208252, + "num_tokens": 442788887.0, + "step": 17257 + }, + { + "epoch": 1.895233911706567, + "grad_norm": 2.2093193531036377, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7226120233535767, + "num_tokens": 442814867.0, + "step": 17258 + }, + { + "epoch": 1.8953437294091806, + "grad_norm": 2.3136298656463623, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7278401851654053, + "num_tokens": 442838900.0, + "step": 17259 + }, + { + "epoch": 1.8954535471117944, + "grad_norm": 2.2572386264801025, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.730365514755249, + "num_tokens": 442864125.0, + "step": 17260 + }, + { + "epoch": 1.8955633648144081, + "grad_norm": 2.1475868225097656, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7111180424690247, + "num_tokens": 442892682.0, + "step": 17261 + }, + { + "epoch": 1.8956731825170219, + "grad_norm": 2.4341330528259277, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7439061999320984, + "num_tokens": 442913927.0, + "step": 17262 + }, + { + "epoch": 1.8957830002196354, + "grad_norm": 2.2666501998901367, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6960835456848145, + "num_tokens": 442938450.0, + "step": 17263 + }, + { + "epoch": 1.895892817922249, + "grad_norm": 2.3758206367492676, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7139719724655151, + "num_tokens": 442962760.0, + "step": 17264 + }, + { + "epoch": 1.8960026356248627, + "grad_norm": 2.1060733795166016, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7210565805435181, + "num_tokens": 442990487.0, + "step": 17265 + }, + { + "epoch": 1.8961124533274765, + "grad_norm": 2.015441417694092, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7106097936630249, + "num_tokens": 443019395.0, + "step": 17266 + }, + { + "epoch": 1.89622227103009, + "grad_norm": 2.3482022285461426, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7196873426437378, + "num_tokens": 443041940.0, + "step": 17267 + }, + { + "epoch": 1.8963320887327038, + "grad_norm": 2.193920850753784, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7215419411659241, + "num_tokens": 443067483.0, + "step": 17268 + }, + { + "epoch": 1.8964419064353173, + "grad_norm": 2.4686450958251953, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.706688642501831, + "num_tokens": 443092122.0, + "step": 17269 + }, + { + "epoch": 1.896551724137931, + "grad_norm": 2.0547916889190674, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7389294505119324, + "num_tokens": 443120810.0, + "step": 17270 + }, + { + "epoch": 1.8966615418405448, + "grad_norm": 2.3618006706237793, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7013244032859802, + "num_tokens": 443145078.0, + "step": 17271 + }, + { + "epoch": 1.8967713595431583, + "grad_norm": 2.273528575897217, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7315362095832825, + "num_tokens": 443169692.0, + "step": 17272 + }, + { + "epoch": 1.8968811772457719, + "grad_norm": 2.1714367866516113, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7270153164863586, + "num_tokens": 443196207.0, + "step": 17273 + }, + { + "epoch": 1.8969909949483856, + "grad_norm": 2.274747610092163, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.731044352054596, + "num_tokens": 443220253.0, + "step": 17274 + }, + { + "epoch": 1.8971008126509994, + "grad_norm": 2.556236505508423, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7200981378555298, + "num_tokens": 443242260.0, + "step": 17275 + }, + { + "epoch": 1.8972106303536131, + "grad_norm": 2.311798095703125, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7214803695678711, + "num_tokens": 443266759.0, + "step": 17276 + }, + { + "epoch": 1.8973204480562267, + "grad_norm": 2.5966062545776367, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7134308815002441, + "num_tokens": 443288016.0, + "step": 17277 + }, + { + "epoch": 1.8974302657588402, + "grad_norm": 1.9928096532821655, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6998634338378906, + "num_tokens": 443318467.0, + "step": 17278 + }, + { + "epoch": 1.897540083461454, + "grad_norm": 2.2418830394744873, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7350386381149292, + "num_tokens": 443344125.0, + "step": 17279 + }, + { + "epoch": 1.8976499011640677, + "grad_norm": 2.0683228969573975, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7448739409446716, + "num_tokens": 443370549.0, + "step": 17280 + }, + { + "epoch": 1.8977597188666813, + "grad_norm": 2.5724337100982666, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7331593036651611, + "num_tokens": 443390053.0, + "step": 17281 + }, + { + "epoch": 1.897869536569295, + "grad_norm": 2.152249813079834, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7172735333442688, + "num_tokens": 443417353.0, + "step": 17282 + }, + { + "epoch": 1.8979793542719086, + "grad_norm": 2.293099880218506, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7161954045295715, + "num_tokens": 443441939.0, + "step": 17283 + }, + { + "epoch": 1.8980891719745223, + "grad_norm": 2.23344349861145, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7107650637626648, + "num_tokens": 443470200.0, + "step": 17284 + }, + { + "epoch": 1.898198989677136, + "grad_norm": 2.284299373626709, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7156046628952026, + "num_tokens": 443495237.0, + "step": 17285 + }, + { + "epoch": 1.8983088073797496, + "grad_norm": 2.1258668899536133, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.727371096611023, + "num_tokens": 443523111.0, + "step": 17286 + }, + { + "epoch": 1.8984186250823631, + "grad_norm": 2.0618228912353516, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7336634397506714, + "num_tokens": 443551613.0, + "step": 17287 + }, + { + "epoch": 1.898528442784977, + "grad_norm": 2.3406031131744385, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.721225380897522, + "num_tokens": 443577196.0, + "step": 17288 + }, + { + "epoch": 1.8986382604875907, + "grad_norm": 2.0685508251190186, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7217077612876892, + "num_tokens": 443608458.0, + "step": 17289 + }, + { + "epoch": 1.8987480781902044, + "grad_norm": 2.1331989765167236, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7068085074424744, + "num_tokens": 443638797.0, + "step": 17290 + }, + { + "epoch": 1.898857895892818, + "grad_norm": 2.436509370803833, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.741391658782959, + "num_tokens": 443661173.0, + "step": 17291 + }, + { + "epoch": 1.8989677135954315, + "grad_norm": 2.5056021213531494, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7364229559898376, + "num_tokens": 443681704.0, + "step": 17292 + }, + { + "epoch": 1.8990775312980452, + "grad_norm": 2.174670696258545, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7117050886154175, + "num_tokens": 443709382.0, + "step": 17293 + }, + { + "epoch": 1.899187349000659, + "grad_norm": 2.822575807571411, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7316663861274719, + "num_tokens": 443727466.0, + "step": 17294 + }, + { + "epoch": 1.8992971667032725, + "grad_norm": 2.0783164501190186, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7154711484909058, + "num_tokens": 443758639.0, + "step": 17295 + }, + { + "epoch": 1.899406984405886, + "grad_norm": 2.490022897720337, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7230197191238403, + "num_tokens": 443780638.0, + "step": 17296 + }, + { + "epoch": 1.8995168021084998, + "grad_norm": 2.613323211669922, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7450374960899353, + "num_tokens": 443799753.0, + "step": 17297 + }, + { + "epoch": 1.8996266198111136, + "grad_norm": 2.60648250579834, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7329376935958862, + "num_tokens": 443818557.0, + "step": 17298 + }, + { + "epoch": 1.8997364375137273, + "grad_norm": 2.758322238922119, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7277600169181824, + "num_tokens": 443837329.0, + "step": 17299 + }, + { + "epoch": 1.8998462552163409, + "grad_norm": 2.0963306427001953, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7006001472473145, + "num_tokens": 443867133.0, + "step": 17300 + }, + { + "epoch": 1.8999560729189544, + "grad_norm": 2.121487855911255, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7152755260467529, + "num_tokens": 443894721.0, + "step": 17301 + }, + { + "epoch": 1.9000658906215682, + "grad_norm": 2.323230266571045, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7103786468505859, + "num_tokens": 443918937.0, + "step": 17302 + }, + { + "epoch": 1.900175708324182, + "grad_norm": 2.079374074935913, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7232155799865723, + "num_tokens": 443948066.0, + "step": 17303 + }, + { + "epoch": 1.9002855260267957, + "grad_norm": 2.077637195587158, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7125442028045654, + "num_tokens": 443977313.0, + "step": 17304 + }, + { + "epoch": 1.9003953437294092, + "grad_norm": 2.195464611053467, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7240858674049377, + "num_tokens": 444002886.0, + "step": 17305 + }, + { + "epoch": 1.9005051614320227, + "grad_norm": 2.353417158126831, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7103651762008667, + "num_tokens": 444026767.0, + "step": 17306 + }, + { + "epoch": 1.9006149791346365, + "grad_norm": 2.2094337940216064, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7481611371040344, + "num_tokens": 444050057.0, + "step": 17307 + }, + { + "epoch": 1.9007247968372503, + "grad_norm": 2.069488286972046, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7401948571205139, + "num_tokens": 444076888.0, + "step": 17308 + }, + { + "epoch": 1.9008346145398638, + "grad_norm": 2.20637583732605, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7076496481895447, + "num_tokens": 444104127.0, + "step": 17309 + }, + { + "epoch": 1.9009444322424773, + "grad_norm": 2.0786967277526855, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.727047860622406, + "num_tokens": 444133430.0, + "step": 17310 + }, + { + "epoch": 1.901054249945091, + "grad_norm": 2.1222071647644043, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.707689106464386, + "num_tokens": 444162034.0, + "step": 17311 + }, + { + "epoch": 1.9011640676477048, + "grad_norm": 2.2306430339813232, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7373487949371338, + "num_tokens": 444187140.0, + "step": 17312 + }, + { + "epoch": 1.9012738853503186, + "grad_norm": 2.3650600910186768, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7195497751235962, + "num_tokens": 444209960.0, + "step": 17313 + }, + { + "epoch": 1.9013837030529321, + "grad_norm": 2.1673531532287598, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7152605056762695, + "num_tokens": 444238482.0, + "step": 17314 + }, + { + "epoch": 1.9014935207555457, + "grad_norm": 2.436410427093506, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7333909273147583, + "num_tokens": 444260504.0, + "step": 17315 + }, + { + "epoch": 1.9016033384581594, + "grad_norm": 2.5078976154327393, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.740401029586792, + "num_tokens": 444281981.0, + "step": 17316 + }, + { + "epoch": 1.9017131561607732, + "grad_norm": 2.1346733570098877, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7173112034797668, + "num_tokens": 444309973.0, + "step": 17317 + }, + { + "epoch": 1.9018229738633867, + "grad_norm": 2.347405195236206, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7149077653884888, + "num_tokens": 444333279.0, + "step": 17318 + }, + { + "epoch": 1.9019327915660005, + "grad_norm": 2.451925277709961, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7288582921028137, + "num_tokens": 444354839.0, + "step": 17319 + }, + { + "epoch": 1.902042609268614, + "grad_norm": 2.035405397415161, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7230921983718872, + "num_tokens": 444383500.0, + "step": 17320 + }, + { + "epoch": 1.9021524269712278, + "grad_norm": 2.3340234756469727, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7274737358093262, + "num_tokens": 444406402.0, + "step": 17321 + }, + { + "epoch": 1.9022622446738415, + "grad_norm": 2.4055070877075195, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7212576866149902, + "num_tokens": 444429528.0, + "step": 17322 + }, + { + "epoch": 1.902372062376455, + "grad_norm": 2.048025369644165, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.700847864151001, + "num_tokens": 444460541.0, + "step": 17323 + }, + { + "epoch": 1.9024818800790686, + "grad_norm": 2.299440622329712, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7331161499023438, + "num_tokens": 444485347.0, + "step": 17324 + }, + { + "epoch": 1.9025916977816824, + "grad_norm": 2.225524663925171, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7431143522262573, + "num_tokens": 444509046.0, + "step": 17325 + }, + { + "epoch": 1.9027015154842961, + "grad_norm": 2.3981106281280518, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7183955311775208, + "num_tokens": 444531525.0, + "step": 17326 + }, + { + "epoch": 1.9028113331869099, + "grad_norm": 2.0803284645080566, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7128273248672485, + "num_tokens": 444560856.0, + "step": 17327 + }, + { + "epoch": 1.9029211508895234, + "grad_norm": 2.1377663612365723, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.718686580657959, + "num_tokens": 444587615.0, + "step": 17328 + }, + { + "epoch": 1.903030968592137, + "grad_norm": 2.653761386871338, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7035273909568787, + "num_tokens": 444609993.0, + "step": 17329 + }, + { + "epoch": 1.9031407862947507, + "grad_norm": 2.0913002490997314, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7320488691329956, + "num_tokens": 444635776.0, + "step": 17330 + }, + { + "epoch": 1.9032506039973645, + "grad_norm": 2.037771701812744, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7115272283554077, + "num_tokens": 444668325.0, + "step": 17331 + }, + { + "epoch": 1.903360421699978, + "grad_norm": 2.274902105331421, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7200844287872314, + "num_tokens": 444693355.0, + "step": 17332 + }, + { + "epoch": 1.9034702394025917, + "grad_norm": 2.0825438499450684, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6959691047668457, + "num_tokens": 444722655.0, + "step": 17333 + }, + { + "epoch": 1.9035800571052053, + "grad_norm": 2.143120765686035, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7251970767974854, + "num_tokens": 444751692.0, + "step": 17334 + }, + { + "epoch": 1.903689874807819, + "grad_norm": 2.2097487449645996, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7190803289413452, + "num_tokens": 444776410.0, + "step": 17335 + }, + { + "epoch": 1.9037996925104328, + "grad_norm": 2.197359561920166, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7040382623672485, + "num_tokens": 444801587.0, + "step": 17336 + }, + { + "epoch": 1.9039095102130463, + "grad_norm": 2.194622278213501, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7233087420463562, + "num_tokens": 444828660.0, + "step": 17337 + }, + { + "epoch": 1.9040193279156599, + "grad_norm": 2.4713194370269775, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7444610595703125, + "num_tokens": 444850620.0, + "step": 17338 + }, + { + "epoch": 1.9041291456182736, + "grad_norm": 1.8931602239608765, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7079778909683228, + "num_tokens": 444886531.0, + "step": 17339 + }, + { + "epoch": 1.9042389633208874, + "grad_norm": 2.3669142723083496, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7236956357955933, + "num_tokens": 444909878.0, + "step": 17340 + }, + { + "epoch": 1.9043487810235011, + "grad_norm": 2.0622055530548096, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7064629197120667, + "num_tokens": 444939347.0, + "step": 17341 + }, + { + "epoch": 1.9044585987261147, + "grad_norm": 1.8597171306610107, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7365198731422424, + "num_tokens": 444974933.0, + "step": 17342 + }, + { + "epoch": 1.9045684164287282, + "grad_norm": 2.2288293838500977, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7242889404296875, + "num_tokens": 444999381.0, + "step": 17343 + }, + { + "epoch": 1.904678234131342, + "grad_norm": 2.100099802017212, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7281720638275146, + "num_tokens": 445025220.0, + "step": 17344 + }, + { + "epoch": 1.9047880518339557, + "grad_norm": 2.1866068840026855, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7330385446548462, + "num_tokens": 445050764.0, + "step": 17345 + }, + { + "epoch": 1.9048978695365693, + "grad_norm": 2.155303955078125, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7185587286949158, + "num_tokens": 445077972.0, + "step": 17346 + }, + { + "epoch": 1.9050076872391828, + "grad_norm": 2.4242024421691895, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7357947826385498, + "num_tokens": 445099035.0, + "step": 17347 + }, + { + "epoch": 1.9051175049417965, + "grad_norm": 2.6022849082946777, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7189753651618958, + "num_tokens": 445119531.0, + "step": 17348 + }, + { + "epoch": 1.9052273226444103, + "grad_norm": 2.2571980953216553, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7177659273147583, + "num_tokens": 445144444.0, + "step": 17349 + }, + { + "epoch": 1.905337140347024, + "grad_norm": 2.2246906757354736, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7197712659835815, + "num_tokens": 445169557.0, + "step": 17350 + }, + { + "epoch": 1.9054469580496376, + "grad_norm": 2.4713540077209473, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7307620644569397, + "num_tokens": 445190097.0, + "step": 17351 + }, + { + "epoch": 1.9055567757522511, + "grad_norm": 2.0773816108703613, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7250989675521851, + "num_tokens": 445220490.0, + "step": 17352 + }, + { + "epoch": 1.9056665934548649, + "grad_norm": 2.283947229385376, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7211938500404358, + "num_tokens": 445246531.0, + "step": 17353 + }, + { + "epoch": 1.9057764111574786, + "grad_norm": 2.377283811569214, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7266354560852051, + "num_tokens": 445269431.0, + "step": 17354 + }, + { + "epoch": 1.9058862288600924, + "grad_norm": 1.9940471649169922, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6854972839355469, + "num_tokens": 445301510.0, + "step": 17355 + }, + { + "epoch": 1.905996046562706, + "grad_norm": 2.5735862255096436, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7468492984771729, + "num_tokens": 445321721.0, + "step": 17356 + }, + { + "epoch": 1.9061058642653195, + "grad_norm": 2.3778696060180664, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7405499219894409, + "num_tokens": 445342732.0, + "step": 17357 + }, + { + "epoch": 1.9062156819679332, + "grad_norm": 2.231401205062866, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7011909484863281, + "num_tokens": 445370101.0, + "step": 17358 + }, + { + "epoch": 1.906325499670547, + "grad_norm": 2.251523971557617, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.737875759601593, + "num_tokens": 445393532.0, + "step": 17359 + }, + { + "epoch": 1.9064353173731605, + "grad_norm": 2.1804442405700684, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7273355722427368, + "num_tokens": 445420189.0, + "step": 17360 + }, + { + "epoch": 1.906545135075774, + "grad_norm": 2.262009620666504, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7405421733856201, + "num_tokens": 445446059.0, + "step": 17361 + }, + { + "epoch": 1.9066549527783878, + "grad_norm": 2.377347707748413, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7207345962524414, + "num_tokens": 445471854.0, + "step": 17362 + }, + { + "epoch": 1.9067647704810016, + "grad_norm": 2.514216899871826, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7262531518936157, + "num_tokens": 445493748.0, + "step": 17363 + }, + { + "epoch": 1.9068745881836153, + "grad_norm": 2.302349090576172, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7363438606262207, + "num_tokens": 445517015.0, + "step": 17364 + }, + { + "epoch": 1.9069844058862289, + "grad_norm": 2.2676377296447754, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7122284770011902, + "num_tokens": 445545240.0, + "step": 17365 + }, + { + "epoch": 1.9070942235888424, + "grad_norm": 2.1451833248138428, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7452924251556396, + "num_tokens": 445570523.0, + "step": 17366 + }, + { + "epoch": 1.9072040412914562, + "grad_norm": 2.155057907104492, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7020787596702576, + "num_tokens": 445599981.0, + "step": 17367 + }, + { + "epoch": 1.90731385899407, + "grad_norm": 2.1784791946411133, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7217872142791748, + "num_tokens": 445627317.0, + "step": 17368 + }, + { + "epoch": 1.9074236766966837, + "grad_norm": 2.2182960510253906, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7392791509628296, + "num_tokens": 445651548.0, + "step": 17369 + }, + { + "epoch": 1.9075334943992972, + "grad_norm": 2.1303622722625732, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7056204676628113, + "num_tokens": 445679413.0, + "step": 17370 + }, + { + "epoch": 1.9076433121019107, + "grad_norm": 1.9191877841949463, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7217622399330139, + "num_tokens": 445714903.0, + "step": 17371 + }, + { + "epoch": 1.9077531298045245, + "grad_norm": 2.0925068855285645, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6868414878845215, + "num_tokens": 445745475.0, + "step": 17372 + }, + { + "epoch": 1.9078629475071383, + "grad_norm": 2.12750244140625, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7017059326171875, + "num_tokens": 445772457.0, + "step": 17373 + }, + { + "epoch": 1.9079727652097518, + "grad_norm": 2.2412476539611816, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7333297729492188, + "num_tokens": 445796486.0, + "step": 17374 + }, + { + "epoch": 1.9080825829123653, + "grad_norm": 2.3760037422180176, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7191615104675293, + "num_tokens": 445819460.0, + "step": 17375 + }, + { + "epoch": 1.908192400614979, + "grad_norm": 2.1630144119262695, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7302943468093872, + "num_tokens": 445844654.0, + "step": 17376 + }, + { + "epoch": 1.9083022183175928, + "grad_norm": 2.2264952659606934, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7176432609558105, + "num_tokens": 445871020.0, + "step": 17377 + }, + { + "epoch": 1.9084120360202066, + "grad_norm": 2.294255018234253, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7366621494293213, + "num_tokens": 445895236.0, + "step": 17378 + }, + { + "epoch": 1.9085218537228201, + "grad_norm": 2.283329486846924, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7172807455062866, + "num_tokens": 445919270.0, + "step": 17379 + }, + { + "epoch": 1.9086316714254337, + "grad_norm": 2.1941657066345215, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7158221006393433, + "num_tokens": 445945902.0, + "step": 17380 + }, + { + "epoch": 1.9087414891280474, + "grad_norm": 2.3085672855377197, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7261743545532227, + "num_tokens": 445969382.0, + "step": 17381 + }, + { + "epoch": 1.9088513068306612, + "grad_norm": 2.4719104766845703, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.721773624420166, + "num_tokens": 445991255.0, + "step": 17382 + }, + { + "epoch": 1.9089611245332747, + "grad_norm": 2.2676026821136475, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7349071502685547, + "num_tokens": 446015812.0, + "step": 17383 + }, + { + "epoch": 1.9090709422358885, + "grad_norm": 2.620192527770996, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7437726259231567, + "num_tokens": 446033626.0, + "step": 17384 + }, + { + "epoch": 1.909180759938502, + "grad_norm": 2.240253448486328, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7163558602333069, + "num_tokens": 446057187.0, + "step": 17385 + }, + { + "epoch": 1.9092905776411158, + "grad_norm": 2.1011178493499756, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7020915150642395, + "num_tokens": 446087585.0, + "step": 17386 + }, + { + "epoch": 1.9094003953437295, + "grad_norm": 2.305910110473633, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7232322096824646, + "num_tokens": 446112653.0, + "step": 17387 + }, + { + "epoch": 1.909510213046343, + "grad_norm": 2.353793144226074, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7156183123588562, + "num_tokens": 446136615.0, + "step": 17388 + }, + { + "epoch": 1.9096200307489566, + "grad_norm": 2.209749221801758, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7160352468490601, + "num_tokens": 446163075.0, + "step": 17389 + }, + { + "epoch": 1.9097298484515703, + "grad_norm": 2.4989776611328125, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7372510433197021, + "num_tokens": 446184558.0, + "step": 17390 + }, + { + "epoch": 1.909839666154184, + "grad_norm": 2.4264698028564453, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7194095849990845, + "num_tokens": 446208159.0, + "step": 17391 + }, + { + "epoch": 1.9099494838567979, + "grad_norm": 1.9729570150375366, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7149139642715454, + "num_tokens": 446239060.0, + "step": 17392 + }, + { + "epoch": 1.9100593015594114, + "grad_norm": 2.6359846591949463, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7155797481536865, + "num_tokens": 446258139.0, + "step": 17393 + }, + { + "epoch": 1.910169119262025, + "grad_norm": 2.257080078125, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7129573822021484, + "num_tokens": 446284941.0, + "step": 17394 + }, + { + "epoch": 1.9102789369646387, + "grad_norm": 2.1847736835479736, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7312064170837402, + "num_tokens": 446309913.0, + "step": 17395 + }, + { + "epoch": 1.9103887546672524, + "grad_norm": 2.0241968631744385, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7146349549293518, + "num_tokens": 446340619.0, + "step": 17396 + }, + { + "epoch": 1.910498572369866, + "grad_norm": 2.3650529384613037, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7556642293930054, + "num_tokens": 446361768.0, + "step": 17397 + }, + { + "epoch": 1.9106083900724797, + "grad_norm": 2.0410633087158203, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7113322019577026, + "num_tokens": 446392468.0, + "step": 17398 + }, + { + "epoch": 1.9107182077750933, + "grad_norm": 2.341118097305298, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7386566400527954, + "num_tokens": 446416886.0, + "step": 17399 + }, + { + "epoch": 1.910828025477707, + "grad_norm": 2.1484909057617188, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7299813032150269, + "num_tokens": 446441952.0, + "step": 17400 + }, + { + "epoch": 1.9109378431803208, + "grad_norm": 2.1687369346618652, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7219681739807129, + "num_tokens": 446468359.0, + "step": 17401 + }, + { + "epoch": 1.9110476608829343, + "grad_norm": 2.083760976791382, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7172408103942871, + "num_tokens": 446495645.0, + "step": 17402 + }, + { + "epoch": 1.9111574785855479, + "grad_norm": 2.7358241081237793, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7245141267776489, + "num_tokens": 446514240.0, + "step": 17403 + }, + { + "epoch": 1.9112672962881616, + "grad_norm": 2.088829755783081, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.720362663269043, + "num_tokens": 446543531.0, + "step": 17404 + }, + { + "epoch": 1.9113771139907754, + "grad_norm": 2.471663475036621, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7314369678497314, + "num_tokens": 446564807.0, + "step": 17405 + }, + { + "epoch": 1.9114869316933891, + "grad_norm": 2.481288194656372, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7349364757537842, + "num_tokens": 446587644.0, + "step": 17406 + }, + { + "epoch": 1.9115967493960027, + "grad_norm": 2.388789415359497, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7345297932624817, + "num_tokens": 446608254.0, + "step": 17407 + }, + { + "epoch": 1.9117065670986162, + "grad_norm": 2.3645923137664795, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.726873517036438, + "num_tokens": 446631048.0, + "step": 17408 + }, + { + "epoch": 1.91181638480123, + "grad_norm": 2.1798219680786133, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7221206426620483, + "num_tokens": 446660033.0, + "step": 17409 + }, + { + "epoch": 1.9119262025038437, + "grad_norm": 2.0507917404174805, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7121509909629822, + "num_tokens": 446689124.0, + "step": 17410 + }, + { + "epoch": 1.9120360202064572, + "grad_norm": 2.3588907718658447, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7099111676216125, + "num_tokens": 446711496.0, + "step": 17411 + }, + { + "epoch": 1.9121458379090708, + "grad_norm": 2.214324951171875, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7327803373336792, + "num_tokens": 446737419.0, + "step": 17412 + }, + { + "epoch": 1.9122556556116845, + "grad_norm": 2.0926849842071533, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.743285059928894, + "num_tokens": 446763604.0, + "step": 17413 + }, + { + "epoch": 1.9123654733142983, + "grad_norm": 2.3677899837493896, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.733157753944397, + "num_tokens": 446787355.0, + "step": 17414 + }, + { + "epoch": 1.912475291016912, + "grad_norm": 2.2264657020568848, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.699324369430542, + "num_tokens": 446813884.0, + "step": 17415 + }, + { + "epoch": 1.9125851087195256, + "grad_norm": 2.339454174041748, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7237966060638428, + "num_tokens": 446838103.0, + "step": 17416 + }, + { + "epoch": 1.9126949264221391, + "grad_norm": 2.151219129562378, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7394399642944336, + "num_tokens": 446864486.0, + "step": 17417 + }, + { + "epoch": 1.9128047441247529, + "grad_norm": 2.2878355979919434, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7388176918029785, + "num_tokens": 446888768.0, + "step": 17418 + }, + { + "epoch": 1.9129145618273666, + "grad_norm": 2.132199764251709, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7427108287811279, + "num_tokens": 446914729.0, + "step": 17419 + }, + { + "epoch": 1.9130243795299804, + "grad_norm": 2.4985945224761963, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7311110496520996, + "num_tokens": 446936544.0, + "step": 17420 + }, + { + "epoch": 1.913134197232594, + "grad_norm": 2.2883224487304688, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7361130118370056, + "num_tokens": 446960745.0, + "step": 17421 + }, + { + "epoch": 1.9132440149352075, + "grad_norm": 2.2679378986358643, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7164166569709778, + "num_tokens": 446987996.0, + "step": 17422 + }, + { + "epoch": 1.9133538326378212, + "grad_norm": 2.4723353385925293, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7112452983856201, + "num_tokens": 447012888.0, + "step": 17423 + }, + { + "epoch": 1.913463650340435, + "grad_norm": 2.4349632263183594, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7213973999023438, + "num_tokens": 447037078.0, + "step": 17424 + }, + { + "epoch": 1.9135734680430485, + "grad_norm": 2.1468310356140137, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.722624659538269, + "num_tokens": 447064555.0, + "step": 17425 + }, + { + "epoch": 1.913683285745662, + "grad_norm": 2.2736599445343018, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7280284762382507, + "num_tokens": 447088701.0, + "step": 17426 + }, + { + "epoch": 1.9137931034482758, + "grad_norm": 2.3431320190429688, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7195297479629517, + "num_tokens": 447112248.0, + "step": 17427 + }, + { + "epoch": 1.9139029211508896, + "grad_norm": 1.8515042066574097, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7149335145950317, + "num_tokens": 447147383.0, + "step": 17428 + }, + { + "epoch": 1.9140127388535033, + "grad_norm": 2.5027852058410645, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7216893434524536, + "num_tokens": 447168575.0, + "step": 17429 + }, + { + "epoch": 1.9141225565561168, + "grad_norm": 2.230624198913574, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7111755013465881, + "num_tokens": 447195711.0, + "step": 17430 + }, + { + "epoch": 1.9142323742587304, + "grad_norm": 2.435100555419922, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.710993766784668, + "num_tokens": 447218169.0, + "step": 17431 + }, + { + "epoch": 1.9143421919613441, + "grad_norm": 2.181105613708496, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7167165279388428, + "num_tokens": 447244691.0, + "step": 17432 + }, + { + "epoch": 1.914452009663958, + "grad_norm": 2.0983269214630127, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7483668327331543, + "num_tokens": 447272374.0, + "step": 17433 + }, + { + "epoch": 1.9145618273665717, + "grad_norm": 2.408505916595459, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7353765964508057, + "num_tokens": 447294213.0, + "step": 17434 + }, + { + "epoch": 1.9146716450691852, + "grad_norm": 2.4317848682403564, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7336363196372986, + "num_tokens": 447315361.0, + "step": 17435 + }, + { + "epoch": 1.9147814627717987, + "grad_norm": 2.3075249195098877, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7022591829299927, + "num_tokens": 447340432.0, + "step": 17436 + }, + { + "epoch": 1.9148912804744125, + "grad_norm": 2.405716896057129, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6986477971076965, + "num_tokens": 447366337.0, + "step": 17437 + }, + { + "epoch": 1.9150010981770262, + "grad_norm": 2.3093652725219727, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7038231492042542, + "num_tokens": 447392601.0, + "step": 17438 + }, + { + "epoch": 1.9151109158796398, + "grad_norm": 2.195396661758423, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7047482132911682, + "num_tokens": 447421342.0, + "step": 17439 + }, + { + "epoch": 1.9152207335822533, + "grad_norm": 2.4853436946868896, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7313916683197021, + "num_tokens": 447443661.0, + "step": 17440 + }, + { + "epoch": 1.915330551284867, + "grad_norm": 2.086163282394409, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7280000448226929, + "num_tokens": 447473447.0, + "step": 17441 + }, + { + "epoch": 1.9154403689874808, + "grad_norm": 2.2727010250091553, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7467525005340576, + "num_tokens": 447497632.0, + "step": 17442 + }, + { + "epoch": 1.9155501866900946, + "grad_norm": 2.0183017253875732, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.707812488079071, + "num_tokens": 447529906.0, + "step": 17443 + }, + { + "epoch": 1.9156600043927081, + "grad_norm": 2.330627202987671, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7256661653518677, + "num_tokens": 447554263.0, + "step": 17444 + }, + { + "epoch": 1.9157698220953217, + "grad_norm": 2.2298669815063477, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7164150476455688, + "num_tokens": 447580390.0, + "step": 17445 + }, + { + "epoch": 1.9158796397979354, + "grad_norm": 2.4243056774139404, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7362678647041321, + "num_tokens": 447602412.0, + "step": 17446 + }, + { + "epoch": 1.9159894575005492, + "grad_norm": 2.212303638458252, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7031223177909851, + "num_tokens": 447629363.0, + "step": 17447 + }, + { + "epoch": 1.9160992752031627, + "grad_norm": 2.0487191677093506, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7038897275924683, + "num_tokens": 447659016.0, + "step": 17448 + }, + { + "epoch": 1.9162090929057765, + "grad_norm": 2.076852798461914, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7295491695404053, + "num_tokens": 447688289.0, + "step": 17449 + }, + { + "epoch": 1.91631891060839, + "grad_norm": 2.0482187271118164, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7122193574905396, + "num_tokens": 447717580.0, + "step": 17450 + }, + { + "epoch": 1.9164287283110037, + "grad_norm": 2.0633444786071777, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7244194149971008, + "num_tokens": 447746149.0, + "step": 17451 + }, + { + "epoch": 1.9165385460136175, + "grad_norm": 2.2915351390838623, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.6976149082183838, + "num_tokens": 447772200.0, + "step": 17452 + }, + { + "epoch": 1.916648363716231, + "grad_norm": 2.2104008197784424, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7163363695144653, + "num_tokens": 447796840.0, + "step": 17453 + }, + { + "epoch": 1.9167581814188446, + "grad_norm": 2.239612579345703, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7245806455612183, + "num_tokens": 447822455.0, + "step": 17454 + }, + { + "epoch": 1.9168679991214583, + "grad_norm": 2.191545248031616, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7189003229141235, + "num_tokens": 447849422.0, + "step": 17455 + }, + { + "epoch": 1.916977816824072, + "grad_norm": 2.18660306930542, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7134265899658203, + "num_tokens": 447874931.0, + "step": 17456 + }, + { + "epoch": 1.9170876345266858, + "grad_norm": 2.4014735221862793, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.737277090549469, + "num_tokens": 447896905.0, + "step": 17457 + }, + { + "epoch": 1.9171974522292994, + "grad_norm": 2.448458671569824, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7038432359695435, + "num_tokens": 447922267.0, + "step": 17458 + }, + { + "epoch": 1.917307269931913, + "grad_norm": 2.141312599182129, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7152707576751709, + "num_tokens": 447949942.0, + "step": 17459 + }, + { + "epoch": 1.9174170876345267, + "grad_norm": 1.9125157594680786, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7119213342666626, + "num_tokens": 447983419.0, + "step": 17460 + }, + { + "epoch": 1.9175269053371404, + "grad_norm": 2.383688449859619, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.735090434551239, + "num_tokens": 448007621.0, + "step": 17461 + }, + { + "epoch": 1.917636723039754, + "grad_norm": 2.130861759185791, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.722099244594574, + "num_tokens": 448035006.0, + "step": 17462 + }, + { + "epoch": 1.9177465407423677, + "grad_norm": 2.190908670425415, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7320539355278015, + "num_tokens": 448059831.0, + "step": 17463 + }, + { + "epoch": 1.9178563584449813, + "grad_norm": 2.449324131011963, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.707821249961853, + "num_tokens": 448082184.0, + "step": 17464 + }, + { + "epoch": 1.917966176147595, + "grad_norm": 2.124272346496582, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7303200960159302, + "num_tokens": 448108161.0, + "step": 17465 + }, + { + "epoch": 1.9180759938502088, + "grad_norm": 2.0478715896606445, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6981126666069031, + "num_tokens": 448139366.0, + "step": 17466 + }, + { + "epoch": 1.9181858115528223, + "grad_norm": 2.133298635482788, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7266026735305786, + "num_tokens": 448166398.0, + "step": 17467 + }, + { + "epoch": 1.9182956292554358, + "grad_norm": 2.326218843460083, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7103357315063477, + "num_tokens": 448191496.0, + "step": 17468 + }, + { + "epoch": 1.9184054469580496, + "grad_norm": 2.3382151126861572, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.721712589263916, + "num_tokens": 448214814.0, + "step": 17469 + }, + { + "epoch": 1.9185152646606634, + "grad_norm": 2.263638734817505, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.714493453502655, + "num_tokens": 448238912.0, + "step": 17470 + }, + { + "epoch": 1.9186250823632771, + "grad_norm": 2.178638458251953, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.695565938949585, + "num_tokens": 448265422.0, + "step": 17471 + }, + { + "epoch": 1.9187349000658906, + "grad_norm": 2.5311245918273926, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7101637125015259, + "num_tokens": 448287876.0, + "step": 17472 + }, + { + "epoch": 1.9188447177685042, + "grad_norm": 2.3200559616088867, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7607858180999756, + "num_tokens": 448310085.0, + "step": 17473 + }, + { + "epoch": 1.918954535471118, + "grad_norm": 2.1801018714904785, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7194870710372925, + "num_tokens": 448336776.0, + "step": 17474 + }, + { + "epoch": 1.9190643531737317, + "grad_norm": 2.0534589290618896, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7357141375541687, + "num_tokens": 448364388.0, + "step": 17475 + }, + { + "epoch": 1.9191741708763452, + "grad_norm": 1.9055067300796509, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7157132625579834, + "num_tokens": 448396743.0, + "step": 17476 + }, + { + "epoch": 1.9192839885789588, + "grad_norm": 2.3949007987976074, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7388191819190979, + "num_tokens": 448418403.0, + "step": 17477 + }, + { + "epoch": 1.9193938062815725, + "grad_norm": 2.0040063858032227, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6993491649627686, + "num_tokens": 448449087.0, + "step": 17478 + }, + { + "epoch": 1.9195036239841863, + "grad_norm": 1.932243824005127, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7217916250228882, + "num_tokens": 448479704.0, + "step": 17479 + }, + { + "epoch": 1.9196134416868, + "grad_norm": 2.5198116302490234, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7306771278381348, + "num_tokens": 448499182.0, + "step": 17480 + }, + { + "epoch": 1.9197232593894136, + "grad_norm": 2.6715829372406006, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7437212467193604, + "num_tokens": 448517972.0, + "step": 17481 + }, + { + "epoch": 1.919833077092027, + "grad_norm": 2.0968363285064697, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7113253474235535, + "num_tokens": 448547257.0, + "step": 17482 + }, + { + "epoch": 1.9199428947946409, + "grad_norm": 2.3644633293151855, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7420606017112732, + "num_tokens": 448568977.0, + "step": 17483 + }, + { + "epoch": 1.9200527124972546, + "grad_norm": 2.1218478679656982, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7049566507339478, + "num_tokens": 448597583.0, + "step": 17484 + }, + { + "epoch": 1.9201625301998684, + "grad_norm": 2.0463078022003174, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7045996189117432, + "num_tokens": 448626560.0, + "step": 17485 + }, + { + "epoch": 1.920272347902482, + "grad_norm": 2.142803430557251, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.706350564956665, + "num_tokens": 448655325.0, + "step": 17486 + }, + { + "epoch": 1.9203821656050954, + "grad_norm": 2.492814302444458, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7563953399658203, + "num_tokens": 448673901.0, + "step": 17487 + }, + { + "epoch": 1.9204919833077092, + "grad_norm": 2.3244760036468506, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7026746869087219, + "num_tokens": 448696957.0, + "step": 17488 + }, + { + "epoch": 1.920601801010323, + "grad_norm": 2.0281198024749756, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7226285934448242, + "num_tokens": 448729445.0, + "step": 17489 + }, + { + "epoch": 1.9207116187129365, + "grad_norm": 2.465758800506592, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7154064178466797, + "num_tokens": 448753717.0, + "step": 17490 + }, + { + "epoch": 1.92082143641555, + "grad_norm": 2.4266791343688965, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7093174457550049, + "num_tokens": 448778762.0, + "step": 17491 + }, + { + "epoch": 1.9209312541181638, + "grad_norm": 3.1602301597595215, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7228976488113403, + "num_tokens": 448798722.0, + "step": 17492 + }, + { + "epoch": 1.9210410718207775, + "grad_norm": 2.1420738697052, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7187031507492065, + "num_tokens": 448828677.0, + "step": 17493 + }, + { + "epoch": 1.9211508895233913, + "grad_norm": 2.627730369567871, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7450917363166809, + "num_tokens": 448849148.0, + "step": 17494 + }, + { + "epoch": 1.9212607072260048, + "grad_norm": 2.211501121520996, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7287534475326538, + "num_tokens": 448874446.0, + "step": 17495 + }, + { + "epoch": 1.9213705249286184, + "grad_norm": 2.1038103103637695, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7223358154296875, + "num_tokens": 448905996.0, + "step": 17496 + }, + { + "epoch": 1.9214803426312321, + "grad_norm": 2.072554111480713, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7499141097068787, + "num_tokens": 448933857.0, + "step": 17497 + }, + { + "epoch": 1.9215901603338459, + "grad_norm": 2.059293270111084, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7200133800506592, + "num_tokens": 448963100.0, + "step": 17498 + }, + { + "epoch": 1.9216999780364594, + "grad_norm": 2.1209592819213867, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7069846987724304, + "num_tokens": 448992217.0, + "step": 17499 + }, + { + "epoch": 1.9218097957390732, + "grad_norm": 2.2344062328338623, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6967917680740356, + "num_tokens": 449019399.0, + "step": 17500 + }, + { + "epoch": 1.9219196134416867, + "grad_norm": 2.283437490463257, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7029428482055664, + "num_tokens": 449044697.0, + "step": 17501 + }, + { + "epoch": 1.9220294311443005, + "grad_norm": 2.604893207550049, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7112594842910767, + "num_tokens": 449064568.0, + "step": 17502 + }, + { + "epoch": 1.9221392488469142, + "grad_norm": 2.4344940185546875, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7359657883644104, + "num_tokens": 449087221.0, + "step": 17503 + }, + { + "epoch": 1.9222490665495278, + "grad_norm": 2.1855835914611816, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7074775695800781, + "num_tokens": 449114506.0, + "step": 17504 + }, + { + "epoch": 1.9223588842521413, + "grad_norm": 2.325016736984253, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7275699377059937, + "num_tokens": 449137560.0, + "step": 17505 + }, + { + "epoch": 1.922468701954755, + "grad_norm": 2.0662646293640137, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6908961534500122, + "num_tokens": 449168076.0, + "step": 17506 + }, + { + "epoch": 1.9225785196573688, + "grad_norm": 2.4256296157836914, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.727186381816864, + "num_tokens": 449189323.0, + "step": 17507 + }, + { + "epoch": 1.9226883373599826, + "grad_norm": 2.57389235496521, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7585803866386414, + "num_tokens": 449207924.0, + "step": 17508 + }, + { + "epoch": 1.922798155062596, + "grad_norm": 2.5765950679779053, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7241179943084717, + "num_tokens": 449228087.0, + "step": 17509 + }, + { + "epoch": 1.9229079727652096, + "grad_norm": 2.1665520668029785, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7220045328140259, + "num_tokens": 449255534.0, + "step": 17510 + }, + { + "epoch": 1.9230177904678234, + "grad_norm": 2.5509238243103027, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7428313493728638, + "num_tokens": 449276711.0, + "step": 17511 + }, + { + "epoch": 1.9231276081704372, + "grad_norm": 1.9634076356887817, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7046204805374146, + "num_tokens": 449310204.0, + "step": 17512 + }, + { + "epoch": 1.9232374258730507, + "grad_norm": 2.0674383640289307, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.712631344795227, + "num_tokens": 449339437.0, + "step": 17513 + }, + { + "epoch": 1.9233472435756644, + "grad_norm": 2.388941764831543, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7142320871353149, + "num_tokens": 449362791.0, + "step": 17514 + }, + { + "epoch": 1.923457061278278, + "grad_norm": 2.259270191192627, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6879507899284363, + "num_tokens": 449389937.0, + "step": 17515 + }, + { + "epoch": 1.9235668789808917, + "grad_norm": 2.2470061779022217, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7334040403366089, + "num_tokens": 449413931.0, + "step": 17516 + }, + { + "epoch": 1.9236766966835055, + "grad_norm": 1.9997022151947021, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7277622222900391, + "num_tokens": 449444044.0, + "step": 17517 + }, + { + "epoch": 1.923786514386119, + "grad_norm": 2.4486846923828125, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7336724400520325, + "num_tokens": 449465326.0, + "step": 17518 + }, + { + "epoch": 1.9238963320887326, + "grad_norm": 2.211251735687256, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7076714038848877, + "num_tokens": 449490958.0, + "step": 17519 + }, + { + "epoch": 1.9240061497913463, + "grad_norm": 2.5359489917755127, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7328875660896301, + "num_tokens": 449511501.0, + "step": 17520 + }, + { + "epoch": 1.92411596749396, + "grad_norm": 2.1929051876068115, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.709531307220459, + "num_tokens": 449537935.0, + "step": 17521 + }, + { + "epoch": 1.9242257851965738, + "grad_norm": 2.2216618061065674, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.704458475112915, + "num_tokens": 449566246.0, + "step": 17522 + }, + { + "epoch": 1.9243356028991874, + "grad_norm": 2.3016629219055176, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7197226285934448, + "num_tokens": 449591285.0, + "step": 17523 + }, + { + "epoch": 1.924445420601801, + "grad_norm": 2.2182505130767822, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7113118171691895, + "num_tokens": 449618193.0, + "step": 17524 + }, + { + "epoch": 1.9245552383044147, + "grad_norm": 2.166116237640381, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7212051153182983, + "num_tokens": 449644562.0, + "step": 17525 + }, + { + "epoch": 1.9246650560070284, + "grad_norm": 2.3477768898010254, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7353549003601074, + "num_tokens": 449668053.0, + "step": 17526 + }, + { + "epoch": 1.924774873709642, + "grad_norm": 2.6364619731903076, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7205204367637634, + "num_tokens": 449687487.0, + "step": 17527 + }, + { + "epoch": 1.9248846914122557, + "grad_norm": 2.393364191055298, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7271032929420471, + "num_tokens": 449711150.0, + "step": 17528 + }, + { + "epoch": 1.9249945091148692, + "grad_norm": 2.2781591415405273, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7251936197280884, + "num_tokens": 449737300.0, + "step": 17529 + }, + { + "epoch": 1.925104326817483, + "grad_norm": 2.4699480533599854, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7022616863250732, + "num_tokens": 449760816.0, + "step": 17530 + }, + { + "epoch": 1.9252141445200968, + "grad_norm": 2.37219500541687, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7209002375602722, + "num_tokens": 449785302.0, + "step": 17531 + }, + { + "epoch": 1.9253239622227103, + "grad_norm": 2.249958038330078, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7510004639625549, + "num_tokens": 449812526.0, + "step": 17532 + }, + { + "epoch": 1.9254337799253238, + "grad_norm": 1.8947196006774902, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7056936025619507, + "num_tokens": 449844472.0, + "step": 17533 + }, + { + "epoch": 1.9255435976279376, + "grad_norm": 2.359433889389038, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7360562086105347, + "num_tokens": 449867121.0, + "step": 17534 + }, + { + "epoch": 1.9256534153305513, + "grad_norm": 2.2225797176361084, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7083334922790527, + "num_tokens": 449894493.0, + "step": 17535 + }, + { + "epoch": 1.925763233033165, + "grad_norm": 2.1683881282806396, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7073085308074951, + "num_tokens": 449921809.0, + "step": 17536 + }, + { + "epoch": 1.9258730507357786, + "grad_norm": 2.1738038063049316, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.721701979637146, + "num_tokens": 449949620.0, + "step": 17537 + }, + { + "epoch": 1.9259828684383922, + "grad_norm": 2.1130287647247314, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7011048793792725, + "num_tokens": 449979377.0, + "step": 17538 + }, + { + "epoch": 1.926092686141006, + "grad_norm": 2.166402816772461, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7355563640594482, + "num_tokens": 450005549.0, + "step": 17539 + }, + { + "epoch": 1.9262025038436197, + "grad_norm": 2.2339565753936768, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7289583086967468, + "num_tokens": 450032044.0, + "step": 17540 + }, + { + "epoch": 1.9263123215462332, + "grad_norm": 2.225524425506592, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7209875583648682, + "num_tokens": 450057527.0, + "step": 17541 + }, + { + "epoch": 1.9264221392488468, + "grad_norm": 2.358668327331543, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7396401166915894, + "num_tokens": 450083028.0, + "step": 17542 + }, + { + "epoch": 1.9265319569514605, + "grad_norm": 2.172503709793091, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.6999326944351196, + "num_tokens": 450113606.0, + "step": 17543 + }, + { + "epoch": 1.9266417746540743, + "grad_norm": 2.462737560272217, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7139362096786499, + "num_tokens": 450135655.0, + "step": 17544 + }, + { + "epoch": 1.926751592356688, + "grad_norm": 2.432920455932617, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7346194982528687, + "num_tokens": 450157059.0, + "step": 17545 + }, + { + "epoch": 1.9268614100593016, + "grad_norm": 2.0437941551208496, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7130138874053955, + "num_tokens": 450186926.0, + "step": 17546 + }, + { + "epoch": 1.926971227761915, + "grad_norm": 2.369184732437134, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7402299046516418, + "num_tokens": 450210092.0, + "step": 17547 + }, + { + "epoch": 1.9270810454645289, + "grad_norm": 2.057427167892456, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.691891610622406, + "num_tokens": 450240722.0, + "step": 17548 + }, + { + "epoch": 1.9271908631671426, + "grad_norm": 2.281679391860962, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7020238041877747, + "num_tokens": 450267168.0, + "step": 17549 + }, + { + "epoch": 1.9273006808697564, + "grad_norm": 2.2716643810272217, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7184394598007202, + "num_tokens": 450292658.0, + "step": 17550 + }, + { + "epoch": 1.92741049857237, + "grad_norm": 2.0163443088531494, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7236549854278564, + "num_tokens": 450323987.0, + "step": 17551 + }, + { + "epoch": 1.9275203162749834, + "grad_norm": 2.196810245513916, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6955664157867432, + "num_tokens": 450351169.0, + "step": 17552 + }, + { + "epoch": 1.9276301339775972, + "grad_norm": 2.2435996532440186, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7184383869171143, + "num_tokens": 450377074.0, + "step": 17553 + }, + { + "epoch": 1.927739951680211, + "grad_norm": 2.365346670150757, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7289140224456787, + "num_tokens": 450398740.0, + "step": 17554 + }, + { + "epoch": 1.9278497693828245, + "grad_norm": 2.068545341491699, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6991179585456848, + "num_tokens": 450426477.0, + "step": 17555 + }, + { + "epoch": 1.927959587085438, + "grad_norm": 2.347804546356201, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7188221216201782, + "num_tokens": 450450993.0, + "step": 17556 + }, + { + "epoch": 1.9280694047880518, + "grad_norm": 2.158561944961548, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7070882320404053, + "num_tokens": 450478009.0, + "step": 17557 + }, + { + "epoch": 1.9281792224906655, + "grad_norm": 2.254683494567871, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7292062044143677, + "num_tokens": 450504241.0, + "step": 17558 + }, + { + "epoch": 1.9282890401932793, + "grad_norm": 2.1998291015625, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7094763517379761, + "num_tokens": 450531200.0, + "step": 17559 + }, + { + "epoch": 1.9283988578958928, + "grad_norm": 2.5956270694732666, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7382689714431763, + "num_tokens": 450550638.0, + "step": 17560 + }, + { + "epoch": 1.9285086755985064, + "grad_norm": 2.0529184341430664, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.69922935962677, + "num_tokens": 450582646.0, + "step": 17561 + }, + { + "epoch": 1.9286184933011201, + "grad_norm": 2.023994207382202, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6892791986465454, + "num_tokens": 450612430.0, + "step": 17562 + }, + { + "epoch": 1.9287283110037339, + "grad_norm": 2.243164539337158, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7109026908874512, + "num_tokens": 450639832.0, + "step": 17563 + }, + { + "epoch": 1.9288381287063474, + "grad_norm": 2.4923553466796875, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7225476503372192, + "num_tokens": 450663903.0, + "step": 17564 + }, + { + "epoch": 1.9289479464089612, + "grad_norm": 2.1054224967956543, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7331399917602539, + "num_tokens": 450691204.0, + "step": 17565 + }, + { + "epoch": 1.9290577641115747, + "grad_norm": 2.233761787414551, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7157219648361206, + "num_tokens": 450714551.0, + "step": 17566 + }, + { + "epoch": 1.9291675818141885, + "grad_norm": 2.152289628982544, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7024356722831726, + "num_tokens": 450742654.0, + "step": 17567 + }, + { + "epoch": 1.9292773995168022, + "grad_norm": 2.7135202884674072, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7178812026977539, + "num_tokens": 450763845.0, + "step": 17568 + }, + { + "epoch": 1.9293872172194158, + "grad_norm": 2.377098798751831, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7190109491348267, + "num_tokens": 450785400.0, + "step": 17569 + }, + { + "epoch": 1.9294970349220293, + "grad_norm": 2.341688394546509, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.705615758895874, + "num_tokens": 450810653.0, + "step": 17570 + }, + { + "epoch": 1.929606852624643, + "grad_norm": 2.3450422286987305, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.724064826965332, + "num_tokens": 450835015.0, + "step": 17571 + }, + { + "epoch": 1.9297166703272568, + "grad_norm": 2.177110433578491, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7009732723236084, + "num_tokens": 450863701.0, + "step": 17572 + }, + { + "epoch": 1.9298264880298706, + "grad_norm": 2.412264347076416, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.73015958070755, + "num_tokens": 450885802.0, + "step": 17573 + }, + { + "epoch": 1.929936305732484, + "grad_norm": 2.658475160598755, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7430651187896729, + "num_tokens": 450905706.0, + "step": 17574 + }, + { + "epoch": 1.9300461234350976, + "grad_norm": 2.406585454940796, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.703091025352478, + "num_tokens": 450928982.0, + "step": 17575 + }, + { + "epoch": 1.9301559411377114, + "grad_norm": 2.4580085277557373, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7271367311477661, + "num_tokens": 450949873.0, + "step": 17576 + }, + { + "epoch": 1.9302657588403251, + "grad_norm": 2.632936716079712, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7240523099899292, + "num_tokens": 450970757.0, + "step": 17577 + }, + { + "epoch": 1.9303755765429387, + "grad_norm": 2.2494356632232666, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7330774068832397, + "num_tokens": 450996812.0, + "step": 17578 + }, + { + "epoch": 1.9304853942455524, + "grad_norm": 2.1248793601989746, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7163623571395874, + "num_tokens": 451023211.0, + "step": 17579 + }, + { + "epoch": 1.930595211948166, + "grad_norm": 2.46701717376709, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7169279456138611, + "num_tokens": 451045108.0, + "step": 17580 + }, + { + "epoch": 1.9307050296507797, + "grad_norm": 2.181414842605591, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7060719728469849, + "num_tokens": 451073418.0, + "step": 17581 + }, + { + "epoch": 1.9308148473533935, + "grad_norm": 2.3327202796936035, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7279617786407471, + "num_tokens": 451097586.0, + "step": 17582 + }, + { + "epoch": 1.930924665056007, + "grad_norm": 2.336308717727661, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7024071216583252, + "num_tokens": 451122304.0, + "step": 17583 + }, + { + "epoch": 1.9310344827586206, + "grad_norm": 2.5872912406921387, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7247896194458008, + "num_tokens": 451142545.0, + "step": 17584 + }, + { + "epoch": 1.9311443004612343, + "grad_norm": 2.323709726333618, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.687768816947937, + "num_tokens": 451167815.0, + "step": 17585 + }, + { + "epoch": 1.931254118163848, + "grad_norm": 2.2001028060913086, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7413972616195679, + "num_tokens": 451193922.0, + "step": 17586 + }, + { + "epoch": 1.9313639358664618, + "grad_norm": 2.15124249458313, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.722087025642395, + "num_tokens": 451218912.0, + "step": 17587 + }, + { + "epoch": 1.9314737535690754, + "grad_norm": 2.145030975341797, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7134758234024048, + "num_tokens": 451247665.0, + "step": 17588 + }, + { + "epoch": 1.931583571271689, + "grad_norm": 2.5535051822662354, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7319005727767944, + "num_tokens": 451268250.0, + "step": 17589 + }, + { + "epoch": 1.9316933889743026, + "grad_norm": 2.3260602951049805, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7258081436157227, + "num_tokens": 451293073.0, + "step": 17590 + }, + { + "epoch": 1.9318032066769164, + "grad_norm": 2.273757219314575, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7139600515365601, + "num_tokens": 451318933.0, + "step": 17591 + }, + { + "epoch": 1.93191302437953, + "grad_norm": 2.0355403423309326, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7254806756973267, + "num_tokens": 451349627.0, + "step": 17592 + }, + { + "epoch": 1.9320228420821435, + "grad_norm": 2.283926010131836, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7238549590110779, + "num_tokens": 451375960.0, + "step": 17593 + }, + { + "epoch": 1.9321326597847572, + "grad_norm": 2.599952459335327, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7240551114082336, + "num_tokens": 451396094.0, + "step": 17594 + }, + { + "epoch": 1.932242477487371, + "grad_norm": 2.1559336185455322, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.716550350189209, + "num_tokens": 451422151.0, + "step": 17595 + }, + { + "epoch": 1.9323522951899847, + "grad_norm": 2.009242296218872, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7449458837509155, + "num_tokens": 451451542.0, + "step": 17596 + }, + { + "epoch": 1.9324621128925983, + "grad_norm": 2.2227962017059326, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7356688976287842, + "num_tokens": 451478400.0, + "step": 17597 + }, + { + "epoch": 1.9325719305952118, + "grad_norm": 2.3348519802093506, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7494232654571533, + "num_tokens": 451500323.0, + "step": 17598 + }, + { + "epoch": 1.9326817482978256, + "grad_norm": 1.9609370231628418, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.714055061340332, + "num_tokens": 451533229.0, + "step": 17599 + }, + { + "epoch": 1.9327915660004393, + "grad_norm": 2.0252652168273926, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7187618017196655, + "num_tokens": 451562141.0, + "step": 17600 + }, + { + "epoch": 1.932901383703053, + "grad_norm": 2.4369006156921387, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7304612398147583, + "num_tokens": 451584253.0, + "step": 17601 + }, + { + "epoch": 1.9330112014056666, + "grad_norm": 2.240241050720215, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7289210557937622, + "num_tokens": 451609675.0, + "step": 17602 + }, + { + "epoch": 1.9331210191082802, + "grad_norm": 2.14481782913208, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7337601184844971, + "num_tokens": 451638152.0, + "step": 17603 + }, + { + "epoch": 1.933230836810894, + "grad_norm": 2.4249744415283203, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7145242691040039, + "num_tokens": 451662186.0, + "step": 17604 + }, + { + "epoch": 1.9333406545135077, + "grad_norm": 2.0105526447296143, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7053729891777039, + "num_tokens": 451694964.0, + "step": 17605 + }, + { + "epoch": 1.9334504722161212, + "grad_norm": 2.360551595687866, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7058213949203491, + "num_tokens": 451719162.0, + "step": 17606 + }, + { + "epoch": 1.9335602899187347, + "grad_norm": 2.218906879425049, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7012919187545776, + "num_tokens": 451747408.0, + "step": 17607 + }, + { + "epoch": 1.9336701076213485, + "grad_norm": 2.5641183853149414, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7326537370681763, + "num_tokens": 451768392.0, + "step": 17608 + }, + { + "epoch": 1.9337799253239623, + "grad_norm": 2.1816694736480713, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7151497602462769, + "num_tokens": 451795615.0, + "step": 17609 + }, + { + "epoch": 1.933889743026576, + "grad_norm": 2.139829635620117, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7096297740936279, + "num_tokens": 451825518.0, + "step": 17610 + }, + { + "epoch": 1.9339995607291895, + "grad_norm": 2.2578978538513184, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7096128463745117, + "num_tokens": 451852650.0, + "step": 17611 + }, + { + "epoch": 1.934109378431803, + "grad_norm": 2.375035285949707, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7268859148025513, + "num_tokens": 451875011.0, + "step": 17612 + }, + { + "epoch": 1.9342191961344168, + "grad_norm": 2.330672025680542, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7305124998092651, + "num_tokens": 451897140.0, + "step": 17613 + }, + { + "epoch": 1.9343290138370306, + "grad_norm": 2.3008973598480225, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7374815940856934, + "num_tokens": 451920296.0, + "step": 17614 + }, + { + "epoch": 1.9344388315396444, + "grad_norm": 2.160698652267456, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7483802437782288, + "num_tokens": 451945362.0, + "step": 17615 + }, + { + "epoch": 1.934548649242258, + "grad_norm": 2.409212350845337, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7410680055618286, + "num_tokens": 451967480.0, + "step": 17616 + }, + { + "epoch": 1.9346584669448714, + "grad_norm": 2.0935347080230713, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7109808325767517, + "num_tokens": 451995533.0, + "step": 17617 + }, + { + "epoch": 1.9347682846474852, + "grad_norm": 2.2463719844818115, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7119675874710083, + "num_tokens": 452022451.0, + "step": 17618 + }, + { + "epoch": 1.934878102350099, + "grad_norm": 2.235930919647217, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7182360887527466, + "num_tokens": 452049202.0, + "step": 17619 + }, + { + "epoch": 1.9349879200527125, + "grad_norm": 2.468590497970581, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7311648726463318, + "num_tokens": 452070195.0, + "step": 17620 + }, + { + "epoch": 1.935097737755326, + "grad_norm": 2.307152271270752, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7112598419189453, + "num_tokens": 452096768.0, + "step": 17621 + }, + { + "epoch": 1.9352075554579398, + "grad_norm": 2.149954319000244, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.7027040719985962, + "num_tokens": 452125179.0, + "step": 17622 + }, + { + "epoch": 1.9353173731605535, + "grad_norm": 2.4621317386627197, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7222474813461304, + "num_tokens": 452145836.0, + "step": 17623 + }, + { + "epoch": 1.9354271908631673, + "grad_norm": 2.584148406982422, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7331844568252563, + "num_tokens": 452165619.0, + "step": 17624 + }, + { + "epoch": 1.9355370085657808, + "grad_norm": 2.15317440032959, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7109746932983398, + "num_tokens": 452192583.0, + "step": 17625 + }, + { + "epoch": 1.9356468262683943, + "grad_norm": 2.26786470413208, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7355850338935852, + "num_tokens": 452216758.0, + "step": 17626 + }, + { + "epoch": 1.935756643971008, + "grad_norm": 2.259185552597046, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6971738338470459, + "num_tokens": 452242971.0, + "step": 17627 + }, + { + "epoch": 1.9358664616736219, + "grad_norm": 2.1476612091064453, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7145082354545593, + "num_tokens": 452269944.0, + "step": 17628 + }, + { + "epoch": 1.9359762793762354, + "grad_norm": 2.1495158672332764, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.729070246219635, + "num_tokens": 452295592.0, + "step": 17629 + }, + { + "epoch": 1.9360860970788492, + "grad_norm": 2.3260529041290283, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7334238290786743, + "num_tokens": 452320335.0, + "step": 17630 + }, + { + "epoch": 1.9361959147814627, + "grad_norm": 2.003335475921631, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.699798583984375, + "num_tokens": 452352235.0, + "step": 17631 + }, + { + "epoch": 1.9363057324840764, + "grad_norm": 2.4337403774261475, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.712774395942688, + "num_tokens": 452375795.0, + "step": 17632 + }, + { + "epoch": 1.9364155501866902, + "grad_norm": 2.1975371837615967, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7530372738838196, + "num_tokens": 452398520.0, + "step": 17633 + }, + { + "epoch": 1.9365253678893037, + "grad_norm": 2.0265719890594482, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7249724864959717, + "num_tokens": 452431321.0, + "step": 17634 + }, + { + "epoch": 1.9366351855919173, + "grad_norm": 1.869105577468872, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7129546999931335, + "num_tokens": 452463062.0, + "step": 17635 + }, + { + "epoch": 1.936745003294531, + "grad_norm": 2.1139612197875977, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.716869592666626, + "num_tokens": 452489461.0, + "step": 17636 + }, + { + "epoch": 1.9368548209971448, + "grad_norm": 2.343109130859375, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7178419232368469, + "num_tokens": 452511775.0, + "step": 17637 + }, + { + "epoch": 1.9369646386997585, + "grad_norm": 2.561819314956665, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7463952302932739, + "num_tokens": 452530828.0, + "step": 17638 + }, + { + "epoch": 1.937074456402372, + "grad_norm": 2.056800603866577, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7380161881446838, + "num_tokens": 452556819.0, + "step": 17639 + }, + { + "epoch": 1.9371842741049856, + "grad_norm": 2.37441349029541, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.701985239982605, + "num_tokens": 452582139.0, + "step": 17640 + }, + { + "epoch": 1.9372940918075994, + "grad_norm": 2.2404825687408447, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6982351541519165, + "num_tokens": 452610206.0, + "step": 17641 + }, + { + "epoch": 1.9374039095102131, + "grad_norm": 2.1684157848358154, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.70725417137146, + "num_tokens": 452636203.0, + "step": 17642 + }, + { + "epoch": 1.9375137272128267, + "grad_norm": 2.338662624359131, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7018001079559326, + "num_tokens": 452660734.0, + "step": 17643 + }, + { + "epoch": 1.9376235449154404, + "grad_norm": 2.185230255126953, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7154129147529602, + "num_tokens": 452688141.0, + "step": 17644 + }, + { + "epoch": 1.937733362618054, + "grad_norm": 2.2283308506011963, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7302855253219604, + "num_tokens": 452712747.0, + "step": 17645 + }, + { + "epoch": 1.9378431803206677, + "grad_norm": 1.873305082321167, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7130205631256104, + "num_tokens": 452745853.0, + "step": 17646 + }, + { + "epoch": 1.9379529980232815, + "grad_norm": 2.1603353023529053, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7309151291847229, + "num_tokens": 452772703.0, + "step": 17647 + }, + { + "epoch": 1.938062815725895, + "grad_norm": 2.152341842651367, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7218762636184692, + "num_tokens": 452800795.0, + "step": 17648 + }, + { + "epoch": 1.9381726334285085, + "grad_norm": 2.2922286987304688, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7204344272613525, + "num_tokens": 452824101.0, + "step": 17649 + }, + { + "epoch": 1.9382824511311223, + "grad_norm": 2.2634341716766357, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7367808818817139, + "num_tokens": 452848148.0, + "step": 17650 + }, + { + "epoch": 1.938392268833736, + "grad_norm": 2.3286404609680176, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7209168076515198, + "num_tokens": 452870514.0, + "step": 17651 + }, + { + "epoch": 1.9385020865363498, + "grad_norm": 2.505009889602661, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7253212928771973, + "num_tokens": 452893818.0, + "step": 17652 + }, + { + "epoch": 1.9386119042389633, + "grad_norm": 2.388540506362915, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7163841724395752, + "num_tokens": 452916963.0, + "step": 17653 + }, + { + "epoch": 1.9387217219415769, + "grad_norm": 2.217430591583252, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.734320878982544, + "num_tokens": 452942703.0, + "step": 17654 + }, + { + "epoch": 1.9388315396441906, + "grad_norm": 2.1782219409942627, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7031177282333374, + "num_tokens": 452971059.0, + "step": 17655 + }, + { + "epoch": 1.9389413573468044, + "grad_norm": 2.580909013748169, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7393761873245239, + "num_tokens": 452991123.0, + "step": 17656 + }, + { + "epoch": 1.939051175049418, + "grad_norm": 2.306478261947632, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7013893127441406, + "num_tokens": 453015479.0, + "step": 17657 + }, + { + "epoch": 1.9391609927520315, + "grad_norm": 2.5037217140197754, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7102724313735962, + "num_tokens": 453036823.0, + "step": 17658 + }, + { + "epoch": 1.9392708104546452, + "grad_norm": 1.8275048732757568, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7230181694030762, + "num_tokens": 453068786.0, + "step": 17659 + }, + { + "epoch": 1.939380628157259, + "grad_norm": 2.0769436359405518, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.7025893926620483, + "num_tokens": 453099358.0, + "step": 17660 + }, + { + "epoch": 1.9394904458598727, + "grad_norm": 2.3341307640075684, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7423121929168701, + "num_tokens": 453122093.0, + "step": 17661 + }, + { + "epoch": 1.9396002635624863, + "grad_norm": 2.4522294998168945, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7238954305648804, + "num_tokens": 453143432.0, + "step": 17662 + }, + { + "epoch": 1.9397100812650998, + "grad_norm": 2.4478535652160645, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7622225880622864, + "num_tokens": 453163517.0, + "step": 17663 + }, + { + "epoch": 1.9398198989677136, + "grad_norm": 2.4383599758148193, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7565129995346069, + "num_tokens": 453182876.0, + "step": 17664 + }, + { + "epoch": 1.9399297166703273, + "grad_norm": 2.3361856937408447, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7036399245262146, + "num_tokens": 453206886.0, + "step": 17665 + }, + { + "epoch": 1.940039534372941, + "grad_norm": 2.0734665393829346, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7359665036201477, + "num_tokens": 453234759.0, + "step": 17666 + }, + { + "epoch": 1.9401493520755546, + "grad_norm": 2.1811413764953613, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7111330032348633, + "num_tokens": 453262311.0, + "step": 17667 + }, + { + "epoch": 1.9402591697781681, + "grad_norm": 2.455371379852295, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.723169207572937, + "num_tokens": 453284755.0, + "step": 17668 + }, + { + "epoch": 1.940368987480782, + "grad_norm": 2.4609711170196533, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7422382831573486, + "num_tokens": 453306934.0, + "step": 17669 + }, + { + "epoch": 1.9404788051833957, + "grad_norm": 2.5957343578338623, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7256242036819458, + "num_tokens": 453327165.0, + "step": 17670 + }, + { + "epoch": 1.9405886228860092, + "grad_norm": 2.15309476852417, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7062822580337524, + "num_tokens": 453353990.0, + "step": 17671 + }, + { + "epoch": 1.9406984405886227, + "grad_norm": 2.2370150089263916, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7198065519332886, + "num_tokens": 453377986.0, + "step": 17672 + }, + { + "epoch": 1.9408082582912365, + "grad_norm": 2.1800267696380615, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7041898965835571, + "num_tokens": 453404131.0, + "step": 17673 + }, + { + "epoch": 1.9409180759938502, + "grad_norm": 1.9923492670059204, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7055754661560059, + "num_tokens": 453435829.0, + "step": 17674 + }, + { + "epoch": 1.941027893696464, + "grad_norm": 2.1220006942749023, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7050527334213257, + "num_tokens": 453464020.0, + "step": 17675 + }, + { + "epoch": 1.9411377113990775, + "grad_norm": 2.3640637397766113, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7279631495475769, + "num_tokens": 453485656.0, + "step": 17676 + }, + { + "epoch": 1.941247529101691, + "grad_norm": 2.2081692218780518, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7238730192184448, + "num_tokens": 453512026.0, + "step": 17677 + }, + { + "epoch": 1.9413573468043048, + "grad_norm": 2.1390976905822754, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6931110620498657, + "num_tokens": 453541507.0, + "step": 17678 + }, + { + "epoch": 1.9414671645069186, + "grad_norm": 2.446826219558716, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7566416263580322, + "num_tokens": 453560623.0, + "step": 17679 + }, + { + "epoch": 1.9415769822095323, + "grad_norm": 2.1148874759674072, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7267266511917114, + "num_tokens": 453587192.0, + "step": 17680 + }, + { + "epoch": 1.9416867999121459, + "grad_norm": 2.455702543258667, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7250696420669556, + "num_tokens": 453608865.0, + "step": 17681 + }, + { + "epoch": 1.9417966176147594, + "grad_norm": 2.409433126449585, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7321960926055908, + "num_tokens": 453629460.0, + "step": 17682 + }, + { + "epoch": 1.9419064353173732, + "grad_norm": 1.9958274364471436, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7161583304405212, + "num_tokens": 453659214.0, + "step": 17683 + }, + { + "epoch": 1.942016253019987, + "grad_norm": 2.3916831016540527, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7315407395362854, + "num_tokens": 453681860.0, + "step": 17684 + }, + { + "epoch": 1.9421260707226005, + "grad_norm": 2.1975879669189453, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7509871125221252, + "num_tokens": 453706872.0, + "step": 17685 + }, + { + "epoch": 1.942235888425214, + "grad_norm": 2.2107014656066895, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7316091060638428, + "num_tokens": 453735131.0, + "step": 17686 + }, + { + "epoch": 1.9423457061278278, + "grad_norm": 2.449066162109375, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7184654474258423, + "num_tokens": 453758466.0, + "step": 17687 + }, + { + "epoch": 1.9424555238304415, + "grad_norm": 2.035731554031372, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7097147107124329, + "num_tokens": 453788035.0, + "step": 17688 + }, + { + "epoch": 1.9425653415330553, + "grad_norm": 2.6483500003814697, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7297878861427307, + "num_tokens": 453807605.0, + "step": 17689 + }, + { + "epoch": 1.9426751592356688, + "grad_norm": 2.071594476699829, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7244898080825806, + "num_tokens": 453837604.0, + "step": 17690 + }, + { + "epoch": 1.9427849769382823, + "grad_norm": 2.4053657054901123, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7344439625740051, + "num_tokens": 453860188.0, + "step": 17691 + }, + { + "epoch": 1.942894794640896, + "grad_norm": 2.021489381790161, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7214505672454834, + "num_tokens": 453891564.0, + "step": 17692 + }, + { + "epoch": 1.9430046123435099, + "grad_norm": 2.0951926708221436, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7095744609832764, + "num_tokens": 453920848.0, + "step": 17693 + }, + { + "epoch": 1.9431144300461234, + "grad_norm": 1.925885558128357, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7276743650436401, + "num_tokens": 453952747.0, + "step": 17694 + }, + { + "epoch": 1.9432242477487371, + "grad_norm": 2.3716623783111572, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7327730655670166, + "num_tokens": 453973732.0, + "step": 17695 + }, + { + "epoch": 1.9433340654513507, + "grad_norm": 2.739793539047241, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7184671759605408, + "num_tokens": 453991835.0, + "step": 17696 + }, + { + "epoch": 1.9434438831539644, + "grad_norm": 2.3369925022125244, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7230229377746582, + "num_tokens": 454016212.0, + "step": 17697 + }, + { + "epoch": 1.9435537008565782, + "grad_norm": 2.2962229251861572, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7144747972488403, + "num_tokens": 454042241.0, + "step": 17698 + }, + { + "epoch": 1.9436635185591917, + "grad_norm": 2.1724019050598145, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7417789697647095, + "num_tokens": 454065785.0, + "step": 17699 + }, + { + "epoch": 1.9437733362618053, + "grad_norm": 2.068643093109131, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7039252519607544, + "num_tokens": 454096014.0, + "step": 17700 + }, + { + "epoch": 1.943883153964419, + "grad_norm": 2.353835344314575, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7191907167434692, + "num_tokens": 454118691.0, + "step": 17701 + }, + { + "epoch": 1.9439929716670328, + "grad_norm": 2.0737242698669434, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7497965097427368, + "num_tokens": 454145082.0, + "step": 17702 + }, + { + "epoch": 1.9441027893696465, + "grad_norm": 2.087249279022217, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7087191343307495, + "num_tokens": 454173552.0, + "step": 17703 + }, + { + "epoch": 1.94421260707226, + "grad_norm": 2.2920541763305664, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7075778841972351, + "num_tokens": 454200238.0, + "step": 17704 + }, + { + "epoch": 1.9443224247748736, + "grad_norm": 2.1763839721679688, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7244041562080383, + "num_tokens": 454225911.0, + "step": 17705 + }, + { + "epoch": 1.9444322424774874, + "grad_norm": 2.5799262523651123, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7336133122444153, + "num_tokens": 454244743.0, + "step": 17706 + }, + { + "epoch": 1.9445420601801011, + "grad_norm": 2.6124603748321533, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7319971323013306, + "num_tokens": 454265042.0, + "step": 17707 + }, + { + "epoch": 1.9446518778827147, + "grad_norm": 2.142526149749756, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7024053335189819, + "num_tokens": 454295268.0, + "step": 17708 + }, + { + "epoch": 1.9447616955853284, + "grad_norm": 2.2467901706695557, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7357693910598755, + "num_tokens": 454319241.0, + "step": 17709 + }, + { + "epoch": 1.944871513287942, + "grad_norm": 2.190401315689087, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7109477519989014, + "num_tokens": 454345545.0, + "step": 17710 + }, + { + "epoch": 1.9449813309905557, + "grad_norm": 2.262566328048706, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7199863791465759, + "num_tokens": 454370207.0, + "step": 17711 + }, + { + "epoch": 1.9450911486931695, + "grad_norm": 2.2952616214752197, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7193675637245178, + "num_tokens": 454394992.0, + "step": 17712 + }, + { + "epoch": 1.945200966395783, + "grad_norm": 2.2305328845977783, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7015271186828613, + "num_tokens": 454421615.0, + "step": 17713 + }, + { + "epoch": 1.9453107840983965, + "grad_norm": 1.924839973449707, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7153429985046387, + "num_tokens": 454453184.0, + "step": 17714 + }, + { + "epoch": 1.9454206018010103, + "grad_norm": 2.286452531814575, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7139908075332642, + "num_tokens": 454476663.0, + "step": 17715 + }, + { + "epoch": 1.945530419503624, + "grad_norm": 2.4379842281341553, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7236035466194153, + "num_tokens": 454499211.0, + "step": 17716 + }, + { + "epoch": 1.9456402372062378, + "grad_norm": 2.0129640102386475, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7069426774978638, + "num_tokens": 454530142.0, + "step": 17717 + }, + { + "epoch": 1.9457500549088513, + "grad_norm": 2.5101664066314697, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7256403565406799, + "num_tokens": 454551351.0, + "step": 17718 + }, + { + "epoch": 1.9458598726114649, + "grad_norm": 2.0680596828460693, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7117994427680969, + "num_tokens": 454580494.0, + "step": 17719 + }, + { + "epoch": 1.9459696903140786, + "grad_norm": 2.5769002437591553, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7252167463302612, + "num_tokens": 454600136.0, + "step": 17720 + }, + { + "epoch": 1.9460795080166924, + "grad_norm": 2.2725255489349365, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7076340913772583, + "num_tokens": 454626427.0, + "step": 17721 + }, + { + "epoch": 1.946189325719306, + "grad_norm": 1.9546732902526855, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7201234102249146, + "num_tokens": 454656933.0, + "step": 17722 + }, + { + "epoch": 1.9462991434219195, + "grad_norm": 2.7308576107025146, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7341666221618652, + "num_tokens": 454675095.0, + "step": 17723 + }, + { + "epoch": 1.9464089611245332, + "grad_norm": 2.067620277404785, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.743152379989624, + "num_tokens": 454701447.0, + "step": 17724 + }, + { + "epoch": 1.946518778827147, + "grad_norm": 2.4912455081939697, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7378191947937012, + "num_tokens": 454722754.0, + "step": 17725 + }, + { + "epoch": 1.9466285965297607, + "grad_norm": 2.392897844314575, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7232039570808411, + "num_tokens": 454745307.0, + "step": 17726 + }, + { + "epoch": 1.9467384142323743, + "grad_norm": 2.1446759700775146, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7429896593093872, + "num_tokens": 454770177.0, + "step": 17727 + }, + { + "epoch": 1.9468482319349878, + "grad_norm": 2.249157667160034, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7197862267494202, + "num_tokens": 454795608.0, + "step": 17728 + }, + { + "epoch": 1.9469580496376016, + "grad_norm": 2.1919894218444824, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7350131273269653, + "num_tokens": 454821214.0, + "step": 17729 + }, + { + "epoch": 1.9470678673402153, + "grad_norm": 2.890825033187866, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7313307523727417, + "num_tokens": 454838316.0, + "step": 17730 + }, + { + "epoch": 1.947177685042829, + "grad_norm": 2.376605272293091, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.707659125328064, + "num_tokens": 454863565.0, + "step": 17731 + }, + { + "epoch": 1.9472875027454426, + "grad_norm": 2.467594623565674, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.729584813117981, + "num_tokens": 454885232.0, + "step": 17732 + }, + { + "epoch": 1.9473973204480561, + "grad_norm": 2.288386106491089, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7240920066833496, + "num_tokens": 454910611.0, + "step": 17733 + }, + { + "epoch": 1.94750713815067, + "grad_norm": 2.415724992752075, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7694523334503174, + "num_tokens": 454931941.0, + "step": 17734 + }, + { + "epoch": 1.9476169558532836, + "grad_norm": 2.5853993892669678, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7615064382553101, + "num_tokens": 454949906.0, + "step": 17735 + }, + { + "epoch": 1.9477267735558972, + "grad_norm": 2.2206737995147705, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7131140232086182, + "num_tokens": 454974710.0, + "step": 17736 + }, + { + "epoch": 1.9478365912585107, + "grad_norm": 2.5446972846984863, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7221930027008057, + "num_tokens": 454996118.0, + "step": 17737 + }, + { + "epoch": 1.9479464089611245, + "grad_norm": 1.980626106262207, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7055110335350037, + "num_tokens": 455026226.0, + "step": 17738 + }, + { + "epoch": 1.9480562266637382, + "grad_norm": 2.3054873943328857, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7240447402000427, + "num_tokens": 455051196.0, + "step": 17739 + }, + { + "epoch": 1.948166044366352, + "grad_norm": 2.389847755432129, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7110015153884888, + "num_tokens": 455074489.0, + "step": 17740 + }, + { + "epoch": 1.9482758620689655, + "grad_norm": 2.157794237136841, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7332309484481812, + "num_tokens": 455099278.0, + "step": 17741 + }, + { + "epoch": 1.948385679771579, + "grad_norm": 2.161435842514038, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7181886434555054, + "num_tokens": 455125166.0, + "step": 17742 + }, + { + "epoch": 1.9484954974741928, + "grad_norm": 2.4432387351989746, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7410805225372314, + "num_tokens": 455147303.0, + "step": 17743 + }, + { + "epoch": 1.9486053151768066, + "grad_norm": 2.192793369293213, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7356952428817749, + "num_tokens": 455175250.0, + "step": 17744 + }, + { + "epoch": 1.94871513287942, + "grad_norm": 2.1901495456695557, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.737296998500824, + "num_tokens": 455199354.0, + "step": 17745 + }, + { + "epoch": 1.9488249505820339, + "grad_norm": 2.1193504333496094, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7294049263000488, + "num_tokens": 455226643.0, + "step": 17746 + }, + { + "epoch": 1.9489347682846474, + "grad_norm": 2.0965640544891357, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7293378114700317, + "num_tokens": 455254607.0, + "step": 17747 + }, + { + "epoch": 1.9490445859872612, + "grad_norm": 2.347416877746582, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7231194972991943, + "num_tokens": 455278430.0, + "step": 17748 + }, + { + "epoch": 1.949154403689875, + "grad_norm": 2.450594663619995, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7288256883621216, + "num_tokens": 455300166.0, + "step": 17749 + }, + { + "epoch": 1.9492642213924884, + "grad_norm": 1.9429398775100708, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7369537353515625, + "num_tokens": 455332704.0, + "step": 17750 + }, + { + "epoch": 1.949374039095102, + "grad_norm": 2.2166473865509033, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.723622739315033, + "num_tokens": 455360103.0, + "step": 17751 + }, + { + "epoch": 1.9494838567977157, + "grad_norm": 2.571918249130249, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7452938556671143, + "num_tokens": 455379548.0, + "step": 17752 + }, + { + "epoch": 1.9495936745003295, + "grad_norm": 2.1067910194396973, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7078384160995483, + "num_tokens": 455410164.0, + "step": 17753 + }, + { + "epoch": 1.9497034922029433, + "grad_norm": 2.4387385845184326, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7412747144699097, + "num_tokens": 455430282.0, + "step": 17754 + }, + { + "epoch": 1.9498133099055568, + "grad_norm": 2.1637043952941895, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.715400218963623, + "num_tokens": 455455319.0, + "step": 17755 + }, + { + "epoch": 1.9499231276081703, + "grad_norm": 1.8844587802886963, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7074443101882935, + "num_tokens": 455489288.0, + "step": 17756 + }, + { + "epoch": 1.950032945310784, + "grad_norm": 2.5729310512542725, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7442718744277954, + "num_tokens": 455508196.0, + "step": 17757 + }, + { + "epoch": 1.9501427630133978, + "grad_norm": 2.041891574859619, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7247244119644165, + "num_tokens": 455538045.0, + "step": 17758 + }, + { + "epoch": 1.9502525807160114, + "grad_norm": 2.341435432434082, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7084602117538452, + "num_tokens": 455560880.0, + "step": 17759 + }, + { + "epoch": 1.9503623984186251, + "grad_norm": 2.2897675037384033, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.690963625907898, + "num_tokens": 455587031.0, + "step": 17760 + }, + { + "epoch": 1.9504722161212387, + "grad_norm": 2.455043315887451, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7366635799407959, + "num_tokens": 455608499.0, + "step": 17761 + }, + { + "epoch": 1.9505820338238524, + "grad_norm": 2.238480567932129, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7053107023239136, + "num_tokens": 455633381.0, + "step": 17762 + }, + { + "epoch": 1.9506918515264662, + "grad_norm": 1.9404479265213013, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7070077061653137, + "num_tokens": 455665844.0, + "step": 17763 + }, + { + "epoch": 1.9508016692290797, + "grad_norm": 2.160598039627075, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7189759612083435, + "num_tokens": 455693564.0, + "step": 17764 + }, + { + "epoch": 1.9509114869316933, + "grad_norm": 2.084911823272705, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7082020044326782, + "num_tokens": 455721551.0, + "step": 17765 + }, + { + "epoch": 1.951021304634307, + "grad_norm": 2.1882622241973877, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7410790920257568, + "num_tokens": 455747125.0, + "step": 17766 + }, + { + "epoch": 1.9511311223369208, + "grad_norm": 2.380704641342163, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7418226003646851, + "num_tokens": 455770099.0, + "step": 17767 + }, + { + "epoch": 1.9512409400395345, + "grad_norm": 2.411149501800537, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7054651975631714, + "num_tokens": 455795211.0, + "step": 17768 + }, + { + "epoch": 1.951350757742148, + "grad_norm": 2.41272234916687, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7359335422515869, + "num_tokens": 455818095.0, + "step": 17769 + }, + { + "epoch": 1.9514605754447616, + "grad_norm": 2.1836838722229004, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.719021201133728, + "num_tokens": 455845773.0, + "step": 17770 + }, + { + "epoch": 1.9515703931473753, + "grad_norm": 2.3423526287078857, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7229939699172974, + "num_tokens": 455869941.0, + "step": 17771 + }, + { + "epoch": 1.951680210849989, + "grad_norm": 1.9952645301818848, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7235596179962158, + "num_tokens": 455902341.0, + "step": 17772 + }, + { + "epoch": 1.9517900285526026, + "grad_norm": 2.4003303050994873, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7571203708648682, + "num_tokens": 455922236.0, + "step": 17773 + }, + { + "epoch": 1.9518998462552162, + "grad_norm": 2.0104622840881348, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7432460784912109, + "num_tokens": 455950607.0, + "step": 17774 + }, + { + "epoch": 1.95200966395783, + "grad_norm": 2.2257518768310547, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7288901805877686, + "num_tokens": 455975821.0, + "step": 17775 + }, + { + "epoch": 1.9521194816604437, + "grad_norm": 1.9774141311645508, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6890053153038025, + "num_tokens": 456008066.0, + "step": 17776 + }, + { + "epoch": 1.9522292993630574, + "grad_norm": 2.832825183868408, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7407099008560181, + "num_tokens": 456024692.0, + "step": 17777 + }, + { + "epoch": 1.952339117065671, + "grad_norm": 2.0204832553863525, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7406179308891296, + "num_tokens": 456054476.0, + "step": 17778 + }, + { + "epoch": 1.9524489347682845, + "grad_norm": 2.0972564220428467, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.723469614982605, + "num_tokens": 456082045.0, + "step": 17779 + }, + { + "epoch": 1.9525587524708983, + "grad_norm": 2.1653833389282227, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7247997522354126, + "num_tokens": 456107959.0, + "step": 17780 + }, + { + "epoch": 1.952668570173512, + "grad_norm": 2.53273344039917, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7253240346908569, + "num_tokens": 456129128.0, + "step": 17781 + }, + { + "epoch": 1.9527783878761258, + "grad_norm": 2.1891021728515625, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7374287843704224, + "num_tokens": 456155639.0, + "step": 17782 + }, + { + "epoch": 1.9528882055787393, + "grad_norm": 2.61251163482666, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7427474856376648, + "num_tokens": 456176120.0, + "step": 17783 + }, + { + "epoch": 1.9529980232813529, + "grad_norm": 2.0789761543273926, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7149556279182434, + "num_tokens": 456207337.0, + "step": 17784 + }, + { + "epoch": 1.9531078409839666, + "grad_norm": 2.395318031311035, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7467288970947266, + "num_tokens": 456229830.0, + "step": 17785 + }, + { + "epoch": 1.9532176586865804, + "grad_norm": 2.2980058193206787, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7080524563789368, + "num_tokens": 456256036.0, + "step": 17786 + }, + { + "epoch": 1.953327476389194, + "grad_norm": 2.2019951343536377, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7403262853622437, + "num_tokens": 456281926.0, + "step": 17787 + }, + { + "epoch": 1.9534372940918074, + "grad_norm": 1.991294503211975, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.725680947303772, + "num_tokens": 456313501.0, + "step": 17788 + }, + { + "epoch": 1.9535471117944212, + "grad_norm": 2.003154754638672, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7118796110153198, + "num_tokens": 456343707.0, + "step": 17789 + }, + { + "epoch": 1.953656929497035, + "grad_norm": 2.034489154815674, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7048519253730774, + "num_tokens": 456373142.0, + "step": 17790 + }, + { + "epoch": 1.9537667471996487, + "grad_norm": 2.4407575130462646, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7400925159454346, + "num_tokens": 456395715.0, + "step": 17791 + }, + { + "epoch": 1.9538765649022622, + "grad_norm": 2.2126359939575195, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.6997048854827881, + "num_tokens": 456424300.0, + "step": 17792 + }, + { + "epoch": 1.9539863826048758, + "grad_norm": 2.162126302719116, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7219475507736206, + "num_tokens": 456451917.0, + "step": 17793 + }, + { + "epoch": 1.9540962003074895, + "grad_norm": 2.2424190044403076, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7383857369422913, + "num_tokens": 456473775.0, + "step": 17794 + }, + { + "epoch": 1.9542060180101033, + "grad_norm": 2.2950520515441895, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7408865094184875, + "num_tokens": 456498051.0, + "step": 17795 + }, + { + "epoch": 1.954315835712717, + "grad_norm": 1.993179202079773, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7234897613525391, + "num_tokens": 456529677.0, + "step": 17796 + }, + { + "epoch": 1.9544256534153306, + "grad_norm": 2.1544439792633057, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7027773261070251, + "num_tokens": 456558038.0, + "step": 17797 + }, + { + "epoch": 1.9545354711179441, + "grad_norm": 2.458191156387329, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7378367185592651, + "num_tokens": 456580317.0, + "step": 17798 + }, + { + "epoch": 1.9546452888205579, + "grad_norm": 2.283388614654541, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7258268594741821, + "num_tokens": 456604971.0, + "step": 17799 + }, + { + "epoch": 1.9547551065231716, + "grad_norm": 2.0710270404815674, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6905028820037842, + "num_tokens": 456635532.0, + "step": 17800 + }, + { + "epoch": 1.9548649242257852, + "grad_norm": 2.257619619369507, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7122187614440918, + "num_tokens": 456662521.0, + "step": 17801 + }, + { + "epoch": 1.9549747419283987, + "grad_norm": 2.210456609725952, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7077479362487793, + "num_tokens": 456690704.0, + "step": 17802 + }, + { + "epoch": 1.9550845596310125, + "grad_norm": 2.345064878463745, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7387443780899048, + "num_tokens": 456714336.0, + "step": 17803 + }, + { + "epoch": 1.9551943773336262, + "grad_norm": 2.0875275135040283, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.710834264755249, + "num_tokens": 456743829.0, + "step": 17804 + }, + { + "epoch": 1.95530419503624, + "grad_norm": 2.2637763023376465, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7160860896110535, + "num_tokens": 456770999.0, + "step": 17805 + }, + { + "epoch": 1.9554140127388535, + "grad_norm": 2.2685978412628174, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7000362873077393, + "num_tokens": 456797808.0, + "step": 17806 + }, + { + "epoch": 1.955523830441467, + "grad_norm": 2.1236371994018555, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.6987349987030029, + "num_tokens": 456828085.0, + "step": 17807 + }, + { + "epoch": 1.9556336481440808, + "grad_norm": 2.386157989501953, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.724431037902832, + "num_tokens": 456850979.0, + "step": 17808 + }, + { + "epoch": 1.9557434658466946, + "grad_norm": 2.184868335723877, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7293510437011719, + "num_tokens": 456877320.0, + "step": 17809 + }, + { + "epoch": 1.955853283549308, + "grad_norm": 2.2238781452178955, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7068313956260681, + "num_tokens": 456904859.0, + "step": 17810 + }, + { + "epoch": 1.9559631012519219, + "grad_norm": 2.645514726638794, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7232818603515625, + "num_tokens": 456923644.0, + "step": 17811 + }, + { + "epoch": 1.9560729189545354, + "grad_norm": 2.0961108207702637, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7091494798660278, + "num_tokens": 456951950.0, + "step": 17812 + }, + { + "epoch": 1.9561827366571491, + "grad_norm": 2.2167086601257324, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6962470412254333, + "num_tokens": 456980010.0, + "step": 17813 + }, + { + "epoch": 1.956292554359763, + "grad_norm": 2.1673505306243896, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7119628190994263, + "num_tokens": 457005691.0, + "step": 17814 + }, + { + "epoch": 1.9564023720623764, + "grad_norm": 2.3579747676849365, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7232156991958618, + "num_tokens": 457028418.0, + "step": 17815 + }, + { + "epoch": 1.95651218976499, + "grad_norm": 2.2648563385009766, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6981933116912842, + "num_tokens": 457054200.0, + "step": 17816 + }, + { + "epoch": 1.9566220074676037, + "grad_norm": 2.1823976039886475, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7234188318252563, + "num_tokens": 457079527.0, + "step": 17817 + }, + { + "epoch": 1.9567318251702175, + "grad_norm": 2.3468639850616455, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.729884922504425, + "num_tokens": 457103242.0, + "step": 17818 + }, + { + "epoch": 1.9568416428728312, + "grad_norm": 2.241443634033203, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7223730683326721, + "num_tokens": 457127895.0, + "step": 17819 + }, + { + "epoch": 1.9569514605754448, + "grad_norm": 2.0572431087493896, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7013556957244873, + "num_tokens": 457161373.0, + "step": 17820 + }, + { + "epoch": 1.9570612782780583, + "grad_norm": 2.57629656791687, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7519561052322388, + "num_tokens": 457180885.0, + "step": 17821 + }, + { + "epoch": 1.957171095980672, + "grad_norm": 2.344764232635498, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7213523387908936, + "num_tokens": 457203282.0, + "step": 17822 + }, + { + "epoch": 1.9572809136832858, + "grad_norm": 2.5564792156219482, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7488471269607544, + "num_tokens": 457222577.0, + "step": 17823 + }, + { + "epoch": 1.9573907313858994, + "grad_norm": 2.479814291000366, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7449396848678589, + "num_tokens": 457243871.0, + "step": 17824 + }, + { + "epoch": 1.9575005490885131, + "grad_norm": 2.595482587814331, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.720473051071167, + "num_tokens": 457264779.0, + "step": 17825 + }, + { + "epoch": 1.9576103667911267, + "grad_norm": 2.53292179107666, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7421488761901855, + "num_tokens": 457285070.0, + "step": 17826 + }, + { + "epoch": 1.9577201844937404, + "grad_norm": 2.1483967304229736, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.736007034778595, + "num_tokens": 457311072.0, + "step": 17827 + }, + { + "epoch": 1.9578300021963542, + "grad_norm": 2.491244077682495, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7003269195556641, + "num_tokens": 457333723.0, + "step": 17828 + }, + { + "epoch": 1.9579398198989677, + "grad_norm": 2.0693254470825195, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7069291472434998, + "num_tokens": 457362200.0, + "step": 17829 + }, + { + "epoch": 1.9580496376015812, + "grad_norm": 2.446558952331543, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7353049516677856, + "num_tokens": 457384789.0, + "step": 17830 + }, + { + "epoch": 1.958159455304195, + "grad_norm": 2.310044527053833, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7040027379989624, + "num_tokens": 457410896.0, + "step": 17831 + }, + { + "epoch": 1.9582692730068088, + "grad_norm": 2.2405881881713867, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.735846996307373, + "num_tokens": 457434419.0, + "step": 17832 + }, + { + "epoch": 1.9583790907094225, + "grad_norm": 1.9315671920776367, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7205706834793091, + "num_tokens": 457467816.0, + "step": 17833 + }, + { + "epoch": 1.958488908412036, + "grad_norm": 2.1234781742095947, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7329333424568176, + "num_tokens": 457495713.0, + "step": 17834 + }, + { + "epoch": 1.9585987261146496, + "grad_norm": 2.1217665672302246, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7166541814804077, + "num_tokens": 457522976.0, + "step": 17835 + }, + { + "epoch": 1.9587085438172633, + "grad_norm": 2.328369140625, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7255637049674988, + "num_tokens": 457547463.0, + "step": 17836 + }, + { + "epoch": 1.958818361519877, + "grad_norm": 2.0455305576324463, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7124658823013306, + "num_tokens": 457577420.0, + "step": 17837 + }, + { + "epoch": 1.9589281792224906, + "grad_norm": 2.0512313842773438, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7127292156219482, + "num_tokens": 457606979.0, + "step": 17838 + }, + { + "epoch": 1.9590379969251042, + "grad_norm": 2.176692485809326, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7278672456741333, + "num_tokens": 457633636.0, + "step": 17839 + }, + { + "epoch": 1.959147814627718, + "grad_norm": 2.0829224586486816, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7274971604347229, + "num_tokens": 457659483.0, + "step": 17840 + }, + { + "epoch": 1.9592576323303317, + "grad_norm": 2.0664291381835938, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7405475974082947, + "num_tokens": 457687622.0, + "step": 17841 + }, + { + "epoch": 1.9593674500329454, + "grad_norm": 2.1327855587005615, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7106488347053528, + "num_tokens": 457716718.0, + "step": 17842 + }, + { + "epoch": 1.959477267735559, + "grad_norm": 2.27895188331604, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.716769814491272, + "num_tokens": 457741990.0, + "step": 17843 + }, + { + "epoch": 1.9595870854381725, + "grad_norm": 2.2375411987304688, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7102853059768677, + "num_tokens": 457768383.0, + "step": 17844 + }, + { + "epoch": 1.9596969031407863, + "grad_norm": 2.082408905029297, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7102130055427551, + "num_tokens": 457796461.0, + "step": 17845 + }, + { + "epoch": 1.9598067208434, + "grad_norm": 2.326646327972412, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7086537480354309, + "num_tokens": 457820041.0, + "step": 17846 + }, + { + "epoch": 1.9599165385460138, + "grad_norm": 2.633937120437622, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7246894836425781, + "num_tokens": 457839314.0, + "step": 17847 + }, + { + "epoch": 1.9600263562486273, + "grad_norm": 2.4161014556884766, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7122209072113037, + "num_tokens": 457862759.0, + "step": 17848 + }, + { + "epoch": 1.9601361739512408, + "grad_norm": 2.314222574234009, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7290416955947876, + "num_tokens": 457886545.0, + "step": 17849 + }, + { + "epoch": 1.9602459916538546, + "grad_norm": 2.65885591506958, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7349718809127808, + "num_tokens": 457906020.0, + "step": 17850 + }, + { + "epoch": 1.9603558093564684, + "grad_norm": 2.304291248321533, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7257741689682007, + "num_tokens": 457930316.0, + "step": 17851 + }, + { + "epoch": 1.960465627059082, + "grad_norm": 2.0550363063812256, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6860941648483276, + "num_tokens": 457960841.0, + "step": 17852 + }, + { + "epoch": 1.9605754447616954, + "grad_norm": 2.462704658508301, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7128180265426636, + "num_tokens": 457982999.0, + "step": 17853 + }, + { + "epoch": 1.9606852624643092, + "grad_norm": 2.326700210571289, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7140067219734192, + "num_tokens": 458007246.0, + "step": 17854 + }, + { + "epoch": 1.960795080166923, + "grad_norm": 2.2823121547698975, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7155267000198364, + "num_tokens": 458032453.0, + "step": 17855 + }, + { + "epoch": 1.9609048978695367, + "grad_norm": 2.2443337440490723, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7177886962890625, + "num_tokens": 458057714.0, + "step": 17856 + }, + { + "epoch": 1.9610147155721502, + "grad_norm": 2.013211965560913, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7026954889297485, + "num_tokens": 458089193.0, + "step": 17857 + }, + { + "epoch": 1.9611245332747638, + "grad_norm": 2.3144896030426025, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7409699559211731, + "num_tokens": 458115522.0, + "step": 17858 + }, + { + "epoch": 1.9612343509773775, + "grad_norm": 2.2886290550231934, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7115569114685059, + "num_tokens": 458142894.0, + "step": 17859 + }, + { + "epoch": 1.9613441686799913, + "grad_norm": 2.599351167678833, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7210649251937866, + "num_tokens": 458165572.0, + "step": 17860 + }, + { + "epoch": 1.961453986382605, + "grad_norm": 2.189913749694824, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7215654253959656, + "num_tokens": 458192568.0, + "step": 17861 + }, + { + "epoch": 1.9615638040852186, + "grad_norm": 2.118997097015381, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7121111750602722, + "num_tokens": 458220885.0, + "step": 17862 + }, + { + "epoch": 1.9616736217878321, + "grad_norm": 2.2522966861724854, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7186551690101624, + "num_tokens": 458247083.0, + "step": 17863 + }, + { + "epoch": 1.9617834394904459, + "grad_norm": 2.15352463722229, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7031437158584595, + "num_tokens": 458275942.0, + "step": 17864 + }, + { + "epoch": 1.9618932571930596, + "grad_norm": 2.436070442199707, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7073432207107544, + "num_tokens": 458299788.0, + "step": 17865 + }, + { + "epoch": 1.9620030748956732, + "grad_norm": 2.2929372787475586, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7041927576065063, + "num_tokens": 458326246.0, + "step": 17866 + }, + { + "epoch": 1.9621128925982867, + "grad_norm": 2.066389322280884, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7232013940811157, + "num_tokens": 458353310.0, + "step": 17867 + }, + { + "epoch": 1.9622227103009005, + "grad_norm": 2.1525251865386963, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7014485001564026, + "num_tokens": 458380319.0, + "step": 17868 + }, + { + "epoch": 1.9623325280035142, + "grad_norm": 2.5443737506866455, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7439759969711304, + "num_tokens": 458400276.0, + "step": 17869 + }, + { + "epoch": 1.962442345706128, + "grad_norm": 1.8831675052642822, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7168096303939819, + "num_tokens": 458433281.0, + "step": 17870 + }, + { + "epoch": 1.9625521634087415, + "grad_norm": 2.344785690307617, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.724437415599823, + "num_tokens": 458456169.0, + "step": 17871 + }, + { + "epoch": 1.962661981111355, + "grad_norm": 2.1716699600219727, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7288872599601746, + "num_tokens": 458483002.0, + "step": 17872 + }, + { + "epoch": 1.9627717988139688, + "grad_norm": 2.1566686630249023, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7275341749191284, + "num_tokens": 458512171.0, + "step": 17873 + }, + { + "epoch": 1.9628816165165826, + "grad_norm": 2.2554497718811035, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7067794799804688, + "num_tokens": 458538672.0, + "step": 17874 + }, + { + "epoch": 1.962991434219196, + "grad_norm": 2.3232638835906982, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7164285182952881, + "num_tokens": 458564452.0, + "step": 17875 + }, + { + "epoch": 1.9631012519218098, + "grad_norm": 2.9024498462677, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7294906377792358, + "num_tokens": 458582789.0, + "step": 17876 + }, + { + "epoch": 1.9632110696244234, + "grad_norm": 2.57121205329895, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7505884170532227, + "num_tokens": 458603927.0, + "step": 17877 + }, + { + "epoch": 1.9633208873270371, + "grad_norm": 2.4786436557769775, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7045075297355652, + "num_tokens": 458628074.0, + "step": 17878 + }, + { + "epoch": 1.963430705029651, + "grad_norm": 2.670593023300171, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7308295965194702, + "num_tokens": 458646505.0, + "step": 17879 + }, + { + "epoch": 1.9635405227322644, + "grad_norm": 2.4433929920196533, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7204370498657227, + "num_tokens": 458669423.0, + "step": 17880 + }, + { + "epoch": 1.963650340434878, + "grad_norm": 2.352508544921875, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7356793880462646, + "num_tokens": 458693519.0, + "step": 17881 + }, + { + "epoch": 1.9637601581374917, + "grad_norm": 2.1390891075134277, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7218817472457886, + "num_tokens": 458722021.0, + "step": 17882 + }, + { + "epoch": 1.9638699758401055, + "grad_norm": 2.092928409576416, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7261044979095459, + "num_tokens": 458749744.0, + "step": 17883 + }, + { + "epoch": 1.9639797935427192, + "grad_norm": 2.1794681549072266, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7325479984283447, + "num_tokens": 458774657.0, + "step": 17884 + }, + { + "epoch": 1.9640896112453328, + "grad_norm": 2.1419286727905273, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6932184100151062, + "num_tokens": 458802970.0, + "step": 17885 + }, + { + "epoch": 1.9641994289479463, + "grad_norm": 2.512674570083618, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7251777648925781, + "num_tokens": 458823916.0, + "step": 17886 + }, + { + "epoch": 1.96430924665056, + "grad_norm": 2.5457258224487305, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7141709327697754, + "num_tokens": 458845241.0, + "step": 17887 + }, + { + "epoch": 1.9644190643531738, + "grad_norm": 2.497117757797241, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7500699162483215, + "num_tokens": 458865346.0, + "step": 17888 + }, + { + "epoch": 1.9645288820557874, + "grad_norm": 2.225614547729492, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7135115265846252, + "num_tokens": 458890014.0, + "step": 17889 + }, + { + "epoch": 1.964638699758401, + "grad_norm": 2.4877750873565674, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7272634506225586, + "num_tokens": 458910719.0, + "step": 17890 + }, + { + "epoch": 1.9647485174610146, + "grad_norm": 2.380547523498535, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7339221239089966, + "num_tokens": 458935206.0, + "step": 17891 + }, + { + "epoch": 1.9648583351636284, + "grad_norm": 2.3792319297790527, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.722825288772583, + "num_tokens": 458958422.0, + "step": 17892 + }, + { + "epoch": 1.9649681528662422, + "grad_norm": 2.3529772758483887, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7095150947570801, + "num_tokens": 458982552.0, + "step": 17893 + }, + { + "epoch": 1.9650779705688557, + "grad_norm": 2.261998414993286, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7121956944465637, + "num_tokens": 459006366.0, + "step": 17894 + }, + { + "epoch": 1.9651877882714692, + "grad_norm": 2.1946630477905273, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7171468734741211, + "num_tokens": 459031378.0, + "step": 17895 + }, + { + "epoch": 1.965297605974083, + "grad_norm": 2.4170961380004883, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7279340028762817, + "num_tokens": 459054725.0, + "step": 17896 + }, + { + "epoch": 1.9654074236766967, + "grad_norm": 2.357759475708008, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7355807423591614, + "num_tokens": 459079166.0, + "step": 17897 + }, + { + "epoch": 1.9655172413793105, + "grad_norm": 2.1989963054656982, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7148433923721313, + "num_tokens": 459104965.0, + "step": 17898 + }, + { + "epoch": 1.965627059081924, + "grad_norm": 2.4101803302764893, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7466816902160645, + "num_tokens": 459126518.0, + "step": 17899 + }, + { + "epoch": 1.9657368767845376, + "grad_norm": 2.4646553993225098, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7057143449783325, + "num_tokens": 459148680.0, + "step": 17900 + }, + { + "epoch": 1.9658466944871513, + "grad_norm": 2.1845695972442627, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7247936129570007, + "num_tokens": 459175792.0, + "step": 17901 + }, + { + "epoch": 1.965956512189765, + "grad_norm": 2.2135696411132812, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7351140975952148, + "num_tokens": 459201360.0, + "step": 17902 + }, + { + "epoch": 1.9660663298923786, + "grad_norm": 2.261531352996826, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.723375678062439, + "num_tokens": 459226435.0, + "step": 17903 + }, + { + "epoch": 1.9661761475949922, + "grad_norm": 2.5818161964416504, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7137856483459473, + "num_tokens": 459247836.0, + "step": 17904 + }, + { + "epoch": 1.966285965297606, + "grad_norm": 1.8636583089828491, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6894081830978394, + "num_tokens": 459285099.0, + "step": 17905 + }, + { + "epoch": 1.9663957830002197, + "grad_norm": 2.2119855880737305, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7572686672210693, + "num_tokens": 459309285.0, + "step": 17906 + }, + { + "epoch": 1.9665056007028334, + "grad_norm": 2.301666736602783, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7414556741714478, + "num_tokens": 459332319.0, + "step": 17907 + }, + { + "epoch": 1.966615418405447, + "grad_norm": 2.2740018367767334, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7189006805419922, + "num_tokens": 459357034.0, + "step": 17908 + }, + { + "epoch": 1.9667252361080605, + "grad_norm": 2.0749595165252686, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7145357131958008, + "num_tokens": 459384840.0, + "step": 17909 + }, + { + "epoch": 1.9668350538106742, + "grad_norm": 1.9485265016555786, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6857967376708984, + "num_tokens": 459419933.0, + "step": 17910 + }, + { + "epoch": 1.966944871513288, + "grad_norm": 2.164684534072876, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7397653460502625, + "num_tokens": 459446638.0, + "step": 17911 + }, + { + "epoch": 1.9670546892159018, + "grad_norm": 2.218144178390503, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7145369052886963, + "num_tokens": 459472807.0, + "step": 17912 + }, + { + "epoch": 1.9671645069185153, + "grad_norm": 2.0490145683288574, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7634943723678589, + "num_tokens": 459497388.0, + "step": 17913 + }, + { + "epoch": 1.9672743246211288, + "grad_norm": 2.2389745712280273, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7347518801689148, + "num_tokens": 459522604.0, + "step": 17914 + }, + { + "epoch": 1.9673841423237426, + "grad_norm": 2.271944284439087, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7067910432815552, + "num_tokens": 459547182.0, + "step": 17915 + }, + { + "epoch": 1.9674939600263563, + "grad_norm": 2.2058889865875244, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7209148406982422, + "num_tokens": 459573109.0, + "step": 17916 + }, + { + "epoch": 1.9676037777289699, + "grad_norm": 2.176117420196533, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7304083108901978, + "num_tokens": 459600302.0, + "step": 17917 + }, + { + "epoch": 1.9677135954315834, + "grad_norm": 2.148124933242798, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7080076336860657, + "num_tokens": 459628320.0, + "step": 17918 + }, + { + "epoch": 1.9678234131341972, + "grad_norm": 2.24078106880188, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7395144701004028, + "num_tokens": 459651969.0, + "step": 17919 + }, + { + "epoch": 1.967933230836811, + "grad_norm": 2.4601662158966064, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7438104152679443, + "num_tokens": 459673292.0, + "step": 17920 + }, + { + "epoch": 1.9680430485394247, + "grad_norm": 2.096351146697998, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7617954611778259, + "num_tokens": 459698459.0, + "step": 17921 + }, + { + "epoch": 1.9681528662420382, + "grad_norm": 2.4954721927642822, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7337157726287842, + "num_tokens": 459719184.0, + "step": 17922 + }, + { + "epoch": 1.9682626839446518, + "grad_norm": 1.9891455173492432, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7177695631980896, + "num_tokens": 459749421.0, + "step": 17923 + }, + { + "epoch": 1.9683725016472655, + "grad_norm": 2.205322027206421, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7272069454193115, + "num_tokens": 459777498.0, + "step": 17924 + }, + { + "epoch": 1.9684823193498793, + "grad_norm": 2.1997649669647217, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7384295463562012, + "num_tokens": 459800526.0, + "step": 17925 + }, + { + "epoch": 1.9685921370524928, + "grad_norm": 2.323723077774048, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7452163100242615, + "num_tokens": 459823840.0, + "step": 17926 + }, + { + "epoch": 1.9687019547551066, + "grad_norm": 2.2013957500457764, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6881136894226074, + "num_tokens": 459853703.0, + "step": 17927 + }, + { + "epoch": 1.96881177245772, + "grad_norm": 2.2648122310638428, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7256323099136353, + "num_tokens": 459878658.0, + "step": 17928 + }, + { + "epoch": 1.9689215901603339, + "grad_norm": 1.9543681144714355, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7033905386924744, + "num_tokens": 459910713.0, + "step": 17929 + }, + { + "epoch": 1.9690314078629476, + "grad_norm": 2.172778606414795, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7551373839378357, + "num_tokens": 459936221.0, + "step": 17930 + }, + { + "epoch": 1.9691412255655611, + "grad_norm": 2.256477117538452, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7212539911270142, + "num_tokens": 459961363.0, + "step": 17931 + }, + { + "epoch": 1.9692510432681747, + "grad_norm": 2.283689260482788, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7041529417037964, + "num_tokens": 459987399.0, + "step": 17932 + }, + { + "epoch": 1.9693608609707884, + "grad_norm": 2.751681089401245, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7421970367431641, + "num_tokens": 460007578.0, + "step": 17933 + }, + { + "epoch": 1.9694706786734022, + "grad_norm": 2.019690990447998, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7094747424125671, + "num_tokens": 460038335.0, + "step": 17934 + }, + { + "epoch": 1.969580496376016, + "grad_norm": 2.2293498516082764, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7107176184654236, + "num_tokens": 460064259.0, + "step": 17935 + }, + { + "epoch": 1.9696903140786295, + "grad_norm": 2.085052490234375, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7201239466667175, + "num_tokens": 460093056.0, + "step": 17936 + }, + { + "epoch": 1.969800131781243, + "grad_norm": 2.091836929321289, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7237328290939331, + "num_tokens": 460122099.0, + "step": 17937 + }, + { + "epoch": 1.9699099494838568, + "grad_norm": 2.341165065765381, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7238264083862305, + "num_tokens": 460145724.0, + "step": 17938 + }, + { + "epoch": 1.9700197671864705, + "grad_norm": 2.715426445007324, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7367196083068848, + "num_tokens": 460164534.0, + "step": 17939 + }, + { + "epoch": 1.970129584889084, + "grad_norm": 2.379904270172119, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7178433537483215, + "num_tokens": 460187698.0, + "step": 17940 + }, + { + "epoch": 1.9702394025916978, + "grad_norm": 2.1361732482910156, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7278961539268494, + "num_tokens": 460214728.0, + "step": 17941 + }, + { + "epoch": 1.9703492202943114, + "grad_norm": 2.303978443145752, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7490240335464478, + "num_tokens": 460236864.0, + "step": 17942 + }, + { + "epoch": 1.9704590379969251, + "grad_norm": 2.3419671058654785, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7428839206695557, + "num_tokens": 460259556.0, + "step": 17943 + }, + { + "epoch": 1.9705688556995389, + "grad_norm": 2.340240001678467, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7245065569877625, + "num_tokens": 460284065.0, + "step": 17944 + }, + { + "epoch": 1.9706786734021524, + "grad_norm": 2.0950863361358643, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7203525304794312, + "num_tokens": 460313807.0, + "step": 17945 + }, + { + "epoch": 1.970788491104766, + "grad_norm": 2.442937135696411, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7169142365455627, + "num_tokens": 460336657.0, + "step": 17946 + }, + { + "epoch": 1.9708983088073797, + "grad_norm": 2.121290445327759, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7079474925994873, + "num_tokens": 460365175.0, + "step": 17947 + }, + { + "epoch": 1.9710081265099935, + "grad_norm": 2.2317659854888916, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7200732231140137, + "num_tokens": 460390908.0, + "step": 17948 + }, + { + "epoch": 1.9711179442126072, + "grad_norm": 1.9597008228302002, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7037856578826904, + "num_tokens": 460422296.0, + "step": 17949 + }, + { + "epoch": 1.9712277619152208, + "grad_norm": 2.1361372470855713, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7106995582580566, + "num_tokens": 460448977.0, + "step": 17950 + }, + { + "epoch": 1.9713375796178343, + "grad_norm": 2.3472580909729004, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7385573387145996, + "num_tokens": 460470931.0, + "step": 17951 + }, + { + "epoch": 1.971447397320448, + "grad_norm": 2.109241008758545, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7136732935905457, + "num_tokens": 460499264.0, + "step": 17952 + }, + { + "epoch": 1.9715572150230618, + "grad_norm": 2.116446018218994, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.715788722038269, + "num_tokens": 460527320.0, + "step": 17953 + }, + { + "epoch": 1.9716670327256753, + "grad_norm": 2.066607713699341, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.72209233045578, + "num_tokens": 460557059.0, + "step": 17954 + }, + { + "epoch": 1.9717768504282889, + "grad_norm": 2.222492218017578, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7237725853919983, + "num_tokens": 460582757.0, + "step": 17955 + }, + { + "epoch": 1.9718866681309026, + "grad_norm": 2.137537717819214, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7162773013114929, + "num_tokens": 460612148.0, + "step": 17956 + }, + { + "epoch": 1.9719964858335164, + "grad_norm": 2.034878730773926, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7030234336853027, + "num_tokens": 460639972.0, + "step": 17957 + }, + { + "epoch": 1.9721063035361301, + "grad_norm": 2.1700239181518555, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7060178518295288, + "num_tokens": 460666842.0, + "step": 17958 + }, + { + "epoch": 1.9722161212387437, + "grad_norm": 2.0887458324432373, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.708124577999115, + "num_tokens": 460696528.0, + "step": 17959 + }, + { + "epoch": 1.9723259389413572, + "grad_norm": 2.2290189266204834, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6910921931266785, + "num_tokens": 460725120.0, + "step": 17960 + }, + { + "epoch": 1.972435756643971, + "grad_norm": 2.221296787261963, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7250876426696777, + "num_tokens": 460752100.0, + "step": 17961 + }, + { + "epoch": 1.9725455743465847, + "grad_norm": 2.0541365146636963, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6983961462974548, + "num_tokens": 460784440.0, + "step": 17962 + }, + { + "epoch": 1.9726553920491985, + "grad_norm": 2.1321933269500732, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.726847767829895, + "num_tokens": 460811216.0, + "step": 17963 + }, + { + "epoch": 1.972765209751812, + "grad_norm": 2.2051525115966797, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7141485214233398, + "num_tokens": 460835596.0, + "step": 17964 + }, + { + "epoch": 1.9728750274544256, + "grad_norm": 2.208085298538208, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7127548456192017, + "num_tokens": 460862347.0, + "step": 17965 + }, + { + "epoch": 1.9729848451570393, + "grad_norm": 2.059818983078003, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.706201434135437, + "num_tokens": 460890513.0, + "step": 17966 + }, + { + "epoch": 1.973094662859653, + "grad_norm": 2.4498798847198486, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7156105041503906, + "num_tokens": 460913007.0, + "step": 17967 + }, + { + "epoch": 1.9732044805622666, + "grad_norm": 1.9511535167694092, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7216301560401917, + "num_tokens": 460944260.0, + "step": 17968 + }, + { + "epoch": 1.9733142982648801, + "grad_norm": 2.3257429599761963, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7292033433914185, + "num_tokens": 460968399.0, + "step": 17969 + }, + { + "epoch": 1.973424115967494, + "grad_norm": 2.5318849086761475, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7183167934417725, + "num_tokens": 460989554.0, + "step": 17970 + }, + { + "epoch": 1.9735339336701077, + "grad_norm": 2.1874258518218994, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7405538558959961, + "num_tokens": 461016646.0, + "step": 17971 + }, + { + "epoch": 1.9736437513727214, + "grad_norm": 2.3283023834228516, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7223374843597412, + "num_tokens": 461040879.0, + "step": 17972 + }, + { + "epoch": 1.973753569075335, + "grad_norm": 2.3315627574920654, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7187533378601074, + "num_tokens": 461063184.0, + "step": 17973 + }, + { + "epoch": 1.9738633867779485, + "grad_norm": 2.0128822326660156, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7405190467834473, + "num_tokens": 461091897.0, + "step": 17974 + }, + { + "epoch": 1.9739732044805622, + "grad_norm": 2.3627357482910156, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7465628981590271, + "num_tokens": 461114988.0, + "step": 17975 + }, + { + "epoch": 1.974083022183176, + "grad_norm": 2.233228921890259, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7207134962081909, + "num_tokens": 461140158.0, + "step": 17976 + }, + { + "epoch": 1.9741928398857898, + "grad_norm": 2.4956743717193604, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7084590792655945, + "num_tokens": 461162735.0, + "step": 17977 + }, + { + "epoch": 1.9743026575884033, + "grad_norm": 2.4374101161956787, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7397844791412354, + "num_tokens": 461184965.0, + "step": 17978 + }, + { + "epoch": 1.9744124752910168, + "grad_norm": 2.5015735626220703, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7497121095657349, + "num_tokens": 461204855.0, + "step": 17979 + }, + { + "epoch": 1.9745222929936306, + "grad_norm": 2.1595327854156494, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7140644788742065, + "num_tokens": 461231464.0, + "step": 17980 + }, + { + "epoch": 1.9746321106962443, + "grad_norm": 2.2335667610168457, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7117531299591064, + "num_tokens": 461257344.0, + "step": 17981 + }, + { + "epoch": 1.9747419283988579, + "grad_norm": 2.3482184410095215, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7412722706794739, + "num_tokens": 461281571.0, + "step": 17982 + }, + { + "epoch": 1.9748517461014714, + "grad_norm": 2.3928728103637695, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.741281270980835, + "num_tokens": 461303339.0, + "step": 17983 + }, + { + "epoch": 1.9749615638040852, + "grad_norm": 2.22316575050354, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.694898247718811, + "num_tokens": 461329583.0, + "step": 17984 + }, + { + "epoch": 1.975071381506699, + "grad_norm": 2.159038543701172, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7262797355651855, + "num_tokens": 461355991.0, + "step": 17985 + }, + { + "epoch": 1.9751811992093127, + "grad_norm": 2.265533208847046, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7144650220870972, + "num_tokens": 461380977.0, + "step": 17986 + }, + { + "epoch": 1.9752910169119262, + "grad_norm": 2.0421435832977295, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6986762285232544, + "num_tokens": 461409581.0, + "step": 17987 + }, + { + "epoch": 1.9754008346145397, + "grad_norm": 2.10591983795166, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7249978184700012, + "num_tokens": 461435167.0, + "step": 17988 + }, + { + "epoch": 1.9755106523171535, + "grad_norm": 2.2685532569885254, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7450821995735168, + "num_tokens": 461457927.0, + "step": 17989 + }, + { + "epoch": 1.9756204700197673, + "grad_norm": 2.517935037612915, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.734471321105957, + "num_tokens": 461478811.0, + "step": 17990 + }, + { + "epoch": 1.9757302877223808, + "grad_norm": 2.521854877471924, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7198958396911621, + "num_tokens": 461500339.0, + "step": 17991 + }, + { + "epoch": 1.9758401054249946, + "grad_norm": 2.4433486461639404, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7285282015800476, + "num_tokens": 461522280.0, + "step": 17992 + }, + { + "epoch": 1.975949923127608, + "grad_norm": 2.48105788230896, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7411744594573975, + "num_tokens": 461542773.0, + "step": 17993 + }, + { + "epoch": 1.9760597408302218, + "grad_norm": 2.2723276615142822, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7047727108001709, + "num_tokens": 461569215.0, + "step": 17994 + }, + { + "epoch": 1.9761695585328356, + "grad_norm": 2.4216437339782715, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7132154703140259, + "num_tokens": 461592143.0, + "step": 17995 + }, + { + "epoch": 1.9762793762354491, + "grad_norm": 2.074751853942871, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7339995503425598, + "num_tokens": 461620169.0, + "step": 17996 + }, + { + "epoch": 1.9763891939380627, + "grad_norm": 2.1453030109405518, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7215080261230469, + "num_tokens": 461646161.0, + "step": 17997 + }, + { + "epoch": 1.9764990116406764, + "grad_norm": 2.087632656097412, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.706121563911438, + "num_tokens": 461675595.0, + "step": 17998 + }, + { + "epoch": 1.9766088293432902, + "grad_norm": 2.156294822692871, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7322651147842407, + "num_tokens": 461701846.0, + "step": 17999 + }, + { + "epoch": 1.976718647045904, + "grad_norm": 2.4762802124023438, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7210540175437927, + "num_tokens": 461722822.0, + "step": 18000 + }, + { + "epoch": 1.9768284647485175, + "grad_norm": 2.412191867828369, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7066256999969482, + "num_tokens": 461746942.0, + "step": 18001 + }, + { + "epoch": 1.976938282451131, + "grad_norm": 2.2534830570220947, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7297312021255493, + "num_tokens": 461770528.0, + "step": 18002 + }, + { + "epoch": 1.9770481001537448, + "grad_norm": 2.2138419151306152, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7201095819473267, + "num_tokens": 461794125.0, + "step": 18003 + }, + { + "epoch": 1.9771579178563585, + "grad_norm": 2.239000082015991, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7040032744407654, + "num_tokens": 461820910.0, + "step": 18004 + }, + { + "epoch": 1.977267735558972, + "grad_norm": 2.270630359649658, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7332980036735535, + "num_tokens": 461847461.0, + "step": 18005 + }, + { + "epoch": 1.9773775532615858, + "grad_norm": 2.502877712249756, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7311612367630005, + "num_tokens": 461869927.0, + "step": 18006 + }, + { + "epoch": 1.9774873709641994, + "grad_norm": 2.222565174102783, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7176811695098877, + "num_tokens": 461896805.0, + "step": 18007 + }, + { + "epoch": 1.977597188666813, + "grad_norm": 2.0080020427703857, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6852322816848755, + "num_tokens": 461927159.0, + "step": 18008 + }, + { + "epoch": 1.9777070063694269, + "grad_norm": 2.1371238231658936, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6885544061660767, + "num_tokens": 461957057.0, + "step": 18009 + }, + { + "epoch": 1.9778168240720404, + "grad_norm": 2.376862049102783, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7209322452545166, + "num_tokens": 461980263.0, + "step": 18010 + }, + { + "epoch": 1.977926641774654, + "grad_norm": 2.2378993034362793, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7151727676391602, + "num_tokens": 462005902.0, + "step": 18011 + }, + { + "epoch": 1.9780364594772677, + "grad_norm": 2.4044017791748047, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7202996611595154, + "num_tokens": 462028869.0, + "step": 18012 + }, + { + "epoch": 1.9781462771798815, + "grad_norm": 2.251444101333618, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7310945987701416, + "num_tokens": 462051940.0, + "step": 18013 + }, + { + "epoch": 1.9782560948824952, + "grad_norm": 2.3745903968811035, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7118474841117859, + "num_tokens": 462073486.0, + "step": 18014 + }, + { + "epoch": 1.9783659125851087, + "grad_norm": 2.27931809425354, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7103502750396729, + "num_tokens": 462099520.0, + "step": 18015 + }, + { + "epoch": 1.9784757302877223, + "grad_norm": 2.5020904541015625, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7141944169998169, + "num_tokens": 462120252.0, + "step": 18016 + }, + { + "epoch": 1.978585547990336, + "grad_norm": 2.2875852584838867, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7073556780815125, + "num_tokens": 462146764.0, + "step": 18017 + }, + { + "epoch": 1.9786953656929498, + "grad_norm": 2.018916368484497, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7267170548439026, + "num_tokens": 462174733.0, + "step": 18018 + }, + { + "epoch": 1.9788051833955633, + "grad_norm": 2.1421399116516113, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7075048685073853, + "num_tokens": 462202775.0, + "step": 18019 + }, + { + "epoch": 1.9789150010981769, + "grad_norm": 2.3055591583251953, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7467970252037048, + "num_tokens": 462226398.0, + "step": 18020 + }, + { + "epoch": 1.9790248188007906, + "grad_norm": 2.0576558113098145, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7352992296218872, + "num_tokens": 462253591.0, + "step": 18021 + }, + { + "epoch": 1.9791346365034044, + "grad_norm": 2.180785894393921, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7126243710517883, + "num_tokens": 462280796.0, + "step": 18022 + }, + { + "epoch": 1.9792444542060181, + "grad_norm": 2.2273995876312256, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7215895652770996, + "num_tokens": 462304773.0, + "step": 18023 + }, + { + "epoch": 1.9793542719086317, + "grad_norm": 2.4988996982574463, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7115299701690674, + "num_tokens": 462326571.0, + "step": 18024 + }, + { + "epoch": 1.9794640896112452, + "grad_norm": 2.283688545227051, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.72139573097229, + "num_tokens": 462352163.0, + "step": 18025 + }, + { + "epoch": 1.979573907313859, + "grad_norm": 2.19724702835083, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7000665664672852, + "num_tokens": 462380112.0, + "step": 18026 + }, + { + "epoch": 1.9796837250164727, + "grad_norm": 2.4271790981292725, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7313414216041565, + "num_tokens": 462402774.0, + "step": 18027 + }, + { + "epoch": 1.9797935427190865, + "grad_norm": 2.11936354637146, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.721061110496521, + "num_tokens": 462430872.0, + "step": 18028 + }, + { + "epoch": 1.9799033604217, + "grad_norm": 2.39898943901062, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.758271336555481, + "num_tokens": 462452857.0, + "step": 18029 + }, + { + "epoch": 1.9800131781243135, + "grad_norm": 2.169776678085327, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7509599924087524, + "num_tokens": 462478869.0, + "step": 18030 + }, + { + "epoch": 1.9801229958269273, + "grad_norm": 2.596977710723877, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7370299696922302, + "num_tokens": 462500240.0, + "step": 18031 + }, + { + "epoch": 1.980232813529541, + "grad_norm": 2.226224660873413, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7212784886360168, + "num_tokens": 462525542.0, + "step": 18032 + }, + { + "epoch": 1.9803426312321546, + "grad_norm": 2.3240582942962646, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7261368036270142, + "num_tokens": 462548463.0, + "step": 18033 + }, + { + "epoch": 1.9804524489347681, + "grad_norm": 2.3442461490631104, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7375984191894531, + "num_tokens": 462571563.0, + "step": 18034 + }, + { + "epoch": 1.9805622666373819, + "grad_norm": 2.418142795562744, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7070482969284058, + "num_tokens": 462593870.0, + "step": 18035 + }, + { + "epoch": 1.9806720843399956, + "grad_norm": 2.305852174758911, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7043542861938477, + "num_tokens": 462620625.0, + "step": 18036 + }, + { + "epoch": 1.9807819020426094, + "grad_norm": 2.3641750812530518, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.714352548122406, + "num_tokens": 462644462.0, + "step": 18037 + }, + { + "epoch": 1.980891719745223, + "grad_norm": 2.2186224460601807, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.69352787733078, + "num_tokens": 462672491.0, + "step": 18038 + }, + { + "epoch": 1.9810015374478365, + "grad_norm": 2.1810431480407715, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7143250703811646, + "num_tokens": 462699904.0, + "step": 18039 + }, + { + "epoch": 1.9811113551504502, + "grad_norm": 2.3168373107910156, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7206674814224243, + "num_tokens": 462723575.0, + "step": 18040 + }, + { + "epoch": 1.981221172853064, + "grad_norm": 2.1133010387420654, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7141610383987427, + "num_tokens": 462752299.0, + "step": 18041 + }, + { + "epoch": 1.9813309905556777, + "grad_norm": 2.3811843395233154, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7466066479682922, + "num_tokens": 462774477.0, + "step": 18042 + }, + { + "epoch": 1.9814408082582913, + "grad_norm": 2.2782249450683594, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.738947331905365, + "num_tokens": 462799236.0, + "step": 18043 + }, + { + "epoch": 1.9815506259609048, + "grad_norm": 2.555985450744629, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7144104838371277, + "num_tokens": 462820308.0, + "step": 18044 + }, + { + "epoch": 1.9816604436635186, + "grad_norm": 2.256382942199707, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7147858738899231, + "num_tokens": 462845696.0, + "step": 18045 + }, + { + "epoch": 1.9817702613661323, + "grad_norm": 2.128359079360962, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7132878303527832, + "num_tokens": 462872653.0, + "step": 18046 + }, + { + "epoch": 1.9818800790687459, + "grad_norm": 2.132540702819824, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7290405035018921, + "num_tokens": 462899933.0, + "step": 18047 + }, + { + "epoch": 1.9819898967713594, + "grad_norm": 2.095975875854492, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7304458022117615, + "num_tokens": 462926802.0, + "step": 18048 + }, + { + "epoch": 1.9820997144739732, + "grad_norm": 2.0502893924713135, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6952057480812073, + "num_tokens": 462957695.0, + "step": 18049 + }, + { + "epoch": 1.982209532176587, + "grad_norm": 2.368535041809082, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7377697229385376, + "num_tokens": 462980593.0, + "step": 18050 + }, + { + "epoch": 1.9823193498792007, + "grad_norm": 1.9900532960891724, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7113105058670044, + "num_tokens": 463011537.0, + "step": 18051 + }, + { + "epoch": 1.9824291675818142, + "grad_norm": 2.411036252975464, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7155349254608154, + "num_tokens": 463034406.0, + "step": 18052 + }, + { + "epoch": 1.9825389852844277, + "grad_norm": 2.380045175552368, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.73041832447052, + "num_tokens": 463056847.0, + "step": 18053 + }, + { + "epoch": 1.9826488029870415, + "grad_norm": 2.2640280723571777, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7349461913108826, + "num_tokens": 463081152.0, + "step": 18054 + }, + { + "epoch": 1.9827586206896552, + "grad_norm": 2.066145181655884, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7063362002372742, + "num_tokens": 463110248.0, + "step": 18055 + }, + { + "epoch": 1.9828684383922688, + "grad_norm": 2.1891820430755615, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7222006916999817, + "num_tokens": 463136594.0, + "step": 18056 + }, + { + "epoch": 1.9829782560948825, + "grad_norm": 2.330374240875244, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7058874368667603, + "num_tokens": 463161388.0, + "step": 18057 + }, + { + "epoch": 1.983088073797496, + "grad_norm": 2.1864960193634033, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7365071773529053, + "num_tokens": 463187082.0, + "step": 18058 + }, + { + "epoch": 1.9831978915001098, + "grad_norm": 2.2207067012786865, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7058620452880859, + "num_tokens": 463214950.0, + "step": 18059 + }, + { + "epoch": 1.9833077092027236, + "grad_norm": 2.4247076511383057, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7168018817901611, + "num_tokens": 463237010.0, + "step": 18060 + }, + { + "epoch": 1.9834175269053371, + "grad_norm": 1.956025242805481, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7145317792892456, + "num_tokens": 463270265.0, + "step": 18061 + }, + { + "epoch": 1.9835273446079507, + "grad_norm": 2.015010356903076, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7229537963867188, + "num_tokens": 463302182.0, + "step": 18062 + }, + { + "epoch": 1.9836371623105644, + "grad_norm": 2.5301074981689453, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7080324292182922, + "num_tokens": 463324768.0, + "step": 18063 + }, + { + "epoch": 1.9837469800131782, + "grad_norm": 2.1762077808380127, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7424991130828857, + "num_tokens": 463350186.0, + "step": 18064 + }, + { + "epoch": 1.983856797715792, + "grad_norm": 2.118035078048706, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7212111353874207, + "num_tokens": 463376791.0, + "step": 18065 + }, + { + "epoch": 1.9839666154184055, + "grad_norm": 2.032911777496338, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7371350526809692, + "num_tokens": 463404289.0, + "step": 18066 + }, + { + "epoch": 1.984076433121019, + "grad_norm": 2.2388908863067627, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7229368686676025, + "num_tokens": 463428691.0, + "step": 18067 + }, + { + "epoch": 1.9841862508236328, + "grad_norm": 2.1621153354644775, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.763814389705658, + "num_tokens": 463451783.0, + "step": 18068 + }, + { + "epoch": 1.9842960685262465, + "grad_norm": 2.5772974491119385, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7260458469390869, + "num_tokens": 463472841.0, + "step": 18069 + }, + { + "epoch": 1.98440588622886, + "grad_norm": 2.1837360858917236, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7412998080253601, + "num_tokens": 463497606.0, + "step": 18070 + }, + { + "epoch": 1.9845157039314738, + "grad_norm": 2.337097644805908, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.7060233950614929, + "num_tokens": 463523181.0, + "step": 18071 + }, + { + "epoch": 1.9846255216340873, + "grad_norm": 1.8376131057739258, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7024450302124023, + "num_tokens": 463558366.0, + "step": 18072 + }, + { + "epoch": 1.984735339336701, + "grad_norm": 2.4884355068206787, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.722454309463501, + "num_tokens": 463581598.0, + "step": 18073 + }, + { + "epoch": 1.9848451570393149, + "grad_norm": 2.1362085342407227, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7218074798583984, + "num_tokens": 463608930.0, + "step": 18074 + }, + { + "epoch": 1.9849549747419284, + "grad_norm": 2.588735342025757, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7347767949104309, + "num_tokens": 463628563.0, + "step": 18075 + }, + { + "epoch": 1.985064792444542, + "grad_norm": 2.8782405853271484, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7615119218826294, + "num_tokens": 463644868.0, + "step": 18076 + }, + { + "epoch": 1.9851746101471557, + "grad_norm": 2.21209979057312, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7336850762367249, + "num_tokens": 463670308.0, + "step": 18077 + }, + { + "epoch": 1.9852844278497694, + "grad_norm": 2.3412277698516846, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7210593223571777, + "num_tokens": 463694843.0, + "step": 18078 + }, + { + "epoch": 1.9853942455523832, + "grad_norm": 2.1767067909240723, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7188190817832947, + "num_tokens": 463722358.0, + "step": 18079 + }, + { + "epoch": 1.9855040632549967, + "grad_norm": 2.3089582920074463, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7410605549812317, + "num_tokens": 463746035.0, + "step": 18080 + }, + { + "epoch": 1.9856138809576103, + "grad_norm": 2.3267290592193604, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7158840298652649, + "num_tokens": 463773554.0, + "step": 18081 + }, + { + "epoch": 1.985723698660224, + "grad_norm": 2.523219347000122, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7092858552932739, + "num_tokens": 463796835.0, + "step": 18082 + }, + { + "epoch": 1.9858335163628378, + "grad_norm": 2.1002089977264404, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7120125889778137, + "num_tokens": 463827096.0, + "step": 18083 + }, + { + "epoch": 1.9859433340654513, + "grad_norm": 2.146963596343994, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7358231544494629, + "num_tokens": 463853446.0, + "step": 18084 + }, + { + "epoch": 1.9860531517680649, + "grad_norm": 2.0717947483062744, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7078735828399658, + "num_tokens": 463882659.0, + "step": 18085 + }, + { + "epoch": 1.9861629694706786, + "grad_norm": 2.261380910873413, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7356127500534058, + "num_tokens": 463906756.0, + "step": 18086 + }, + { + "epoch": 1.9862727871732924, + "grad_norm": 2.354153633117676, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7481978535652161, + "num_tokens": 463930984.0, + "step": 18087 + }, + { + "epoch": 1.9863826048759061, + "grad_norm": 2.1682910919189453, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.68790602684021, + "num_tokens": 463957723.0, + "step": 18088 + }, + { + "epoch": 1.9864924225785197, + "grad_norm": 2.3693125247955322, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7269549369812012, + "num_tokens": 463980373.0, + "step": 18089 + }, + { + "epoch": 1.9866022402811332, + "grad_norm": 2.3241539001464844, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7203229665756226, + "num_tokens": 464004466.0, + "step": 18090 + }, + { + "epoch": 1.986712057983747, + "grad_norm": 1.9153883457183838, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6930828094482422, + "num_tokens": 464036641.0, + "step": 18091 + }, + { + "epoch": 1.9868218756863607, + "grad_norm": 2.250108242034912, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7418568134307861, + "num_tokens": 464061077.0, + "step": 18092 + }, + { + "epoch": 1.9869316933889745, + "grad_norm": 2.206566333770752, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.718204140663147, + "num_tokens": 464087465.0, + "step": 18093 + }, + { + "epoch": 1.987041511091588, + "grad_norm": 2.0102765560150146, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.713516116142273, + "num_tokens": 464118844.0, + "step": 18094 + }, + { + "epoch": 1.9871513287942015, + "grad_norm": 2.2861623764038086, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7180918455123901, + "num_tokens": 464143848.0, + "step": 18095 + }, + { + "epoch": 1.9872611464968153, + "grad_norm": 2.3435051441192627, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7090266942977905, + "num_tokens": 464166823.0, + "step": 18096 + }, + { + "epoch": 1.987370964199429, + "grad_norm": 2.441147565841675, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7251808047294617, + "num_tokens": 464190061.0, + "step": 18097 + }, + { + "epoch": 1.9874807819020426, + "grad_norm": 2.341477870941162, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7068461179733276, + "num_tokens": 464213293.0, + "step": 18098 + }, + { + "epoch": 1.9875905996046561, + "grad_norm": 2.151045322418213, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7204764485359192, + "num_tokens": 464238725.0, + "step": 18099 + }, + { + "epoch": 1.9877004173072699, + "grad_norm": 2.1135125160217285, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7132638692855835, + "num_tokens": 464269123.0, + "step": 18100 + }, + { + "epoch": 1.9878102350098836, + "grad_norm": 2.117696523666382, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7185662984848022, + "num_tokens": 464297829.0, + "step": 18101 + }, + { + "epoch": 1.9879200527124974, + "grad_norm": 2.3277244567871094, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7616850733757019, + "num_tokens": 464319552.0, + "step": 18102 + }, + { + "epoch": 1.988029870415111, + "grad_norm": 2.084864377975464, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7169961929321289, + "num_tokens": 464349910.0, + "step": 18103 + }, + { + "epoch": 1.9881396881177245, + "grad_norm": 2.132615327835083, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7278394103050232, + "num_tokens": 464377117.0, + "step": 18104 + }, + { + "epoch": 1.9882495058203382, + "grad_norm": 2.2588274478912354, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7585224509239197, + "num_tokens": 464401756.0, + "step": 18105 + }, + { + "epoch": 1.988359323522952, + "grad_norm": 2.2944486141204834, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7338151931762695, + "num_tokens": 464425997.0, + "step": 18106 + }, + { + "epoch": 1.9884691412255655, + "grad_norm": 2.570518732070923, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7093312740325928, + "num_tokens": 464447375.0, + "step": 18107 + }, + { + "epoch": 1.9885789589281793, + "grad_norm": 2.179363489151001, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7353068590164185, + "num_tokens": 464472294.0, + "step": 18108 + }, + { + "epoch": 1.9886887766307928, + "grad_norm": 2.4876456260681152, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7378942966461182, + "num_tokens": 464491278.0, + "step": 18109 + }, + { + "epoch": 1.9887985943334066, + "grad_norm": 2.196261167526245, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7192076444625854, + "num_tokens": 464523155.0, + "step": 18110 + }, + { + "epoch": 1.9889084120360203, + "grad_norm": 2.353858709335327, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7522579431533813, + "num_tokens": 464543932.0, + "step": 18111 + }, + { + "epoch": 1.9890182297386338, + "grad_norm": 2.1517868041992188, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7057119011878967, + "num_tokens": 464570154.0, + "step": 18112 + }, + { + "epoch": 1.9891280474412474, + "grad_norm": 1.9041414260864258, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7239748239517212, + "num_tokens": 464602852.0, + "step": 18113 + }, + { + "epoch": 1.9892378651438611, + "grad_norm": 2.2513368129730225, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7097852826118469, + "num_tokens": 464626426.0, + "step": 18114 + }, + { + "epoch": 1.989347682846475, + "grad_norm": 2.5462167263031006, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7198802828788757, + "num_tokens": 464646602.0, + "step": 18115 + }, + { + "epoch": 1.9894575005490887, + "grad_norm": 2.471968412399292, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7280685901641846, + "num_tokens": 464668754.0, + "step": 18116 + }, + { + "epoch": 1.9895673182517022, + "grad_norm": 1.8669545650482178, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7027350068092346, + "num_tokens": 464703610.0, + "step": 18117 + }, + { + "epoch": 1.9896771359543157, + "grad_norm": 2.5746853351593018, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7330923080444336, + "num_tokens": 464723284.0, + "step": 18118 + }, + { + "epoch": 1.9897869536569295, + "grad_norm": 2.2268869876861572, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7349194288253784, + "num_tokens": 464748485.0, + "step": 18119 + }, + { + "epoch": 1.9898967713595432, + "grad_norm": 2.0520009994506836, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7011170983314514, + "num_tokens": 464778251.0, + "step": 18120 + }, + { + "epoch": 1.9900065890621568, + "grad_norm": 2.2719855308532715, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.718977689743042, + "num_tokens": 464800923.0, + "step": 18121 + }, + { + "epoch": 1.9901164067647705, + "grad_norm": 2.204921245574951, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7170863151550293, + "num_tokens": 464827770.0, + "step": 18122 + }, + { + "epoch": 1.990226224467384, + "grad_norm": 2.213765859603882, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7318276166915894, + "num_tokens": 464853634.0, + "step": 18123 + }, + { + "epoch": 1.9903360421699978, + "grad_norm": 1.9430345296859741, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7236602306365967, + "num_tokens": 464884537.0, + "step": 18124 + }, + { + "epoch": 1.9904458598726116, + "grad_norm": 2.3929717540740967, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.736514151096344, + "num_tokens": 464906539.0, + "step": 18125 + }, + { + "epoch": 1.9905556775752251, + "grad_norm": 2.2634620666503906, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7438970804214478, + "num_tokens": 464929920.0, + "step": 18126 + }, + { + "epoch": 1.9906654952778386, + "grad_norm": 2.198529005050659, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7128996849060059, + "num_tokens": 464956035.0, + "step": 18127 + }, + { + "epoch": 1.9907753129804524, + "grad_norm": 2.3194739818573, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7163445949554443, + "num_tokens": 464979870.0, + "step": 18128 + }, + { + "epoch": 1.9908851306830662, + "grad_norm": 2.0902583599090576, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7145329713821411, + "num_tokens": 465009966.0, + "step": 18129 + }, + { + "epoch": 1.99099494838568, + "grad_norm": 2.324866771697998, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.712903618812561, + "num_tokens": 465034729.0, + "step": 18130 + }, + { + "epoch": 1.9911047660882935, + "grad_norm": 2.323928117752075, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7234622836112976, + "num_tokens": 465058138.0, + "step": 18131 + }, + { + "epoch": 1.991214583790907, + "grad_norm": 2.1513655185699463, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.712882399559021, + "num_tokens": 465085043.0, + "step": 18132 + }, + { + "epoch": 1.9913244014935207, + "grad_norm": 2.5060436725616455, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7589163780212402, + "num_tokens": 465103915.0, + "step": 18133 + }, + { + "epoch": 1.9914342191961345, + "grad_norm": 2.196826457977295, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7391430735588074, + "num_tokens": 465128947.0, + "step": 18134 + }, + { + "epoch": 1.991544036898748, + "grad_norm": 2.265316963195801, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7107523679733276, + "num_tokens": 465153935.0, + "step": 18135 + }, + { + "epoch": 1.9916538546013618, + "grad_norm": 2.5634632110595703, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.751093327999115, + "num_tokens": 465175537.0, + "step": 18136 + }, + { + "epoch": 1.9917636723039753, + "grad_norm": 2.3335297107696533, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6903693079948425, + "num_tokens": 465200869.0, + "step": 18137 + }, + { + "epoch": 1.991873490006589, + "grad_norm": 2.2950446605682373, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7392042875289917, + "num_tokens": 465223578.0, + "step": 18138 + }, + { + "epoch": 1.9919833077092028, + "grad_norm": 2.11171293258667, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7096972465515137, + "num_tokens": 465252469.0, + "step": 18139 + }, + { + "epoch": 1.9920931254118164, + "grad_norm": 2.1047656536102295, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7078375220298767, + "num_tokens": 465282460.0, + "step": 18140 + }, + { + "epoch": 1.99220294311443, + "grad_norm": 1.9414446353912354, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6986327171325684, + "num_tokens": 465315589.0, + "step": 18141 + }, + { + "epoch": 1.9923127608170437, + "grad_norm": 2.0781657695770264, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.693572998046875, + "num_tokens": 465344681.0, + "step": 18142 + }, + { + "epoch": 1.9924225785196574, + "grad_norm": 2.3029580116271973, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7378934621810913, + "num_tokens": 465367651.0, + "step": 18143 + }, + { + "epoch": 1.9925323962222712, + "grad_norm": 2.212700843811035, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7323645949363708, + "num_tokens": 465392738.0, + "step": 18144 + }, + { + "epoch": 1.9926422139248847, + "grad_norm": 2.1708712577819824, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7286897897720337, + "num_tokens": 465418503.0, + "step": 18145 + }, + { + "epoch": 1.9927520316274983, + "grad_norm": 2.044192314147949, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7024203538894653, + "num_tokens": 465450546.0, + "step": 18146 + }, + { + "epoch": 1.992861849330112, + "grad_norm": 2.5408949851989746, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7186235785484314, + "num_tokens": 465471772.0, + "step": 18147 + }, + { + "epoch": 1.9929716670327258, + "grad_norm": 2.2889490127563477, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7079200148582458, + "num_tokens": 465496427.0, + "step": 18148 + }, + { + "epoch": 1.9930814847353393, + "grad_norm": 2.0983052253723145, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7053653001785278, + "num_tokens": 465526116.0, + "step": 18149 + }, + { + "epoch": 1.9931913024379528, + "grad_norm": 1.9528682231903076, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7087922692298889, + "num_tokens": 465558009.0, + "step": 18150 + }, + { + "epoch": 1.9933011201405666, + "grad_norm": 2.140869379043579, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7382440567016602, + "num_tokens": 465582748.0, + "step": 18151 + }, + { + "epoch": 1.9934109378431804, + "grad_norm": 1.974332332611084, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7284373044967651, + "num_tokens": 465610599.0, + "step": 18152 + }, + { + "epoch": 1.993520755545794, + "grad_norm": 1.9396644830703735, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7099358439445496, + "num_tokens": 465642313.0, + "step": 18153 + }, + { + "epoch": 1.9936305732484076, + "grad_norm": 2.026761770248413, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6954523921012878, + "num_tokens": 465672591.0, + "step": 18154 + }, + { + "epoch": 1.9937403909510212, + "grad_norm": 2.393620491027832, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7195651531219482, + "num_tokens": 465696151.0, + "step": 18155 + }, + { + "epoch": 1.993850208653635, + "grad_norm": 2.276874303817749, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.751192033290863, + "num_tokens": 465718848.0, + "step": 18156 + }, + { + "epoch": 1.9939600263562487, + "grad_norm": 2.190526247024536, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7123718857765198, + "num_tokens": 465747685.0, + "step": 18157 + }, + { + "epoch": 1.9940698440588625, + "grad_norm": 2.5272653102874756, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7311592102050781, + "num_tokens": 465768492.0, + "step": 18158 + }, + { + "epoch": 1.994179661761476, + "grad_norm": 2.1122453212738037, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.729576051235199, + "num_tokens": 465794339.0, + "step": 18159 + }, + { + "epoch": 1.9942894794640895, + "grad_norm": 2.148319721221924, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7222204804420471, + "num_tokens": 465820779.0, + "step": 18160 + }, + { + "epoch": 1.9943992971667033, + "grad_norm": 2.482189178466797, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7050080895423889, + "num_tokens": 465843542.0, + "step": 18161 + }, + { + "epoch": 1.994509114869317, + "grad_norm": 2.313683032989502, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7319438457489014, + "num_tokens": 465867395.0, + "step": 18162 + }, + { + "epoch": 1.9946189325719306, + "grad_norm": 2.2490296363830566, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7219904661178589, + "num_tokens": 465891804.0, + "step": 18163 + }, + { + "epoch": 1.994728750274544, + "grad_norm": 2.4521594047546387, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7312661409378052, + "num_tokens": 465914486.0, + "step": 18164 + }, + { + "epoch": 1.9948385679771579, + "grad_norm": 2.4169228076934814, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7344173192977905, + "num_tokens": 465937605.0, + "step": 18165 + }, + { + "epoch": 1.9949483856797716, + "grad_norm": 2.4677717685699463, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.714030921459198, + "num_tokens": 465960254.0, + "step": 18166 + }, + { + "epoch": 1.9950582033823854, + "grad_norm": 2.1431291103363037, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7168617248535156, + "num_tokens": 465987117.0, + "step": 18167 + }, + { + "epoch": 1.995168021084999, + "grad_norm": 2.1734063625335693, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7180953025817871, + "num_tokens": 466014134.0, + "step": 18168 + }, + { + "epoch": 1.9952778387876124, + "grad_norm": 2.2754099369049072, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7049349546432495, + "num_tokens": 466040897.0, + "step": 18169 + }, + { + "epoch": 1.9953876564902262, + "grad_norm": 1.9701510667800903, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7014292478561401, + "num_tokens": 466072674.0, + "step": 18170 + }, + { + "epoch": 1.99549747419284, + "grad_norm": 2.1940805912017822, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.737757682800293, + "num_tokens": 466099299.0, + "step": 18171 + }, + { + "epoch": 1.9956072918954535, + "grad_norm": 2.455382823944092, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7346742153167725, + "num_tokens": 466120376.0, + "step": 18172 + }, + { + "epoch": 1.9957171095980673, + "grad_norm": 2.553286075592041, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7357741594314575, + "num_tokens": 466141693.0, + "step": 18173 + }, + { + "epoch": 1.9958269273006808, + "grad_norm": 2.1158289909362793, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7175855040550232, + "num_tokens": 466169170.0, + "step": 18174 + }, + { + "epoch": 1.9959367450032945, + "grad_norm": 2.343855381011963, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7182599306106567, + "num_tokens": 466195084.0, + "step": 18175 + }, + { + "epoch": 1.9960465627059083, + "grad_norm": 2.0905020236968994, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7154130339622498, + "num_tokens": 466226498.0, + "step": 18176 + }, + { + "epoch": 1.9961563804085218, + "grad_norm": 2.3668315410614014, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7350367903709412, + "num_tokens": 466250182.0, + "step": 18177 + }, + { + "epoch": 1.9962661981111354, + "grad_norm": 2.1981046199798584, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7218688726425171, + "num_tokens": 466274016.0, + "step": 18178 + }, + { + "epoch": 1.9963760158137491, + "grad_norm": 2.066147804260254, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6914207339286804, + "num_tokens": 466303380.0, + "step": 18179 + }, + { + "epoch": 1.9964858335163629, + "grad_norm": 2.200098752975464, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7124510407447815, + "num_tokens": 466329273.0, + "step": 18180 + }, + { + "epoch": 1.9965956512189766, + "grad_norm": 2.109096050262451, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6993648409843445, + "num_tokens": 466357154.0, + "step": 18181 + }, + { + "epoch": 1.9967054689215902, + "grad_norm": 2.0498228073120117, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7084352970123291, + "num_tokens": 466385595.0, + "step": 18182 + }, + { + "epoch": 1.9968152866242037, + "grad_norm": 1.9840315580368042, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7311738729476929, + "num_tokens": 466415508.0, + "step": 18183 + }, + { + "epoch": 1.9969251043268175, + "grad_norm": 2.3990025520324707, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7181887626647949, + "num_tokens": 466437103.0, + "step": 18184 + }, + { + "epoch": 1.9970349220294312, + "grad_norm": 2.4993152618408203, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7140941619873047, + "num_tokens": 466458768.0, + "step": 18185 + }, + { + "epoch": 1.9971447397320448, + "grad_norm": 2.1321303844451904, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7108628749847412, + "num_tokens": 466487013.0, + "step": 18186 + }, + { + "epoch": 1.9972545574346585, + "grad_norm": 2.027545690536499, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7222858667373657, + "num_tokens": 466518715.0, + "step": 18187 + }, + { + "epoch": 1.997364375137272, + "grad_norm": 2.663356304168701, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7308062314987183, + "num_tokens": 466539108.0, + "step": 18188 + }, + { + "epoch": 1.9974741928398858, + "grad_norm": 2.4046335220336914, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7329833507537842, + "num_tokens": 466562880.0, + "step": 18189 + }, + { + "epoch": 1.9975840105424996, + "grad_norm": 2.207967758178711, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7270934581756592, + "num_tokens": 466589585.0, + "step": 18190 + }, + { + "epoch": 1.997693828245113, + "grad_norm": 2.1162726879119873, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.6976704001426697, + "num_tokens": 466618821.0, + "step": 18191 + }, + { + "epoch": 1.9978036459477266, + "grad_norm": 2.4908559322357178, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7230792045593262, + "num_tokens": 466639990.0, + "step": 18192 + }, + { + "epoch": 1.9979134636503404, + "grad_norm": 2.156412363052368, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7042646408081055, + "num_tokens": 466668541.0, + "step": 18193 + }, + { + "epoch": 1.9980232813529542, + "grad_norm": 2.114194869995117, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6918463706970215, + "num_tokens": 466698154.0, + "step": 18194 + }, + { + "epoch": 1.998133099055568, + "grad_norm": 2.2212932109832764, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7371777296066284, + "num_tokens": 466722944.0, + "step": 18195 + }, + { + "epoch": 1.9982429167581814, + "grad_norm": 2.033491849899292, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7104830741882324, + "num_tokens": 466753006.0, + "step": 18196 + }, + { + "epoch": 1.998352734460795, + "grad_norm": 2.453094959259033, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7101881504058838, + "num_tokens": 466776445.0, + "step": 18197 + }, + { + "epoch": 1.9984625521634087, + "grad_norm": 2.3824915885925293, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7465223073959351, + "num_tokens": 466799008.0, + "step": 18198 + }, + { + "epoch": 1.9985723698660225, + "grad_norm": 2.3102076053619385, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7335169315338135, + "num_tokens": 466822418.0, + "step": 18199 + }, + { + "epoch": 1.998682187568636, + "grad_norm": 2.5589418411254883, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7134063839912415, + "num_tokens": 466843577.0, + "step": 18200 + }, + { + "epoch": 1.9987920052712496, + "grad_norm": 2.252307891845703, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7400009632110596, + "num_tokens": 466866979.0, + "step": 18201 + }, + { + "epoch": 1.9989018229738633, + "grad_norm": 2.189452648162842, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7010281085968018, + "num_tokens": 466896661.0, + "step": 18202 + }, + { + "epoch": 1.999011640676477, + "grad_norm": 2.5723800659179688, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7552800178527832, + "num_tokens": 466918589.0, + "step": 18203 + }, + { + "epoch": 1.9991214583790908, + "grad_norm": 2.5652098655700684, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.748008131980896, + "num_tokens": 466939308.0, + "step": 18204 + }, + { + "epoch": 1.9992312760817044, + "grad_norm": 2.281175374984741, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7248058319091797, + "num_tokens": 466963643.0, + "step": 18205 + }, + { + "epoch": 1.999341093784318, + "grad_norm": 2.2615532875061035, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7034668326377869, + "num_tokens": 466990568.0, + "step": 18206 + }, + { + "epoch": 1.9994509114869317, + "grad_norm": 2.408552408218384, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7338019609451294, + "num_tokens": 467013809.0, + "step": 18207 + }, + { + "epoch": 1.9995607291895454, + "grad_norm": 2.1967737674713135, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7490490078926086, + "num_tokens": 467038482.0, + "step": 18208 + }, + { + "epoch": 1.9996705468921592, + "grad_norm": 1.9150054454803467, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7151707410812378, + "num_tokens": 467072471.0, + "step": 18209 + }, + { + "epoch": 1.9997803645947727, + "grad_norm": 2.4749984741210938, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7339646220207214, + "num_tokens": 467093891.0, + "step": 18210 + }, + { + "epoch": 1.9998901822973862, + "grad_norm": 2.287174940109253, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7235186100006104, + "num_tokens": 467120102.0, + "step": 18211 + }, + { + "epoch": 2.0, + "grad_norm": 2.2093117237091064, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7482165694236755, + "num_tokens": 467143956.0, + "step": 18212 + }, + { + "epoch": 2.0, + "step": 18212, + "total_flos": 2.1035281197777814e+19, + "train_loss": 0.9530878761967747, + "train_runtime": 21335.3439, + "train_samples_per_second": 13.657, + "train_steps_per_second": 0.854 + } + ], + "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": 2.1035281197777814e+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..3ab2f0c --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0da9ef9aa5b2d2c5a31a70ed4ace1c8692ba8975ff53373210b697072350007 +size 13329